{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:ZHYJ5H46FDEGLOJUB5EVZGXO2Y","short_pith_number":"pith:ZHYJ5H46","schema_version":"1.0","canonical_sha256":"c9f09e9f9e28c865b9340f495c9aeed636af6ef39f38bc95d57f27efcfffa43b","source":{"kind":"arxiv","id":"2305.15067","version":3},"attestation_state":"computed","paper":{"title":"Not All Metrics Are Guilty: Improving NLG Evaluation by Diversifying References","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dongdong Zhang, Furu Wei, Haoyang Huang, Hongyuan Lu, Tianyi Tang, Tom Kocmi, Wayne Xin Zhao, Yuchen Eleanor Jiang","submitted_at":"2023-05-24T11:53:29Z","abstract_excerpt":"Most research about natural language generation (NLG) relies on evaluation benchmarks with limited references for a sample, which may result in poor correlations with human judgements. The underlying reason is that one semantic meaning can actually be expressed in different forms, and the evaluation with a single or few references may not accurately reflect the quality of the model's hypotheses. To address this issue, this paper presents a simple and effective method, named Div-Ref, to enhance existing evaluation benchmarks by enriching the number of references. We leverage large language mode"},"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":"2305.15067","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-05-24T11:53:29Z","cross_cats_sorted":[],"title_canon_sha256":"9db56a87b31e44b97e559b7d39eef53fd2982d946e2bac17ab555e61fe737404","abstract_canon_sha256":"f466c1ffca30846db49129d99685aad5ced8eb939121e8a9d4124f8e1eaca46e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:23:09.143922Z","signature_b64":"99gfp6aj5h/8XI+T1RY8dsxxPY/tSDvlB5jejiKoPuI4YmBoDM+zR9ZjlxX1kjYQ05r6CH1k7aVYHXauOKdsDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c9f09e9f9e28c865b9340f495c9aeed636af6ef39f38bc95d57f27efcfffa43b","last_reissued_at":"2026-07-05T08:23:09.143439Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:23:09.143439Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Not All Metrics Are Guilty: Improving NLG Evaluation by Diversifying References","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dongdong Zhang, Furu Wei, Haoyang Huang, Hongyuan Lu, Tianyi Tang, Tom Kocmi, Wayne Xin Zhao, Yuchen Eleanor Jiang","submitted_at":"2023-05-24T11:53:29Z","abstract_excerpt":"Most research about natural language generation (NLG) relies on evaluation benchmarks with limited references for a sample, which may result in poor correlations with human judgements. The underlying reason is that one semantic meaning can actually be expressed in different forms, and the evaluation with a single or few references may not accurately reflect the quality of the model's hypotheses. To address this issue, this paper presents a simple and effective method, named Div-Ref, to enhance existing evaluation benchmarks by enriching the number of references. We leverage large language mode"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.15067","kind":"arxiv","version":3},"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/2305.15067/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":"2305.15067","created_at":"2026-07-05T08:23:09.143496+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.15067v3","created_at":"2026-07-05T08:23:09.143496+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.15067","created_at":"2026-07-05T08:23:09.143496+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZHYJ5H46FDEG","created_at":"2026-07-05T08:23:09.143496+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZHYJ5H46FDEGLOJU","created_at":"2026-07-05T08:23:09.143496+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZHYJ5H46","created_at":"2026-07-05T08:23:09.143496+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2411.17204","citing_title":"Strategic Prompting for Conversational Tasks: A Comparative Analysis of Large Language Models Across Diverse Conversational Tasks","ref_index":45,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZHYJ5H46FDEGLOJUB5EVZGXO2Y","json":"https://pith.science/pith/ZHYJ5H46FDEGLOJUB5EVZGXO2Y.json","graph_json":"https://pith.science/api/pith-number/ZHYJ5H46FDEGLOJUB5EVZGXO2Y/graph.json","events_json":"https://pith.science/api/pith-number/ZHYJ5H46FDEGLOJUB5EVZGXO2Y/events.json","paper":"https://pith.science/paper/ZHYJ5H46"},"agent_actions":{"view_html":"https://pith.science/pith/ZHYJ5H46FDEGLOJUB5EVZGXO2Y","download_json":"https://pith.science/pith/ZHYJ5H46FDEGLOJUB5EVZGXO2Y.json","view_paper":"https://pith.science/paper/ZHYJ5H46","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.15067&json=true","fetch_graph":"https://pith.science/api/pith-number/ZHYJ5H46FDEGLOJUB5EVZGXO2Y/graph.json","fetch_events":"https://pith.science/api/pith-number/ZHYJ5H46FDEGLOJUB5EVZGXO2Y/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZHYJ5H46FDEGLOJUB5EVZGXO2Y/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZHYJ5H46FDEGLOJUB5EVZGXO2Y/action/storage_attestation","attest_author":"https://pith.science/pith/ZHYJ5H46FDEGLOJUB5EVZGXO2Y/action/author_attestation","sign_citation":"https://pith.science/pith/ZHYJ5H46FDEGLOJUB5EVZGXO2Y/action/citation_signature","submit_replication":"https://pith.science/pith/ZHYJ5H46FDEGLOJUB5EVZGXO2Y/action/replication_record"}},"created_at":"2026-07-05T08:23:09.143496+00:00","updated_at":"2026-07-05T08:23:09.143496+00:00"}