{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:R22L6ARN4PC6SZTNK7EAOWIV5B","short_pith_number":"pith:R22L6ARN","schema_version":"1.0","canonical_sha256":"8eb4bf022de3c5e9666d57c8075915e8787cc4edcf44eb4c24ffdcb7b6543548","source":{"kind":"arxiv","id":"1909.03087","version":1},"attestation_state":"computed","paper":{"title":"ACUTE-EVAL: Improved Dialogue Evaluation with Optimized Questions and Multi-turn Comparisons","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jason Weston, Margaret Li, Stephen Roller","submitted_at":"2019-09-06T18:44:49Z","abstract_excerpt":"While dialogue remains an important end-goal of natural language research, the difficulty of evaluation is an oft-quoted reason why it remains troublesome to make real progress towards its solution. Evaluation difficulties are actually two-fold: not only do automatic metrics not correlate well with human judgments, but also human judgments themselves are in fact difficult to measure. The two most used human judgment tests, single-turn pairwise evaluation and multi-turn Likert scores, both have serious flaws as we discuss in this work.\n  We instead provide a novel procedure involving comparing "},"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":"1909.03087","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-09-06T18:44:49Z","cross_cats_sorted":[],"title_canon_sha256":"1c1737366acfa6e78aba3b736b503aaceba4de433f88636d3d2b0af98e978768","abstract_canon_sha256":"f612d1c448928cbeacedcf342a46c3138b63f53a5d0a522b4693602d9f56eef2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:03:02.574236Z","signature_b64":"pOepMBYGX0EOnrAiPycm7IwLSxiyVwtokbyVWu55GQQT0MBbdWrjGzlmNBHJSgbkvV/fFS/OiNcmmEiBB7PUCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8eb4bf022de3c5e9666d57c8075915e8787cc4edcf44eb4c24ffdcb7b6543548","last_reissued_at":"2026-07-05T00:03:02.573894Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:03:02.573894Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ACUTE-EVAL: Improved Dialogue Evaluation with Optimized Questions and Multi-turn Comparisons","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jason Weston, Margaret Li, Stephen Roller","submitted_at":"2019-09-06T18:44:49Z","abstract_excerpt":"While dialogue remains an important end-goal of natural language research, the difficulty of evaluation is an oft-quoted reason why it remains troublesome to make real progress towards its solution. Evaluation difficulties are actually two-fold: not only do automatic metrics not correlate well with human judgments, but also human judgments themselves are in fact difficult to measure. The two most used human judgment tests, single-turn pairwise evaluation and multi-turn Likert scores, both have serious flaws as we discuss in this work.\n  We instead provide a novel procedure involving comparing "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1909.03087","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/1909.03087/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":"1909.03087","created_at":"2026-07-05T00:03:02.573950+00:00"},{"alias_kind":"arxiv_version","alias_value":"1909.03087v1","created_at":"2026-07-05T00:03:02.573950+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1909.03087","created_at":"2026-07-05T00:03:02.573950+00:00"},{"alias_kind":"pith_short_12","alias_value":"R22L6ARN4PC6","created_at":"2026-07-05T00:03:02.573950+00:00"},{"alias_kind":"pith_short_16","alias_value":"R22L6ARN4PC6SZTN","created_at":"2026-07-05T00:03:02.573950+00:00"},{"alias_kind":"pith_short_8","alias_value":"R22L6ARN","created_at":"2026-07-05T00:03:02.573950+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2410.04155","citing_title":"Toxic Subword Pruning for Dialogue Response Generation on Large Language Models","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2009.01325","citing_title":"Learning to summarize from human feedback","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2005.11401","citing_title":"Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks","ref_index":37,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/R22L6ARN4PC6SZTNK7EAOWIV5B","json":"https://pith.science/pith/R22L6ARN4PC6SZTNK7EAOWIV5B.json","graph_json":"https://pith.science/api/pith-number/R22L6ARN4PC6SZTNK7EAOWIV5B/graph.json","events_json":"https://pith.science/api/pith-number/R22L6ARN4PC6SZTNK7EAOWIV5B/events.json","paper":"https://pith.science/paper/R22L6ARN"},"agent_actions":{"view_html":"https://pith.science/pith/R22L6ARN4PC6SZTNK7EAOWIV5B","download_json":"https://pith.science/pith/R22L6ARN4PC6SZTNK7EAOWIV5B.json","view_paper":"https://pith.science/paper/R22L6ARN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1909.03087&json=true","fetch_graph":"https://pith.science/api/pith-number/R22L6ARN4PC6SZTNK7EAOWIV5B/graph.json","fetch_events":"https://pith.science/api/pith-number/R22L6ARN4PC6SZTNK7EAOWIV5B/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/R22L6ARN4PC6SZTNK7EAOWIV5B/action/timestamp_anchor","attest_storage":"https://pith.science/pith/R22L6ARN4PC6SZTNK7EAOWIV5B/action/storage_attestation","attest_author":"https://pith.science/pith/R22L6ARN4PC6SZTNK7EAOWIV5B/action/author_attestation","sign_citation":"https://pith.science/pith/R22L6ARN4PC6SZTNK7EAOWIV5B/action/citation_signature","submit_replication":"https://pith.science/pith/R22L6ARN4PC6SZTNK7EAOWIV5B/action/replication_record"}},"created_at":"2026-07-05T00:03:02.573950+00:00","updated_at":"2026-07-05T00:03:02.573950+00:00"}