{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ETKKTK4XWBUMLXKGUEGW4YHMPA","short_pith_number":"pith:ETKKTK4X","schema_version":"1.0","canonical_sha256":"24d4a9ab97b068c5dd46a10d6e60ec782d7b9abc82b0f475efdd4a0021259fb9","source":{"kind":"arxiv","id":"2605.15299","version":1},"attestation_state":"computed","paper":{"title":"Fortress: A Case Study in Stabilizing Search Recommendations via Temporal Data Augmentation and Feature Pruning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Fortress stabilizes search recommendation models by pruning features that introduce temporal volatility in prediction scores.","cross_cats":["cs.AI"],"primary_cat":"cs.IR","authors_text":"Babak Seyed Aghazadeh, Chris Alvino, Dayvid V. R. Oliveira, Jia Huang, Jinda Han, Kailash Thiyagarajan, Milind Pandurang Jagre, Puja Das, Zhinan Cheng","submitted_at":"2026-05-14T18:13:05Z","abstract_excerpt":"In search and recommendation systems, predictive models often suffer from temporal instability when certain input features introduce volatility in output scores. This instability can degrade model reliability and user experience especially in multi-stage systems where consistent predictions are critical for downstream decision making. We introduce Fortress, a general framework for enhancing model stability and accuracy by identifying and pruning features that contribute to inconsistent prediction scores over time. Fortress leverages historical snapshots temporally partitioned datasets capturin"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2605.15299","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.IR","submitted_at":"2026-05-14T18:13:05Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"567150ce3402cc1d15ab72a9c11de8acdd6d6fb86f148780a78b5469e6452bd4","abstract_canon_sha256":"2f727eb2536d89d647d63ff22883b9a3e26440dd7ab6992f2e6399710e161247"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:00:51.415500Z","signature_b64":"YMFGZIpt2EjvAXLu3Jv8sJcKf1u0yHQWOuva3eoH2GHPScbEZF3ZYngFVWu0d0qDWzYTVrYgzMhFE6twZMTgBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"24d4a9ab97b068c5dd46a10d6e60ec782d7b9abc82b0f475efdd4a0021259fb9","last_reissued_at":"2026-05-20T00:00:51.414684Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:00:51.414684Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Fortress: A Case Study in Stabilizing Search Recommendations via Temporal Data Augmentation and Feature Pruning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Fortress stabilizes search recommendation models by pruning features that introduce temporal volatility in prediction scores.","cross_cats":["cs.AI"],"primary_cat":"cs.IR","authors_text":"Babak Seyed Aghazadeh, Chris Alvino, Dayvid V. R. Oliveira, Jia Huang, Jinda Han, Kailash Thiyagarajan, Milind Pandurang Jagre, Puja Das, Zhinan Cheng","submitted_at":"2026-05-14T18:13:05Z","abstract_excerpt":"In search and recommendation systems, predictive models often suffer from temporal instability when certain input features introduce volatility in output scores. This instability can degrade model reliability and user experience especially in multi-stage systems where consistent predictions are critical for downstream decision making. We introduce Fortress, a general framework for enhancing model stability and accuracy by identifying and pruning features that contribute to inconsistent prediction scores over time. Fortress leverages historical snapshots temporally partitioned datasets capturin"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Fortress mitigates this trade-off by suppressing the volatility of engagement signals while retaining their predictive value leading to more stable and accurate models.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the features identified as instability-inducing in the historical snapshots are causally responsible for temporal score fluctuations and that their removal will not materially harm generalization on future data.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Fortress stabilizes query-to-app relevance models by pruning features that cause inconsistent predictions across time periods while retaining predictive power from engagement signals.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Fortress stabilizes search recommendation models by pruning features that introduce temporal volatility in prediction scores.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"8d7f4d1d0239eb2c894a7ee2bbe40d822139e4a9f7650807ad42eacf4674a38d"},"source":{"id":"2605.15299","kind":"arxiv","version":1},"verdict":{"id":"ec7298f9-7aec-43c0-a6f8-a3b240f7cb98","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T16:02:23.902917Z","strongest_claim":"Fortress mitigates this trade-off by suppressing the volatility of engagement signals while retaining their predictive value leading to more stable and accurate models.","one_line_summary":"Fortress stabilizes query-to-app relevance models by pruning features that cause inconsistent predictions across time periods while retaining predictive power from engagement signals.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the features identified as instability-inducing in the historical snapshots are causally responsible for temporal score fluctuations and that their removal will not materially harm generalization on future data.","pith_extraction_headline":"Fortress stabilizes search recommendation models by pruning features that introduce temporal volatility in prediction scores."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.15299/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T16:31:18.356632Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T16:16:11.974797Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T14:41:54.230857Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T13:33:22.781295Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"1c2989e01cdd1a6347322cda6dbb9961fd8068e630b52fe306e527743e6bfef0"},"references":{"count":20,"sample":[{"doi":"","year":2016,"title":"Tianqi Chen and Carlos Guestrin. 2016. Xgboost: A scalable tree boosting system. In Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining. 785–794","work_id":"748a8dc5-4f88-41d5-b932-237d8f4a6601","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Wenqi Fan. 2024. Recommender systems in the era of large language models (llms). IEEE Transactions on Knowledge and Data Engineering (2024), 1–20","work_id":"c6e5243f-4e04-4b2e-97ba-55a5c6d2c288","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Apple intelligence foundation language models","work_id":"73def087-9f61-47aa-b280-01d67f522dec","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Yupeng Hou, Junjie Zhang, Zihan Lin, Hongyu Lu, Ruobing Xie, Julian McAuley, and Wayne Xin Zhao. 2024. Large language models are zero- shot rankers for recommender systems. In European Conference on I","work_id":"1cc47959-3359-4115-a29a-55457d372355","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Gareth James, Daniela Witten, Trevor Hastie, and Robert Tibshirani","work_id":"945ea6e2-1c1a-4c6e-8d36-639f6ab89682","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":20,"snapshot_sha256":"3243a9cef8ad0047bcac594bac9255853333d2cea83967b3728187e0d0d259b6","internal_anchors":1},"formal_canon":{"evidence_count":2,"snapshot_sha256":"113d703e9bd7dbef58040daf9110fe29476d16d678201afd08e8fb6dc2d59358"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2605.15299","created_at":"2026-05-20T00:00:51.414813+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.15299v1","created_at":"2026-05-20T00:00:51.414813+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15299","created_at":"2026-05-20T00:00:51.414813+00:00"},{"alias_kind":"pith_short_12","alias_value":"ETKKTK4XWBUM","created_at":"2026-05-20T00:00:51.414813+00:00"},{"alias_kind":"pith_short_16","alias_value":"ETKKTK4XWBUMLXKG","created_at":"2026-05-20T00:00:51.414813+00:00"},{"alias_kind":"pith_short_8","alias_value":"ETKKTK4X","created_at":"2026-05-20T00:00:51.414813+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ETKKTK4XWBUMLXKGUEGW4YHMPA","json":"https://pith.science/pith/ETKKTK4XWBUMLXKGUEGW4YHMPA.json","graph_json":"https://pith.science/api/pith-number/ETKKTK4XWBUMLXKGUEGW4YHMPA/graph.json","events_json":"https://pith.science/api/pith-number/ETKKTK4XWBUMLXKGUEGW4YHMPA/events.json","paper":"https://pith.science/paper/ETKKTK4X"},"agent_actions":{"view_html":"https://pith.science/pith/ETKKTK4XWBUMLXKGUEGW4YHMPA","download_json":"https://pith.science/pith/ETKKTK4XWBUMLXKGUEGW4YHMPA.json","view_paper":"https://pith.science/paper/ETKKTK4X","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.15299&json=true","fetch_graph":"https://pith.science/api/pith-number/ETKKTK4XWBUMLXKGUEGW4YHMPA/graph.json","fetch_events":"https://pith.science/api/pith-number/ETKKTK4XWBUMLXKGUEGW4YHMPA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ETKKTK4XWBUMLXKGUEGW4YHMPA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ETKKTK4XWBUMLXKGUEGW4YHMPA/action/storage_attestation","attest_author":"https://pith.science/pith/ETKKTK4XWBUMLXKGUEGW4YHMPA/action/author_attestation","sign_citation":"https://pith.science/pith/ETKKTK4XWBUMLXKGUEGW4YHMPA/action/citation_signature","submit_replication":"https://pith.science/pith/ETKKTK4XWBUMLXKGUEGW4YHMPA/action/replication_record"}},"created_at":"2026-05-20T00:00:51.414813+00:00","updated_at":"2026-05-20T00:00:51.414813+00:00"}