{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:3YAY5J2SGEHFO7QSZWKIP4QTNL","short_pith_number":"pith:3YAY5J2S","schema_version":"1.0","canonical_sha256":"de018ea752310e577e12cd9487f2136afb2f7d73fa73e9af2c64d1b56933dc9d","source":{"kind":"arxiv","id":"2305.17721","version":1},"attestation_state":"computed","paper":{"title":"Rethinking Masked Language Modeling for Chinese Spelling Correction","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hai Zhao, Hongqiu Wu, Shaohua Zhang, Yuchen Zhang","submitted_at":"2023-05-28T13:19:12Z","abstract_excerpt":"In this paper, we study Chinese Spelling Correction (CSC) as a joint decision made by two separate models: a language model and an error model. Through empirical analysis, we find that fine-tuning BERT tends to over-fit the error model while under-fit the language model, resulting in poor generalization to out-of-distribution error patterns. Given that BERT is the backbone of most CSC models, this phenomenon has a significant negative impact. To address this issue, we are releasing a multi-domain benchmark LEMON, with higher quality and diversity than existing benchmarks, to allow a comprehens"},"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":"2305.17721","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-05-28T13:19:12Z","cross_cats_sorted":[],"title_canon_sha256":"232b347e6e8dbec943dc9a5fc8c89ae41897a51ce61d3d9c250ae0d1ceaf891f","abstract_canon_sha256":"e013af93bdae3a2a36f215705485cbb0d3d56c9abe5e09f7fdf2715432b4bdf4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:14:47.790530Z","signature_b64":"SWn7SLWZJQhFQnZPnK3anPU9J2PTUTfZPes3x25CfhODGCXhJQzzmHvo6jJGPegWWQnlxZ0wmlMe/VzAAbXvBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"de018ea752310e577e12cd9487f2136afb2f7d73fa73e9af2c64d1b56933dc9d","last_reissued_at":"2026-07-05T06:14:47.790065Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:14:47.790065Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Rethinking Masked Language Modeling for Chinese Spelling Correction","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hai Zhao, Hongqiu Wu, Shaohua Zhang, Yuchen Zhang","submitted_at":"2023-05-28T13:19:12Z","abstract_excerpt":"In this paper, we study Chinese Spelling Correction (CSC) as a joint decision made by two separate models: a language model and an error model. Through empirical analysis, we find that fine-tuning BERT tends to over-fit the error model while under-fit the language model, resulting in poor generalization to out-of-distribution error patterns. Given that BERT is the backbone of most CSC models, this phenomenon has a significant negative impact. To address this issue, we are releasing a multi-domain benchmark LEMON, with higher quality and diversity than existing benchmarks, to allow a comprehens"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.17721","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/2305.17721/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":"2305.17721","created_at":"2026-07-05T06:14:47.790114+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.17721v1","created_at":"2026-07-05T06:14:47.790114+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.17721","created_at":"2026-07-05T06:14:47.790114+00:00"},{"alias_kind":"pith_short_12","alias_value":"3YAY5J2SGEHF","created_at":"2026-07-05T06:14:47.790114+00:00"},{"alias_kind":"pith_short_16","alias_value":"3YAY5J2SGEHFO7QS","created_at":"2026-07-05T06:14:47.790114+00:00"},{"alias_kind":"pith_short_8","alias_value":"3YAY5J2S","created_at":"2026-07-05T06:14:47.790114+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2412.12832","citing_title":"DSGram: Dynamic Weighting Sub-Metrics for Grammatical Error Correction in the Era of Large Language Models","ref_index":26,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3YAY5J2SGEHFO7QSZWKIP4QTNL","json":"https://pith.science/pith/3YAY5J2SGEHFO7QSZWKIP4QTNL.json","graph_json":"https://pith.science/api/pith-number/3YAY5J2SGEHFO7QSZWKIP4QTNL/graph.json","events_json":"https://pith.science/api/pith-number/3YAY5J2SGEHFO7QSZWKIP4QTNL/events.json","paper":"https://pith.science/paper/3YAY5J2S"},"agent_actions":{"view_html":"https://pith.science/pith/3YAY5J2SGEHFO7QSZWKIP4QTNL","download_json":"https://pith.science/pith/3YAY5J2SGEHFO7QSZWKIP4QTNL.json","view_paper":"https://pith.science/paper/3YAY5J2S","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.17721&json=true","fetch_graph":"https://pith.science/api/pith-number/3YAY5J2SGEHFO7QSZWKIP4QTNL/graph.json","fetch_events":"https://pith.science/api/pith-number/3YAY5J2SGEHFO7QSZWKIP4QTNL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3YAY5J2SGEHFO7QSZWKIP4QTNL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3YAY5J2SGEHFO7QSZWKIP4QTNL/action/storage_attestation","attest_author":"https://pith.science/pith/3YAY5J2SGEHFO7QSZWKIP4QTNL/action/author_attestation","sign_citation":"https://pith.science/pith/3YAY5J2SGEHFO7QSZWKIP4QTNL/action/citation_signature","submit_replication":"https://pith.science/pith/3YAY5J2SGEHFO7QSZWKIP4QTNL/action/replication_record"}},"created_at":"2026-07-05T06:14:47.790114+00:00","updated_at":"2026-07-05T06:14:47.790114+00:00"}