{"as_of":"2026-08-04T20:58:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:9b65a42bea1ff436ae6d65daa18aef69ef66b242c220e1a40933d19077791b81","coverage":[{"denominator":87,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":87,"source":"paper_references, paper_reference_links","source_observed_at":"2026-06-25T19:19:50.420396Z","state":"measured"},{"denominator":87,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":87,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-04T06:34:03.388597+00:00","state":"measured"},{"denominator":0,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":0,"source":"paper_references, paper_reference_links","source_observed_at":null,"state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":0,"source":"cited_works","source_observed_at":null,"state":"measured"}],"external_citation_measurements":[],"inbound":[],"links":{"evidence":"/evidence","html":"/paper/2606.26036/citation-record","integrity":"/paper/2606.26036/integrity","json":"/paper/2606.26036/citation-record.json","paper":"/paper/2606.26036"},"outbound":[{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Second-order stochastic optimization for machine learn- ing in linear time.Journal of Machine Learning Re- search, 18(116):1–40, 2017","venue":null,"work_id":null,"year":2017},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:37cafb5ff78ed81923e87198db2b14fa8c83884ca97bed8cba877f79043a640b","observation_id":"eb6cdc65-12a3-4d12-b3e5-53ba31e2e5d4","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Can generative ai models extract deeper sentiments as compared to tra- ditional deep learning algorithms?IEEE Intelligent Systems, 39(2):5–10, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:aba562745e2df29f31f6a0c7971efd5b6865ac59c7f8e070b841f871527ff29f","observation_id":"2bcaf717-ea6d-4fee-af20-bdbae7a3fe18","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2305.10403","last_updated":"2023-09-13T20:35:45Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2023-05-17T17:46:53Z","title":"PaLM 2 Technical Report","version":3},"cited_work":{"arxiv_id":"2305.10403","doi":"10.48550/arxiv.2305.10403","metadata_source":"pith","pith_arxiv_id":"2305.10403","snapshot_observed_at":"2026-07-10T13:57:07.008081Z","title":"PaLM 2 Technical Report","venue":"cs.CL","work_id":"905ee9a7-ea61-4a94-bd62-2600cbe3e315","year":2023},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"cited_paper":"/paper/2305.10403","citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:20f94f9fe9e92c8d5bbf40154da36f2c35e47bf9190048bacb3e5d0be3bdcd24","observation_id":"8715a95c-df98-4b2d-84e6-49e30d5ff37e","resolution":{"observed_at":"2026-07-04T21:00:09.071036Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-04T06:34:03.388597+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-04T06:34:03.388597+00:00","source":"crossref"},{"observed_at":"2026-08-04T06:33:57.428241+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Introducing claude 4, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:d0dc46b4cd8d1f7a1e3210699f14905b773f6f740a3a34ee98ecb78a6ca0e913","observation_id":"017ad96d-deef-4c4f-85cd-ee4aadef2a25","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Summarizing news: Unleashing the power of bart, gpt-2, t5, and pega- sus models in text summarization","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:07f506072694a152767404fcb841a3f95881810980736e529ddce8ed022ba8b7","observation_id":"e3e43d48-4a40-44cb-a3a4-b58de0e1679f","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Guardbench: A large-scale benchmark for guardrail models","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:efa240cf24ceb0ddf034ad0333b8544c7c6a80aa13cfbeeca544a812b7f54b97","observation_id":"8e0c3564-e684-4db9-9d68-d95c1e2fca1c","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Machine un- learning","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:4c3e0af921f5824ae415a3ad97b6860428d11ec630120bd43f5c8c0fd8a98c79","observation_id":"a460fa0f-ac02-4304-ab12-e1ab8408329c","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"TweetNLP: Cutting- edge natural language processing for social media","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:a310d64c541478966eeb919b7278afdbd2581c6497fcb3158f950012b2dd96f5","observation_id":"4fa1a6ca-daf1-4c37-afc4-38dc141431a9","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Towards making systems forget with machine unlearning","venue":null,"work_id":null,"year":2015},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:a3e599e111c8a74bc75119c7f96b93677c57cdef3f3eb76f6fafdf0dad7f558e","observation_id":"21f23894-dfd2-4887-8a1e-804062c50691","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Pruning strategies for back- door defense in llms","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:e9cb8b101d952207708afc9b66d87a89504d053c4b6b4cd6967a0551102e9f9a","observation_id":"99404189-b224-4ed4-8ffc-f3d91a07dd68","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1811.03728","last_updated":"2018-11-09T01:08:00Z","snapshot_observed_at":"2026-07-06T07:13:39.005894Z","submitted_at":"2018-11-09T01:08:00Z","title":"Detecting Backdoor Attacks on Deep Neural Networks by Activation Clustering","version":1},"cited_work":{"arxiv_id":"1811.03728","doi":null,"metadata_source":"pith","pith_arxiv_id":"1811.03728","snapshot_observed_at":"2026-07-04T21:00:09.063525Z","title":"Detecting Backdoor Attacks on Deep Neural Networks by Activation Clustering","venue":"cs.LG","work_id":"2e1136ed-bfba-455c-979e-f617263c8b5e","year":2018},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":11,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"cited_paper":"/paper/1811.03728","citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:ba1cfa64ec22e62c651176437908af605a0918c625ea5ba22a28ea6bf5b97610","observation_id":"ad9c5a94-9ac6-452a-a2df-ea6ac76042cc","resolution":{"observed_at":"2026-07-04T21:00:09.065668Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-04T06:34:03.388597+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-04T06:34:03.388597+00:00","source":"crossref"},{"observed_at":"2026-08-04T06:33:57.428241+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"A thorough examination of the cnn/daily mail reading comprehension task","venue":null,"work_id":null,"year":2016},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:f0bd4ae0e229e2d4c303268c93cd9a99850f219a66b15c79a0c256234c2ac5d4","observation_id":"4148720e-0ace-4a26-b293-9344d0057875","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Badnl: Backdoor attacks against nlp mod- els with semantic-preserving improvements","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":13,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:135e329644b67e50ea3931bd79b50102946e2511be720e7f0ba7cd7a098144e4","observation_id":"6551c652-8c50-4a92-8cad-dc99ecf998d2","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Sdd: Self-degraded defense against malicious fine-tuning","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:598b53124721e27d332ea26c799e24e0248e53720245460424114f735d0a5b93","observation_id":"5372bdf6-70da-4d83-b404-26668075350a","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Gonzalez, Ion Stoica, and Eric P","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":15,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:a38a27ba92f6e292a13c59c41d962d2c479c6f97b0da45e8bffc9f2557bda5fd","observation_id":"1dc83893-e269-410e-b3c3-fa4726ce4d21","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Forget unlearning: Towards true data-deletion in machine learning","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":16,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:0974e0db29f0197171eab4d227551ca638a3e35aa2baf2cabc6425ab2cda7b6f","observation_id":"9877d1e3-c170-4415-be7c-e36bddd9c430","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Scaling instruction-finetuned language models.Journal of Ma- chine Learning Research, 25(70):1–53, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":17,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:0397165532410c1c0a44eebfb0e1d0b589268e8b3bee91ff064949e8734495d0","observation_id":"6fa6a541-c059-4f34-9112-2555d47400f6","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"A discourse-aware attention model for ab- stractive summarization of long documents","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":18,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:cbb41a14a2d37ec5dad87662598a11cf15bde93abb3e0f5d042564169cce5c57","observation_id":"8b79489f-2f02-42b5-8f7d-6c3e064b7112","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Certi- fied adversarial robustness via randomized smoothing","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":19,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:56a49147a8c20ce0ea4ce555a01660efed10051ea0fd26cdadba06c2d6cb0be2","observation_id":"5bdd9a06-8376-473f-98ec-263e898bc906","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Wikisum: Coherent summariza- tion dataset for efficient human-evaluation","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":20,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:0fb5c5fc0a845559faeaf97aa7c17fa2a641e0c6b70c1e7f0bfdb023a5c376ea","observation_id":"596f4fe6-3124-4e8c-9167-17590ea19007","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Structured in- formation extraction from scientific text with large lan- guage models.Nature communications, 15(1):1418, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":21,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:d91aeb64a317195681b124c6f122bbd4ad994c85ec7721a94b7b9aeddf70864c","observation_id":"58798f41-fe66-4aa1-a634-2b0cf5a593d3","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"The llama 3 herd of models.arXiv e-prints, pages arXiv–2407, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":22,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:813b265c0b4ae2052cd4b662929031fe4154a8b3373731e9da8b8e142796ff77","observation_id":"085f207d-352b-4699-ae43-8a4fdd126162","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"The algorithmic foun- dations of differential privacy.Foundations and Trends® in Theoretical Computer Science, 9(3-4):211–407, 2014","venue":null,"work_id":null,"year":2014},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":23,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:918d9489d131d4a51b9bd48a930a707a36574291bee9fc402df82139d7ffb2a4","observation_id":"d26b0f0c-3249-416d-b92f-b3d9c271c123","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1712.06751","last_updated":"2018-05-24T16:43:45Z","snapshot_observed_at":"2026-07-06T06:15:02.045009Z","submitted_at":"2017-12-19T02:15:19Z","title":"HotFlip: White-Box Adversarial Examples for Text Classification","version":2},"cited_work":{"arxiv_id":"1712.06751","doi":null,"metadata_source":"pith","pith_arxiv_id":"1712.06751","snapshot_observed_at":"2026-07-04T21:00:09.078340Z","title":"HotFlip: White-Box Adversarial Examples for Text Classification","venue":"cs.CL","work_id":"4437bc40-bf35-46a7-adb4-3051437b4028","year":2017},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":24,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"cited_paper":"/paper/1712.06751","citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:9ada8d8030797ffaad703204d155abb8424321e3f50bd17757473503de0cd881","observation_id":"cb190af6-bf31-4fd3-afd1-7fceaaf526e6","resolution":{"observed_at":"2026-07-04T21:00:09.079511Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-04T06:34:03.388597+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-04T06:34:03.388597+00:00","source":"crossref"},{"observed_at":"2026-08-04T06:33:57.428241+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Multi-news: A large-scale multi-document summarization dataset and abstractive hierarchical model","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":25,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:3fc2eaedbbc7b9adad32ebf8c9c3f9cb8cc7f133ef82307ebbf8fadfa4723d39","observation_id":"ceef7bcf-d751-4c81-88c2-64ed180275e3","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Attack to defend: Exploiting adversarial attacks for detecting poisoned models","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":26,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:600a30eeea8d84c46032892f370a5527d56aa6a5a765b7867d9879932dc7065a","observation_id":"2d052db2-aced-467e-b990-516d610f2259","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Strip: A de- fence against trojan attacks on deep neural networks","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":27,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:d420616c9b140e7f240a70e31b2633742cd0b27b1204dafa71273f14207c0976","observation_id":"59e1d459-1d79-4969-8dc0-f3e59754a0f4","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Eternal sunshine of the spotless net: Selective forgetting in deep networks","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":28,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:9da8741de27c38f75ef721f62e06c877c14182e000424192de373711491de955","observation_id":"0384e2a3-83be-4efa-8fb7-e30b1fd6baf6","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1412.6572","last_updated":"2015-03-20T20:19:16Z","snapshot_observed_at":"2026-07-06T04:04:16.777653Z","submitted_at":"2014-12-20T01:17:12Z","title":"Explaining and Harnessing Adversarial Examples","version":3},"cited_work":{"arxiv_id":"1412.6572","doi":"10.48550/arxiv.1412.6572","metadata_source":"pith","pith_arxiv_id":"1412.6572","snapshot_observed_at":"2026-07-10T19:57:34.026407Z","title":"Explaining and Harnessing Adversarial Examples","venue":"stat.ML","work_id":"2cedf8f6-7539-4c49-8136-f42a20487146","year":2014},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":29,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"cited_paper":"/paper/1412.6572","citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:b9cfc5e4c759caccb6704107e3137e0a7058a840e78c0398f24c7d91f5374d7d","observation_id":"2dbfdf83-15b9-4978-b07e-0ca57429a07c","resolution":{"observed_at":"2026-07-04T21:00:09.068340Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-04T06:34:03.388597+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-04T06:34:03.388597+00:00","source":"crossref"},{"observed_at":"2026-08-04T06:33:57.428241+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Flight of the pegasus? comparing transformers on few- shot and zero-shot multi-document abstractive summa- rization","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":30,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:9663a3e6b8af616777267a47efc0f5096ed046930e90524f1a3afbe9a3b8dbc5","observation_id":"b618a2f9-77cf-492e-8d44-5274e06388cf","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1909.04028","last_updated":"2019-09-08T05:50:18Z","snapshot_observed_at":"2026-08-03T19:17:56.283000Z","submitted_at":"2019-09-08T05:50:18Z","title":"Countering the Effects of Lead Bias in News Summarization via Multi-Stage Training and Auxiliary Losses","version":1},"cited_work":{"arxiv_id":"1909.04028","doi":null,"metadata_source":"arxiv_reference","pith_arxiv_id":"1909.04028","snapshot_observed_at":"2026-07-04T21:00:09.049949Z","title":"Countering the effects of lead bias in news summarization via multi-stage training and auxiliary losses.arXiv preprint arXiv:1909.04028, 2019","venue":null,"work_id":"b87c57c4-e65f-4eb9-b209-2cb3869fe1ab","year":1909},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":31,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"cited_paper":"/paper/1909.04028","citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:a56049265b6f7b6732b6c744e670d116c326101df22212cae5d6bc6baebb6369","observation_id":"b3d9ddff-11f9-4200-bec9-5ad98c99f817","resolution":{"observed_at":"2026-07-04T21:00:09.051813Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-04T06:34:03.388597+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-04T06:34:03.388597+00:00","source":"crossref"},{"observed_at":"2026-08-04T06:33:57.428241+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Certified data removal from machine learning models","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":32,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:7f2d6d4af6dd8465721d30e70f3fd38614829fe3c94d4afaecef341896324a79","observation_id":"f45b68c8-867a-4531-ab5a-4aa649a44926","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Adaptive machine unlearning.Advances in Neural Information Processing Systems, 34:16319–16330, 2021","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":33,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:da218228bd0ea661c91f2649c1a8f351d4e034ce7c065d43f9cc424b8f6c4820","observation_id":"bf36c18c-9d13-4d8d-9c27-986ac4ee862e","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Generative language models exhibit social identity bi- ases.Nature Computational Science, 5(1):65–75, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":34,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:19e9dbdf4622f4c3509be2ec51b99a0f6be699bfdfc994c0732356ac8234d25c","observation_id":"06e861f1-6c54-4178-bbc2-adfbaa049e75","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Adaptive defense against harmful fine- tuning for large language models via bayesian data scheduler","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":35,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:8fb483b2fbf073b65a33db74189abc25c44b3f9b066af09ccb71ced105339cdf","observation_id":"6023ad4b-0e64-4280-8ba0-a298e6ca15c1","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Antidote: Post-fine- tuning safety alignment for large language models against harmful fine-tuning attack","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":36,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:89b4a60f12e20f603d5d0413ac5c55418733632ccc94f55be49739ee4311851d","observation_id":"425aa7f7-7c9d-468c-b649-78da44e0ffff","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2410.21276","last_updated":"2024-10-25T17:43:01Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2024-10-25T17:43:01Z","title":"GPT-4o System Card","version":1},"cited_work":{"arxiv_id":"2410.21276","doi":"10.1177/15248380231178756","metadata_source":"pith","pith_arxiv_id":"2410.21276","snapshot_observed_at":"2026-07-11T11:50:26.030339Z","title":"GPT-4o System Card","venue":"cs.CL","work_id":"f37bf1c7-4964-4e56-9762-d20da8d9009f","year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":37,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"cited_paper":"/paper/2410.21276","citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:6a3f4a0bd4911f8ae8bd2f77db362b6b7430ba9a4fcdaf8947543d3bf3af98f6","observation_id":"05feafc3-1995-4c38-90f8-2491ebbdbc38","resolution":{"observed_at":"2026-07-04T21:00:09.074486Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-04T06:34:03.388597+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-04T06:34:03.388597+00:00","source":"crossref"},{"observed_at":"2026-08-04T06:33:57.428241+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Knowledge unlearning for mitigating privacy risks in language models","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":38,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:14323892284e51fc2b53daf45067671eed38ece73323dbef500919f392375188","observation_id":"6295a362-6327-4e33-8720-0c47aa62a8c7","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2310.06825","last_updated":"2023-10-10T17:54:58Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2023-10-10T17:54:58Z","title":"Mistral 7B","version":1},"cited_work":{"arxiv_id":"2310.06825","doi":"10.48550/arxiv.2310.06825","metadata_source":"pith","pith_arxiv_id":"2310.06825","snapshot_observed_at":"2026-07-10T14:47:14.537301Z","title":"Mistral 7B","venue":"cs.CL","work_id":"eb5e1305-ad11-4875-ad8d-ad8b8f697599","year":2023},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":39,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"cited_paper":"/paper/2310.06825","citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:8f26862ff89cebe804a4d143f6d4bd948ec13b506ff791fe9777b9e055e6ede0","observation_id":"10ee2f34-39af-40fc-b267-daf7dd461da1","resolution":{"observed_at":"2026-07-04T21:00:09.077139Z","resolver_source":"local_arxiv","status":"malformed_identifier"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-04T06:34:03.388597+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-04T06:34:03.388597+00:00","source":"crossref"},{"observed_at":"2026-08-02T03:08:12.282824+00:00","source":"crossref_status_cache"},{"observed_at":"2026-08-02T03:08:12.282824+00:00","source":"openalex_status_cache"},{"observed_at":"2026-08-04T06:33:57.428241+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Is bert really robust? a strong baseline for nat- ural language attack on text classification and entailment","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":40,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:2c14a8eda993c13b4e7aec7b1206b6d05ef8e959d04936f2974c260b3e8e3366","observation_id":"3024e743-2020-4bb0-987e-c36fb0d896f9","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Understanding black- box predictions via influence functions","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":41,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:28495359ec5ac642c45253d90c295506f8176d1bad4fb660dd68130806bdb234","observation_id":"87c6c9d6-2a48-4069-a441-3628fd0abf56","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Gender bias and stereotypes in large language models","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":42,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:36f3c7459acfe1dfe2738c01c1f8fe7a670578ff38b3a17c39809185c7af25e2","observation_id":"9427311d-93f9-4f49-bfae-6d4e888a041e","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Investigating implicit bias in large language models: A large-scale study of over 50 llms","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":43,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:d203645f838d41e24ee59b93f4aa87e3447329602df40080cd22be2abf3dc034","observation_id":"7a3430c7-0467-42a9-b943-4bb73bacd6aa","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Datainf: Efficiently estimating data influence in lora- tuned llms and diffusion models","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":44,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:2d037572b3784b509ce4ef9be4e4fde83dfbf7b1f4889eb3faf9d86fd45cd3a1","observation_id":"9eb4bb5e-e7eb-472a-89c2-bdae45c6695d","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1910.13461","last_updated":"2019-10-29T18:01:00Z","snapshot_observed_at":"2026-07-06T08:33:12.534026Z","submitted_at":"2019-10-29T18:01:00Z","title":"BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension","version":1},"cited_work":{"arxiv_id":"1910.13461","doi":"10.48550/arxiv.1910.13461","metadata_source":"pith","pith_arxiv_id":"1910.13461","snapshot_observed_at":"2026-07-10T20:47:34.539793Z","title":"BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension","venue":"cs.CL","work_id":"7ab72623-0d1b-41fc-96e9-181246f2ea00","year":2019},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":45,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"cited_paper":"/paper/1910.13461","citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:7c7e656af0d108e2448b2fe729521dab69cf6bfa9cd5cc7012402c8c48e0f111","observation_id":"35267433-e757-4b32-a2e1-8b0e64cc8947","resolution":{"observed_at":"2026-07-04T21:00:09.057166Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-04T06:34:03.388597+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-04T06:34:03.388597+00:00","source":"crossref"},{"observed_at":"2026-08-04T06:33:57.428241+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Efficiently generat- ing sentence-level textual adversarial examples with seq2seq stacked auto-encoder.Expert Systems with Ap- plications, 213:119170, 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":46,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:dd7277ca1c3f5049dc3c43a5e44f3832cbc41695fc616a640e1bd6b7da182a89","observation_id":"6dbc112e-4b3b-4a5e-910c-4939b81a766f","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Text adver- sarial purification as defense against adversarial attacks","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":47,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:6451769203460580961cd80a41a1cd371c757f1a31758eee6f5a88472383893a","observation_id":"a9a245af-0ed9-4f97-9182-795eb49698c2","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"ROUGE: A package for automatic eval- uation of summaries","venue":null,"work_id":null,"year":2004},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":48,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:61cbe1b5f5241669283739eedf2cb3df8c0bddf4aa1df805f2f242ac47cefffb","observation_id":"00527617-a8b4-4279-a9cc-fd79fe46a0b8","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Joint character-level word embedding and adversarial stability training to defend adversarial text","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":49,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:81ae2cfbd1b675771fbfa00c3891e149bcbe30d71040cb64908919536c5acb6e","observation_id":"705915d7-bc53-4a85-a97d-bcb7b81c4a1a","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"On learning to summarize with large language models as references","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":50,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:cf744a8a07b0f56a6ca79dfb1ce6905762a8ac9e04135e36f50d1eadcf475045","observation_id":"76192794-276b-4873-8313-b65c68c26edd","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Modality- aware neuron pruning for unlearning in multi- modal large language models","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":51,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:a5fba141a59b87d11713e22c4a015f0e3fc56f08532f3d4376c69f1daa08fca2","observation_id":"dcd95374-5a56-4937-98dc-26baad1c9538","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Multi-xscience: A large-scale dataset for extreme multi-document sum- marization of scientific articles","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":52,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:3673e591fe6cb4b3c0452c2a87debe5227e400cd9712d72e6f4a57c44c812b65","observation_id":"e790fcd0-dc66-459f-b762-dbd741870966","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Towards deep learning models resistant to adversarial attacks","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":53,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:f6f55b25d48b656d88efa1943ea0dd408e7170e4d2c3144e141430a75dbb8098","observation_id":"07012f10-eb41-48ee-89d7-dbbdffac9ff3","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Adversarial training methods for semi-supervised text classification","venue":null,"work_id":null,"year":2017},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":54,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:3b4b06441856c46ab69d9daa82a38a520936b220f1a3ae683cf8b0d0359235d2","observation_id":"eaca5dd9-6842-43bf-ad9c-654eda596969","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Adversarial text purification: A large language model approach for defense","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":55,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:aef74d31e8d0cbf29534f0030bc1fe9fee6462daa51653fc579e29ab54248ca8","observation_id":"d3e48206-941d-41c5-b006-7081bb440440","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Sum- marunner: A recurrent neural network based sequence model for extractive summarization of documents","venue":null,"work_id":null,"year":2017},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":56,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:39eec610f769835dd741097c02a614fc044820d03cc3658c1a4f46192c6db99f","observation_id":"aa6f6014-46f1-41a3-9719-6a9510bd085e","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"A survey of machine unlearning.ACM Transactions on Intelligent Systems and Technology, 16(5):1–46, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":57,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:ad630def542f46a212bb6ffb2a24bb984110d258ed64bef111b3dba78eaca50d","observation_id":"ece11160-796d-4913-897d-e4ed3b4b30fe","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Textguard: Provable defense against backdoor attacks on text classification","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":58,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:2cdf7d5307f3ed2e081684a3a1f1310b695d32576b3c2a16b8ae1b321b29bf9a","observation_id":"da74e412-a538-4612-8967-3f006dab3404","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Estimating training data influence by trac- ing gradient descent.Advances in Neural Information Processing Systems, 33:19920–19930, 2020","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":59,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:0acb2cf6221eff720562d5597c6267a2bf8ab608ff674bf9a3220e35c6de3802","observation_id":"278bc3f8-ec6c-463e-95e7-c3e872bdf02e","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":null,"venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":60,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:f9d5e23b6eb4fb57744d81cd1740e072b5d294b2469acf50dbc2c6b8ed424210","observation_id":"4cd1474f-cbf0-4c59-8472-431b07b48f7c","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Exploring the limits of transfer learn- ing with a unified text-to-text transformer.The Journal of Machine Learning Research, 21(1):5485–5551, 2020","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":61,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:fa75d502fff3edb60591076af9a842794668a8aef73b91741152b446f1e1a061","observation_id":"9b6139d6-89c8-449a-80d0-b55642919784","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"On context utilization in summarization with large language models","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":62,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:8383efe566e75e4b9e2e4e51b1628575be48f1988ab26c6a50b3a63c5e633829","observation_id":"5cf47f4a-def2-4249-a204-bc84b0353150","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Sentence-bert: Sen- tence embeddings using siamese bert-networks","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":63,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:8a837cc4e7a63ed38e197f00a550773b15c450c22f1b04d7cfb036dbf817a265","observation_id":"8e4f0803-b9b8-457f-971f-e89d44c70762","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Representation nois- ing: A defence mechanism against harmful finetuning","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":64,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:d9b35afeac6784015141121c3cbf78ff4e5db4ce4537067d278847f2c9b0a60d","observation_id":"d7fe8151-a442-4c28-b39e-391335b69533","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Immuniza- tion against harmful fine-tuning attacks","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":65,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:841c841e6c1288c0f8dd0549ecc49cf258962da180141b14470d96d8aa11fe7f","observation_id":"05269805-6e05-447a-b53b-87f087ddb29a","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"A maximum entropy approach to adaptive statistical language modelling.Computer speech and language, 10(3):187, 1996","venue":null,"work_id":null,"year":1996},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":66,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:bf8c353b0f392d578eae74d7680258cd8a916480e39003ae31bd8afb5a7fd19d","observation_id":"e4103cd9-8fe7-4bfb-a527-03caf9de8517","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Rome was built in 1776: A case study on factual correctness in knowledge-grounded response generation","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":67,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:d5e916a13c4e0221b5e48994995d8294787f56a190e7bf829d08cd08d12a81b0","observation_id":"99b8ab68-0135-4a69-844d-48f00c324692","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Evaluating performance of transformer models for dialogue summarization: A comparison of t5-base, t5-small, and bart-base","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":68,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:4adacb71bc05c7b9c06c4c59638781cec850fbaf9b9a124c6a307ded3e5160cc","observation_id":"130c25af-84da-4724-b614-3d3281f06508","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Textshield: Beyond successfully detecting adversarial 17 sentences in text classification","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":69,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:8be6e4a6f3074f3218026ac0f1430f94042787e45aee003e64e7c099de96da6c","observation_id":"f974e4a5-314e-4471-bc53-c93878745f27","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"what shapes your bias?","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":70,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:16b108cef8d43feedfc289d98a71cf356b05a63f23ef9cedb8ba69edeb2c151b","observation_id":"4c5ce92e-f228-4751-9d22-c1245623397f","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Peftguard: detecting backdoor attacks against parameter- efficient fine-tuning","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":71,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:731ca44d20eba8fecadf437d26f6ab0d9b4ba78ca3ecfeb166ea7178e88df891","observation_id":"c7968ac6-0cce-4617-93f2-4f4d0053762e","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Fever: a large-scale dataset for fact extraction and verification","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":72,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:db848b322fdcb57155b6d5a55710aeb7dfda06c63d3829b65e8b04f438d05372","observation_id":"73c731f6-9c06-44b6-ae50-08ec02b5d5ed","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Attacks against abstractive text summarization models through lead bias and influence functions","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":73,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:9b10e2d8912055d6fd5c5a8a2a39eed66d8825f240f3b3f652e38b8686bf4e54","observation_id":"2c0e1d12-c5f7-4b1c-80ba-c6dfd9a890e9","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Unrolling sgd: Understanding factors influencing machine unlearning","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":74,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:40afed1b42f2ec07a6aac9e7d29c06c550a2cc8c90d9609b54e7631c795aa017","observation_id":"9287bb55-6600-4410-8209-c1c8fd7c8ce4","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Spectral signatures in backdoor attacks.Advances in neural information processing systems, 31, 2018","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":75,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:5185b76ae9280823577e2fe37ca4a03bdc30d0809c33c83d8f27ec249be11bf5","observation_id":"6cc86b16-7391-47f4-9068-8c46a60de85b","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Concealed data poisoning attacks on nlp models","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":76,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:920d679cc2be8b99a78ece9379df73373b1f1e13a9912b3e98da20276a41730d","observation_id":"47c48506-8041-41f5-bcc9-d3b18c4e1c1b","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Neu- ral cleanse: Identifying and mitigating backdoor attacks in neural networks","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":77,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:8e9ce30ca0e8d548145e92075e27240dedefbeafa43894d2540c01feb2c9dd8a","observation_id":"cafa6457-c87a-4018-bb4b-9bbedbb10e71","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2205.01287","last_updated":"2022-06-11T03:42:27Z","snapshot_observed_at":"2026-07-06T13:06:05.298886Z","submitted_at":"2022-05-03T03:44:03Z","title":"SemAttack: Natural Textual Attacks via Different Semantic Spaces","version":3},"cited_work":{"arxiv_id":"2205.01287","doi":null,"metadata_source":"arxiv_reference","pith_arxiv_id":"2205.01287","snapshot_observed_at":"2026-07-04T21:00:09.057633Z","title":"Semattack: natural textual attacks via differ- ent semantic spaces.arXiv preprint arXiv:2205.01287, 2022","venue":null,"work_id":"67025f24-6d86-49c6-83e2-38d76edad213","year":2022},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":78,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"cited_paper":"/paper/2205.01287","citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:cd772360f6094e681bbf5a706a54f6514aeb909af65ee89dd2f5b98689ce5ced","observation_id":"cebed54e-308f-49c0-a0a0-c2881f677261","resolution":{"observed_at":"2026-07-04T21:00:09.059395Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-04T06:34:03.388597+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-04T06:34:03.388597+00:00","source":"crossref"},{"observed_at":"2026-08-04T06:33:57.428241+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Improving neural language modeling via adversarial training","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":79,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:f6583016a4736da938e6faa888d3f77d1d08de17404e9eb50d6e4e9b24bfb131","observation_id":"a55c1db1-8af1-4364-b880-4a0a44169e1a","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Mitigating data poisoning in text classification with differential privacy","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":80,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:cc2e8f398d7e08b0f6d4d6f3af5d157515a90caa2a4b258e4cc8a78ac538c3bd","observation_id":"8504d6ef-1482-432c-8a91-1d735c30b35c","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.09388","last_updated":"2025-05-14T13:41:34Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2025-05-14T13:41:34Z","title":"Qwen3 Technical Report","version":1},"cited_work":{"arxiv_id":"2505.09388","doi":"10.1016/j.aiopen.2022.12","metadata_source":"pith","pith_arxiv_id":"2505.09388","snapshot_observed_at":"2026-07-11T11:50:26.030339Z","title":"Qwen3 Technical Report","venue":"cs.CL","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","year":2025},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":81,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"cited_paper":"/paper/2505.09388","citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:cdc1082d2545aaf87b6cbd8fe3b75ea62531f797727e186623ee0e788ec3ae1b","observation_id":"5f6377c5-2f90-4e56-bea1-e8374137fb63","resolution":{"observed_at":"2026-07-04T21:00:09.062250Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-04T06:34:03.388597+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-04T06:34:03.388597+00:00","source":"crossref"},{"observed_at":"2026-08-04T06:33:57.428241+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Machine unlearn- ing of pre-trained large language models","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":82,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:7a3acda4a1ab672a382549dd2818f45f532d29502d88cf0f6436f5007f6bddfb","observation_id":"f259fb7c-ccc8-4188-840d-36b5b72e486e","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Clibe: Detecting dynamic back- doors in transformer-based nlp models.Network and Distributed System Security (NDSS) Symposium, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":83,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:9dbbf8df12f8cfbdcf3bc88490d7d125ab1bb29e242baf34a35ae75401582fc0","observation_id":"8d2b888f-35e3-48d4-984b-e9486d07ea90","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Alignscore: Evaluating factual consistency with a uni- fied alignment function","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":84,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:8b89b04a745de37f2ad7384d0c0e60a4354346ae816046075383ad556bf4b079","observation_id":"68dcc53c-46e6-4cc4-adee-956a5aba0717","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"Pegasus: Pre-training with extracted gap-sentences for abstractive summarization","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":85,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:47d6386e80371e1651d5c27e002dc5ea6f51360942ee71481c82b9adb231bed9","observation_id":"a99210a0-48a1-406f-b701-9b3714026f2d","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"A comprehensive survey of abstractive text summarization based on deep learning.Computa- tional intelligence and neuroscience, 2022(1):7132226, 2022","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":86,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:3a323c41536afebd53464513724741d7d349f69f63ff571a26bd6a0becf0a547","observation_id":"abfeb607-acb4-4e1a-a089-96ba257b2be5","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-06-25T19:19:50.420396Z","title":"silicon nanowire design","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning","version":1},"reference_index":87,"source":"pdf_text","source_observed_at":"2026-06-25T19:19:50.420396Z"},"links":{"citing_paper":"/paper/2606.26036"},"observation_digest":"sha256:e152a1930026134fcb473bdcf0945ba44b63b66350c4d0d3092e581eea275a04","observation_id":"01b46de1-a663-4e0b-a616-68470f0e2f70","resolution":{"observed_at":"2026-06-25T19:19:50.420396Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"paper":{"arxiv_id":"2606.26036","last_updated":"2026-06-24T17:12:42Z","latest_version":1,"primary_category":"cs.CL","snapshot_observed_at":"2026-08-02T17:50:27.411206Z","submitted_at":"2026-06-24T17:12:42Z","title":"Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning"},"reference_resolution":{"displayed":87,"state_counts":{"malformed_identifier":1,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":77,"verified_exact":9,"verified_fuzzy":0},"total_outbound_references":87},"refusal":"A citation records a reference. It does not transfer a finding from one paper to another.","schema":"pith.paper-citation-record.v1","standing_sources":[{"observed_at":"2026-08-04T06:34:03.388597+00:00","source":"crossref"},{"observed_at":"2026-08-04T06:33:57.428241+00:00","source":"retraction_watch"}],"thesis":"As of 4 August 2026, this Paper Citation Record lists 87 of 87 outbound references and 0 inbound Pith citation observations for arXiv:2606.26036."}