{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:763DPQM5PXK6B4NNUHRUXKK3ST","short_pith_number":"pith:763DPQM5","schema_version":"1.0","canonical_sha256":"ffb637c19d7dd5e0f1ada1e34ba95b94ef4e7ed59ce62da3abdd09281578b349","source":{"kind":"arxiv","id":"2503.05888","version":2},"attestation_state":"computed","paper":{"title":"QG-SMS: Enhancing Test Item Analysis via Student Modeling and Simulation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CY","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bang Nguyen, Lawrence Angrave, Meng Jiang, Mengxia Yu, Tingting Du","submitted_at":"2025-03-07T19:21:59Z","abstract_excerpt":"While the Question Generation (QG) task has been increasingly adopted in educational assessments, its evaluation remains limited by approaches that lack a clear connection to the educational values of test items. In this work, we introduce test item analysis, a method frequently used by educators to assess test question quality, into QG evaluation. Specifically, we construct pairs of candidate questions that differ in quality across dimensions such as topic coverage, item difficulty, item discrimination, and distractor efficiency. We then examine whether existing QG evaluation approaches can e"},"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":"2503.05888","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-03-07T19:21:59Z","cross_cats_sorted":["cs.AI","cs.CY","cs.LG"],"title_canon_sha256":"4f70de4c543970dc45b0561631c4532737a4e064ad7150d68ea0c3d0664f0c44","abstract_canon_sha256":"de057cdbf0d6c03ec070341df30c7f5b3e292c54c54520d49a532d4a0cff4459"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:24:15.899736Z","signature_b64":"CNXRutIMrfsMIfPs+K/icChUj2FPNl43xJTH4Xr9MoSjBebJOYZfxndpvFigNb3NW9Zjx6aeBdN775ft4OtYDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ffb637c19d7dd5e0f1ada1e34ba95b94ef4e7ed59ce62da3abdd09281578b349","last_reissued_at":"2026-07-05T11:24:15.899165Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:24:15.899165Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"QG-SMS: Enhancing Test Item Analysis via Student Modeling and Simulation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CY","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bang Nguyen, Lawrence Angrave, Meng Jiang, Mengxia Yu, Tingting Du","submitted_at":"2025-03-07T19:21:59Z","abstract_excerpt":"While the Question Generation (QG) task has been increasingly adopted in educational assessments, its evaluation remains limited by approaches that lack a clear connection to the educational values of test items. In this work, we introduce test item analysis, a method frequently used by educators to assess test question quality, into QG evaluation. Specifically, we construct pairs of candidate questions that differ in quality across dimensions such as topic coverage, item difficulty, item discrimination, and distractor efficiency. We then examine whether existing QG evaluation approaches can e"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.05888","kind":"arxiv","version":2},"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/2503.05888/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":"2503.05888","created_at":"2026-07-05T11:24:15.899225+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.05888v2","created_at":"2026-07-05T11:24:15.899225+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.05888","created_at":"2026-07-05T11:24:15.899225+00:00"},{"alias_kind":"pith_short_12","alias_value":"763DPQM5PXK6","created_at":"2026-07-05T11:24:15.899225+00:00"},{"alias_kind":"pith_short_16","alias_value":"763DPQM5PXK6B4NN","created_at":"2026-07-05T11:24:15.899225+00:00"},{"alias_kind":"pith_short_8","alias_value":"763DPQM5","created_at":"2026-07-05T11:24:15.899225+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/763DPQM5PXK6B4NNUHRUXKK3ST","json":"https://pith.science/pith/763DPQM5PXK6B4NNUHRUXKK3ST.json","graph_json":"https://pith.science/api/pith-number/763DPQM5PXK6B4NNUHRUXKK3ST/graph.json","events_json":"https://pith.science/api/pith-number/763DPQM5PXK6B4NNUHRUXKK3ST/events.json","paper":"https://pith.science/paper/763DPQM5"},"agent_actions":{"view_html":"https://pith.science/pith/763DPQM5PXK6B4NNUHRUXKK3ST","download_json":"https://pith.science/pith/763DPQM5PXK6B4NNUHRUXKK3ST.json","view_paper":"https://pith.science/paper/763DPQM5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.05888&json=true","fetch_graph":"https://pith.science/api/pith-number/763DPQM5PXK6B4NNUHRUXKK3ST/graph.json","fetch_events":"https://pith.science/api/pith-number/763DPQM5PXK6B4NNUHRUXKK3ST/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/763DPQM5PXK6B4NNUHRUXKK3ST/action/timestamp_anchor","attest_storage":"https://pith.science/pith/763DPQM5PXK6B4NNUHRUXKK3ST/action/storage_attestation","attest_author":"https://pith.science/pith/763DPQM5PXK6B4NNUHRUXKK3ST/action/author_attestation","sign_citation":"https://pith.science/pith/763DPQM5PXK6B4NNUHRUXKK3ST/action/citation_signature","submit_replication":"https://pith.science/pith/763DPQM5PXK6B4NNUHRUXKK3ST/action/replication_record"}},"created_at":"2026-07-05T11:24:15.899225+00:00","updated_at":"2026-07-05T11:24:15.899225+00:00"}