{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:T5UTGPG7CVZZT7TKDAS6VHQLJH","short_pith_number":"pith:T5UTGPG7","schema_version":"1.0","canonical_sha256":"9f69333cdf157399fe6a1825ea9e0b49e74097988cb2ca470b9a60665db69d12","source":{"kind":"arxiv","id":"2005.03724","version":1},"attestation_state":"computed","paper":{"title":"SUPERT: Towards New Frontiers in Unsupervised Evaluation Metrics for Multi-Document Summarization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.IR"],"primary_cat":"cs.CL","authors_text":"Steffen Eger, Wei Zhao, Yang Gao","submitted_at":"2020-05-07T19:54:24Z","abstract_excerpt":"We study unsupervised multi-document summarization evaluation metrics, which require neither human-written reference summaries nor human annotations (e.g. preferences, ratings, etc.). We propose SUPERT, which rates the quality of a summary by measuring its semantic similarity with a pseudo reference summary, i.e. selected salient sentences from the source documents, using contextualized embeddings and soft token alignment techniques. Compared to the state-of-the-art unsupervised evaluation metrics, SUPERT correlates better with human ratings by 18-39%. Furthermore, we use SUPERT as rewards to "},"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":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2005.03724","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-05-07T19:54:24Z","cross_cats_sorted":["cs.IR"],"title_canon_sha256":"1fcfc02022b41dc0755d00c7e8de4a853996986d0d82ec6b06538a4ef7b8b24c","abstract_canon_sha256":"f2cf1b0ed3fede075fbef3c3f301519ea7717700135202b0bfcdc69abc629d42"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:01:25.344341Z","signature_b64":"pIvyFYXaHDSWG4JJ121iuKz7dSLoAitkLchV6Xxz7BbvrjdpszqXhvQnPsoOWVk8wFEsczHgM1Q3yApitr+mDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9f69333cdf157399fe6a1825ea9e0b49e74097988cb2ca470b9a60665db69d12","last_reissued_at":"2026-07-05T01:01:25.343888Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:01:25.343888Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SUPERT: Towards New Frontiers in Unsupervised Evaluation Metrics for Multi-Document Summarization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.IR"],"primary_cat":"cs.CL","authors_text":"Steffen Eger, Wei Zhao, Yang Gao","submitted_at":"2020-05-07T19:54:24Z","abstract_excerpt":"We study unsupervised multi-document summarization evaluation metrics, which require neither human-written reference summaries nor human annotations (e.g. preferences, ratings, etc.). We propose SUPERT, which rates the quality of a summary by measuring its semantic similarity with a pseudo reference summary, i.e. selected salient sentences from the source documents, using contextualized embeddings and soft token alignment techniques. Compared to the state-of-the-art unsupervised evaluation metrics, SUPERT correlates better with human ratings by 18-39%. Furthermore, we use SUPERT as rewards to "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2005.03724","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2005.03724/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"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":"2005.03724","created_at":"2026-07-05T01:01:25.343971+00:00"},{"alias_kind":"arxiv_version","alias_value":"2005.03724v1","created_at":"2026-07-05T01:01:25.343971+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2005.03724","created_at":"2026-07-05T01:01:25.343971+00:00"},{"alias_kind":"pith_short_12","alias_value":"T5UTGPG7CVZZ","created_at":"2026-07-05T01:01:25.343971+00:00"},{"alias_kind":"pith_short_16","alias_value":"T5UTGPG7CVZZT7TK","created_at":"2026-07-05T01:01:25.343971+00:00"},{"alias_kind":"pith_short_8","alias_value":"T5UTGPG7","created_at":"2026-07-05T01:01:25.343971+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.07653","citing_title":"An Automated Length-Aware Quality Metric for Summarization","ref_index":2020,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/T5UTGPG7CVZZT7TKDAS6VHQLJH","json":"https://pith.science/pith/T5UTGPG7CVZZT7TKDAS6VHQLJH.json","graph_json":"https://pith.science/api/pith-number/T5UTGPG7CVZZT7TKDAS6VHQLJH/graph.json","events_json":"https://pith.science/api/pith-number/T5UTGPG7CVZZT7TKDAS6VHQLJH/events.json","paper":"https://pith.science/paper/T5UTGPG7"},"agent_actions":{"view_html":"https://pith.science/pith/T5UTGPG7CVZZT7TKDAS6VHQLJH","download_json":"https://pith.science/pith/T5UTGPG7CVZZT7TKDAS6VHQLJH.json","view_paper":"https://pith.science/paper/T5UTGPG7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2005.03724&json=true","fetch_graph":"https://pith.science/api/pith-number/T5UTGPG7CVZZT7TKDAS6VHQLJH/graph.json","fetch_events":"https://pith.science/api/pith-number/T5UTGPG7CVZZT7TKDAS6VHQLJH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/T5UTGPG7CVZZT7TKDAS6VHQLJH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/T5UTGPG7CVZZT7TKDAS6VHQLJH/action/storage_attestation","attest_author":"https://pith.science/pith/T5UTGPG7CVZZT7TKDAS6VHQLJH/action/author_attestation","sign_citation":"https://pith.science/pith/T5UTGPG7CVZZT7TKDAS6VHQLJH/action/citation_signature","submit_replication":"https://pith.science/pith/T5UTGPG7CVZZT7TKDAS6VHQLJH/action/replication_record"}},"created_at":"2026-07-05T01:01:25.343971+00:00","updated_at":"2026-07-05T01:01:25.343971+00:00"}