{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7TNII6BVBTYFNP2S57V2X2GYHO","short_pith_number":"pith:7TNII6BV","schema_version":"1.0","canonical_sha256":"fcda8478350cf056bf52efebabe8d83b8388226f0cc23f8c582a0fd5f21d044d","source":{"kind":"arxiv","id":"2608.02738","version":1},"attestation_state":"computed","paper":{"title":"Knowledge-Geometry Decoupling: Refreshable Pretrained Transfer for Streaming Recommendation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Anxiang Zeng, Bangyang Hong, Cong Fu, Guidong Lei, Hui Li, Kangle Wu, Kun Wang, Yabo Ni, Yuhong Chen, Yuxuan Zhu, Yu Zhao, Zhiluohan Guo, Zixuan Wang","submitted_at":"2026-08-03T18:00:05Z","abstract_excerpt":"Industrial recommenders increasingly adopt the pretrain-then-transfer paradigm, yet behavioral distribution drift raises two questions: what to learn from behavior sequences, and how to transfer the learned knowledge while the pretrained model is continually refreshed. To resolve them, we propose Knowledge-Geometry Decoupling (KGD). For what to learn, conventional next-token prediction treats adjacency as dependency and may encode spurious transitions across unrelated sessions. We introduce Behavioral Multi-Token Prediction (BMTP) to retain only collaboratively or semantically related future i"},"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":"2608.02738","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.IR","submitted_at":"2026-08-03T18:00:05Z","cross_cats_sorted":[],"title_canon_sha256":"0e23786da5cf57c6bd8740db0de078d3536f8560bbb79718bcf7e79478a58aca","abstract_canon_sha256":"45fd4c4d1060b78c14bda8a0704c7cf6c74c0eb7a2d6592d3802fea629bd5228"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-05T00:39:44.243049Z","signature_b64":"YluP1sMMMe3zI2avzbBsqjPzwcALrDJec0dxDxrVfZFzNIjhEfYesKN2t0AQqEnzK0baffQGOEnhUTu4mIDDBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fcda8478350cf056bf52efebabe8d83b8388226f0cc23f8c582a0fd5f21d044d","last_reissued_at":"2026-08-05T00:39:44.240437Z","signature_status":"signed_v1","first_computed_at":"2026-08-05T00:39:44.240437Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Knowledge-Geometry Decoupling: Refreshable Pretrained Transfer for Streaming Recommendation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Anxiang Zeng, Bangyang Hong, Cong Fu, Guidong Lei, Hui Li, Kangle Wu, Kun Wang, Yabo Ni, Yuhong Chen, Yuxuan Zhu, Yu Zhao, Zhiluohan Guo, Zixuan Wang","submitted_at":"2026-08-03T18:00:05Z","abstract_excerpt":"Industrial recommenders increasingly adopt the pretrain-then-transfer paradigm, yet behavioral distribution drift raises two questions: what to learn from behavior sequences, and how to transfer the learned knowledge while the pretrained model is continually refreshed. To resolve them, we propose Knowledge-Geometry Decoupling (KGD). For what to learn, conventional next-token prediction treats adjacency as dependency and may encode spurious transitions across unrelated sessions. We introduce Behavioral Multi-Token Prediction (BMTP) to retain only collaboratively or semantically related future i"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.02738","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/2608.02738/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":"2608.02738","created_at":"2026-08-05T00:39:44.241454+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.02738v1","created_at":"2026-08-05T00:39:44.241454+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.02738","created_at":"2026-08-05T00:39:44.241454+00:00"},{"alias_kind":"pith_short_12","alias_value":"7TNII6BVBTYF","created_at":"2026-08-05T00:39:44.241454+00:00"},{"alias_kind":"pith_short_16","alias_value":"7TNII6BVBTYFNP2S","created_at":"2026-08-05T00:39:44.241454+00:00"},{"alias_kind":"pith_short_8","alias_value":"7TNII6BV","created_at":"2026-08-05T00:39:44.241454+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/7TNII6BVBTYFNP2S57V2X2GYHO","json":"https://pith.science/pith/7TNII6BVBTYFNP2S57V2X2GYHO.json","graph_json":"https://pith.science/api/pith-number/7TNII6BVBTYFNP2S57V2X2GYHO/graph.json","events_json":"https://pith.science/api/pith-number/7TNII6BVBTYFNP2S57V2X2GYHO/events.json","paper":"https://pith.science/paper/7TNII6BV"},"agent_actions":{"view_html":"https://pith.science/pith/7TNII6BVBTYFNP2S57V2X2GYHO","download_json":"https://pith.science/pith/7TNII6BVBTYFNP2S57V2X2GYHO.json","view_paper":"https://pith.science/paper/7TNII6BV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.02738&json=true","fetch_graph":"https://pith.science/api/pith-number/7TNII6BVBTYFNP2S57V2X2GYHO/graph.json","fetch_events":"https://pith.science/api/pith-number/7TNII6BVBTYFNP2S57V2X2GYHO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7TNII6BVBTYFNP2S57V2X2GYHO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7TNII6BVBTYFNP2S57V2X2GYHO/action/storage_attestation","attest_author":"https://pith.science/pith/7TNII6BVBTYFNP2S57V2X2GYHO/action/author_attestation","sign_citation":"https://pith.science/pith/7TNII6BVBTYFNP2S57V2X2GYHO/action/citation_signature","submit_replication":"https://pith.science/pith/7TNII6BVBTYFNP2S57V2X2GYHO/action/replication_record"}},"created_at":"2026-08-05T00:39:44.241454+00:00","updated_at":"2026-08-05T00:39:44.241454+00:00"}