{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:QVXCVK3UGBCPQVKVN2DJMRMWZA","short_pith_number":"pith:QVXCVK3U","schema_version":"1.0","canonical_sha256":"856e2aab743044f855556e86964596c812fb3e920f4b961a2dc673bc261e43fd","source":{"kind":"arxiv","id":"2306.00248","version":1},"attestation_state":"computed","paper":{"title":"TransAct: Transformer-based Realtime User Action Model for Recommendation at Pinterest","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.IR","authors_text":"Andrew Zhai, Dhruvil Deven Badani, Nazanin Farahpour, Neng Gu, Nikil Pancha, Pong Eksombatchai, Po-Wei Wang, Saurabh Vishwas Joshi, Xue Xia, Zhiyuan Zhang","submitted_at":"2023-05-31T23:45:29Z","abstract_excerpt":"Sequential models that encode user activity for next action prediction have become a popular design choice for building web-scale personalized recommendation systems. Traditional methods of sequential recommendation either utilize end-to-end learning on realtime user actions, or learn user representations separately in an offline batch-generated manner. This paper (1) presents Pinterest's ranking architecture for Homefeed, our personalized recommendation product and the largest engagement surface; (2) proposes TransAct, a sequential model that extracts users' short-term preferences from their "},"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":"2306.00248","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.IR","submitted_at":"2023-05-31T23:45:29Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"abceec165bcad0ecce57917c2849f33dda3cbbf42c4e001603efe3e8aaffbc5f","abstract_canon_sha256":"eaa57aaf34f209dcbe98a32b7d19a1ef443e2316d382fc8c146eb42d9669da1e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:14:22.579055Z","signature_b64":"AA7e9l4mZmwgPeDBULporR5KOjY7Alst6ZqdIe/ePUFvozwaUyY8d9zYF2PeBKEspjD9k23X9YhEu3BXHaT3AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"856e2aab743044f855556e86964596c812fb3e920f4b961a2dc673bc261e43fd","last_reissued_at":"2026-07-05T11:14:22.578655Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:14:22.578655Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TransAct: Transformer-based Realtime User Action Model for Recommendation at Pinterest","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.IR","authors_text":"Andrew Zhai, Dhruvil Deven Badani, Nazanin Farahpour, Neng Gu, Nikil Pancha, Pong Eksombatchai, Po-Wei Wang, Saurabh Vishwas Joshi, Xue Xia, Zhiyuan Zhang","submitted_at":"2023-05-31T23:45:29Z","abstract_excerpt":"Sequential models that encode user activity for next action prediction have become a popular design choice for building web-scale personalized recommendation systems. Traditional methods of sequential recommendation either utilize end-to-end learning on realtime user actions, or learn user representations separately in an offline batch-generated manner. This paper (1) presents Pinterest's ranking architecture for Homefeed, our personalized recommendation product and the largest engagement surface; (2) proposes TransAct, a sequential model that extracts users' short-term preferences from their "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2306.00248","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/2306.00248/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":"2306.00248","created_at":"2026-07-05T11:14:22.578710+00:00"},{"alias_kind":"arxiv_version","alias_value":"2306.00248v1","created_at":"2026-07-05T11:14:22.578710+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2306.00248","created_at":"2026-07-05T11:14:22.578710+00:00"},{"alias_kind":"pith_short_12","alias_value":"QVXCVK3UGBCP","created_at":"2026-07-05T11:14:22.578710+00:00"},{"alias_kind":"pith_short_16","alias_value":"QVXCVK3UGBCPQVKV","created_at":"2026-07-05T11:14:22.578710+00:00"},{"alias_kind":"pith_short_8","alias_value":"QVXCVK3U","created_at":"2026-07-05T11:14:22.578710+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.09027","citing_title":"A Contextual-Aware Position Encoding for Sequential Recommendation","ref_index":22,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QVXCVK3UGBCPQVKVN2DJMRMWZA","json":"https://pith.science/pith/QVXCVK3UGBCPQVKVN2DJMRMWZA.json","graph_json":"https://pith.science/api/pith-number/QVXCVK3UGBCPQVKVN2DJMRMWZA/graph.json","events_json":"https://pith.science/api/pith-number/QVXCVK3UGBCPQVKVN2DJMRMWZA/events.json","paper":"https://pith.science/paper/QVXCVK3U"},"agent_actions":{"view_html":"https://pith.science/pith/QVXCVK3UGBCPQVKVN2DJMRMWZA","download_json":"https://pith.science/pith/QVXCVK3UGBCPQVKVN2DJMRMWZA.json","view_paper":"https://pith.science/paper/QVXCVK3U","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2306.00248&json=true","fetch_graph":"https://pith.science/api/pith-number/QVXCVK3UGBCPQVKVN2DJMRMWZA/graph.json","fetch_events":"https://pith.science/api/pith-number/QVXCVK3UGBCPQVKVN2DJMRMWZA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QVXCVK3UGBCPQVKVN2DJMRMWZA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QVXCVK3UGBCPQVKVN2DJMRMWZA/action/storage_attestation","attest_author":"https://pith.science/pith/QVXCVK3UGBCPQVKVN2DJMRMWZA/action/author_attestation","sign_citation":"https://pith.science/pith/QVXCVK3UGBCPQVKVN2DJMRMWZA/action/citation_signature","submit_replication":"https://pith.science/pith/QVXCVK3UGBCPQVKVN2DJMRMWZA/action/replication_record"}},"created_at":"2026-07-05T11:14:22.578710+00:00","updated_at":"2026-07-05T11:14:22.578710+00:00"}