{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:RDARYRAZC2CDM7MRQCMKWELPC3","short_pith_number":"pith:RDARYRAZ","schema_version":"1.0","canonical_sha256":"88c11c44191684367d918098ab116f16c6709bcb73436be114015b3172813a13","source":{"kind":"arxiv","id":"2202.06317","version":2},"attestation_state":"computed","paper":{"title":"Off-Policy Evaluation for Large Action Spaces via Embeddings","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Thorsten Joachims, Yuta Saito","submitted_at":"2022-02-13T14:00:09Z","abstract_excerpt":"Off-policy evaluation (OPE) in contextual bandits has seen rapid adoption in real-world systems, since it enables offline evaluation of new policies using only historic log data. Unfortunately, when the number of actions is large, existing OPE estimators -- most of which are based on inverse propensity score weighting -- degrade severely and can suffer from extreme bias and variance. This foils the use of OPE in many applications from recommender systems to language models. To overcome this issue, we propose a new OPE estimator that leverages marginalized importance weights when action embeddi"},"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":"2202.06317","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-02-13T14:00:09Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"39902ee2724f71cc6e9a12937425d7c513844e7e42ab9782129285209ed9079d","abstract_canon_sha256":"cecaca42350bcde254f25fc655bc2a098f4ba8a6b5ff2bcc626b7eb014e21327"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:32:14.633333Z","signature_b64":"V/OAgJuBsSqfvI4c/7HixprpBYN2OrbsLtktTKRBlt5VGxNcRz83WzNrMOUKsfGKiOtQas8BPHxBmjCW6dy8CA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"88c11c44191684367d918098ab116f16c6709bcb73436be114015b3172813a13","last_reissued_at":"2026-07-05T04:32:14.632850Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:32:14.632850Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Off-Policy Evaluation for Large Action Spaces via Embeddings","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Thorsten Joachims, Yuta Saito","submitted_at":"2022-02-13T14:00:09Z","abstract_excerpt":"Off-policy evaluation (OPE) in contextual bandits has seen rapid adoption in real-world systems, since it enables offline evaluation of new policies using only historic log data. Unfortunately, when the number of actions is large, existing OPE estimators -- most of which are based on inverse propensity score weighting -- degrade severely and can suffer from extreme bias and variance. This foils the use of OPE in many applications from recommender systems to language models. To overcome this issue, we propose a new OPE estimator that leverages marginalized importance weights when action embeddi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.06317","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/2202.06317/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":"2202.06317","created_at":"2026-07-05T04:32:14.632905+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.06317v2","created_at":"2026-07-05T04:32:14.632905+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.06317","created_at":"2026-07-05T04:32:14.632905+00:00"},{"alias_kind":"pith_short_12","alias_value":"RDARYRAZC2CD","created_at":"2026-07-05T04:32:14.632905+00:00"},{"alias_kind":"pith_short_16","alias_value":"RDARYRAZC2CDM7MR","created_at":"2026-07-05T04:32:14.632905+00:00"},{"alias_kind":"pith_short_8","alias_value":"RDARYRAZ","created_at":"2026-07-05T04:32:14.632905+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2509.03456","citing_title":"Off-Policy Learning in Large Action Spaces: Optimization Matters More Than Estimation","ref_index":24,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RDARYRAZC2CDM7MRQCMKWELPC3","json":"https://pith.science/pith/RDARYRAZC2CDM7MRQCMKWELPC3.json","graph_json":"https://pith.science/api/pith-number/RDARYRAZC2CDM7MRQCMKWELPC3/graph.json","events_json":"https://pith.science/api/pith-number/RDARYRAZC2CDM7MRQCMKWELPC3/events.json","paper":"https://pith.science/paper/RDARYRAZ"},"agent_actions":{"view_html":"https://pith.science/pith/RDARYRAZC2CDM7MRQCMKWELPC3","download_json":"https://pith.science/pith/RDARYRAZC2CDM7MRQCMKWELPC3.json","view_paper":"https://pith.science/paper/RDARYRAZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.06317&json=true","fetch_graph":"https://pith.science/api/pith-number/RDARYRAZC2CDM7MRQCMKWELPC3/graph.json","fetch_events":"https://pith.science/api/pith-number/RDARYRAZC2CDM7MRQCMKWELPC3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RDARYRAZC2CDM7MRQCMKWELPC3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RDARYRAZC2CDM7MRQCMKWELPC3/action/storage_attestation","attest_author":"https://pith.science/pith/RDARYRAZC2CDM7MRQCMKWELPC3/action/author_attestation","sign_citation":"https://pith.science/pith/RDARYRAZC2CDM7MRQCMKWELPC3/action/citation_signature","submit_replication":"https://pith.science/pith/RDARYRAZC2CDM7MRQCMKWELPC3/action/replication_record"}},"created_at":"2026-07-05T04:32:14.632905+00:00","updated_at":"2026-07-05T04:32:14.632905+00:00"}