{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:62YH447NMEDPS5WIQYNU7BCZLX","short_pith_number":"pith:62YH447N","schema_version":"1.0","canonical_sha256":"f6b07e73ed6106f976c8861b4f84595ddf6668247360deee76b8d7dcf1802349","source":{"kind":"arxiv","id":"2111.14592","version":8},"attestation_state":"computed","paper":{"title":"GALAXY: A Generative Pre-trained Model for Task-Oriented Dialog with Semi-Supervised Learning and Explicit Policy Injection","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dermot Liu, Fei Huang, Jian Sun, Luo Si, Min Yang, Peng Jiang, Wanwei He, Yinhe Zheng, Yinpei Dai, Yongbin Li, Yuchuan Wu, Zheng Cao","submitted_at":"2021-11-29T15:24:36Z","abstract_excerpt":"Pre-trained models have proved to be powerful in enhancing task-oriented dialog systems. However, current pre-training methods mainly focus on enhancing dialog understanding and generation tasks while neglecting the exploitation of dialog policy. In this paper, we propose GALAXY, a novel pre-trained dialog model that explicitly learns dialog policy from limited labeled dialogs and large-scale unlabeled dialog corpora via semi-supervised learning. Specifically, we introduce a dialog act prediction task for policy optimization during pre-training and employ a consistency regularization term to r"},"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":"2111.14592","kind":"arxiv","version":8},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2021-11-29T15:24:36Z","cross_cats_sorted":[],"title_canon_sha256":"dc753e80e568adf95147cce3bb65985fcc28f4c3c4189ea1d4a73833cc4209d9","abstract_canon_sha256":"d9c9196ea6ddd0cec74207648d9e9a4eca949abbdef5e22308be44d36173243b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:09:16.678550Z","signature_b64":"SK4ShyA9FpsG1BuaNfcDzCez0EaDZur4G9adh5MYPbkexNNQFOHXq+QD1VylzpRJlXWMI2DVk0Iz84/5PPUFDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f6b07e73ed6106f976c8861b4f84595ddf6668247360deee76b8d7dcf1802349","last_reissued_at":"2026-07-05T04:09:16.678080Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:09:16.678080Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GALAXY: A Generative Pre-trained Model for Task-Oriented Dialog with Semi-Supervised Learning and Explicit Policy Injection","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dermot Liu, Fei Huang, Jian Sun, Luo Si, Min Yang, Peng Jiang, Wanwei He, Yinhe Zheng, Yinpei Dai, Yongbin Li, Yuchuan Wu, Zheng Cao","submitted_at":"2021-11-29T15:24:36Z","abstract_excerpt":"Pre-trained models have proved to be powerful in enhancing task-oriented dialog systems. However, current pre-training methods mainly focus on enhancing dialog understanding and generation tasks while neglecting the exploitation of dialog policy. In this paper, we propose GALAXY, a novel pre-trained dialog model that explicitly learns dialog policy from limited labeled dialogs and large-scale unlabeled dialog corpora via semi-supervised learning. Specifically, we introduce a dialog act prediction task for policy optimization during pre-training and employ a consistency regularization term to r"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2111.14592","kind":"arxiv","version":8},"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/2111.14592/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":"2111.14592","created_at":"2026-07-05T04:09:16.678139+00:00"},{"alias_kind":"arxiv_version","alias_value":"2111.14592v8","created_at":"2026-07-05T04:09:16.678139+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2111.14592","created_at":"2026-07-05T04:09:16.678139+00:00"},{"alias_kind":"pith_short_12","alias_value":"62YH447NMEDP","created_at":"2026-07-05T04:09:16.678139+00:00"},{"alias_kind":"pith_short_16","alias_value":"62YH447NMEDPS5WI","created_at":"2026-07-05T04:09:16.678139+00:00"},{"alias_kind":"pith_short_8","alias_value":"62YH447N","created_at":"2026-07-05T04:09:16.678139+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2603.22003","citing_title":"VP-VLA: Visual Prompting as an Interface for Vision-Language-Action Models","ref_index":11,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/62YH447NMEDPS5WIQYNU7BCZLX","json":"https://pith.science/pith/62YH447NMEDPS5WIQYNU7BCZLX.json","graph_json":"https://pith.science/api/pith-number/62YH447NMEDPS5WIQYNU7BCZLX/graph.json","events_json":"https://pith.science/api/pith-number/62YH447NMEDPS5WIQYNU7BCZLX/events.json","paper":"https://pith.science/paper/62YH447N"},"agent_actions":{"view_html":"https://pith.science/pith/62YH447NMEDPS5WIQYNU7BCZLX","download_json":"https://pith.science/pith/62YH447NMEDPS5WIQYNU7BCZLX.json","view_paper":"https://pith.science/paper/62YH447N","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2111.14592&json=true","fetch_graph":"https://pith.science/api/pith-number/62YH447NMEDPS5WIQYNU7BCZLX/graph.json","fetch_events":"https://pith.science/api/pith-number/62YH447NMEDPS5WIQYNU7BCZLX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/62YH447NMEDPS5WIQYNU7BCZLX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/62YH447NMEDPS5WIQYNU7BCZLX/action/storage_attestation","attest_author":"https://pith.science/pith/62YH447NMEDPS5WIQYNU7BCZLX/action/author_attestation","sign_citation":"https://pith.science/pith/62YH447NMEDPS5WIQYNU7BCZLX/action/citation_signature","submit_replication":"https://pith.science/pith/62YH447NMEDPS5WIQYNU7BCZLX/action/replication_record"}},"created_at":"2026-07-05T04:09:16.678139+00:00","updated_at":"2026-07-05T04:09:16.678139+00:00"}