{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:DORALZQJRZ2ZL45UVV7UABRXI5","short_pith_number":"pith:DORALZQJ","schema_version":"1.0","canonical_sha256":"1ba205e6098e7595f3b4ad7f40063747763c12811f33a9a2eea940afb0b68b04","source":{"kind":"arxiv","id":"2110.06918","version":3},"attestation_state":"computed","paper":{"title":"Salient Phrase Aware Dense Retrieval: Can a Dense Retriever Imitate a Sparse One?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Anchit Gupta, Barlas O\\u{g}uz, Kushal Lakhotia, Patrick Lewis, Sonal Gupta, Stan Peshterliev, Wen-tau Yih, Xilun Chen, Yashar Mehdad","submitted_at":"2021-10-13T17:56:19Z","abstract_excerpt":"Despite their recent popularity and well-known advantages, dense retrievers still lag behind sparse methods such as BM25 in their ability to reliably match salient phrases and rare entities in the query and to generalize to out-of-domain data. It has been argued that this is an inherent limitation of dense models. We rebut this claim by introducing the Salient Phrase Aware Retriever (SPAR), a dense retriever with the lexical matching capacity of a sparse model. We show that a dense Lexical Model {\\Lambda} can be trained to imitate a sparse one, and SPAR is built by augmenting a standard dense "},"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":"2110.06918","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-10-13T17:56:19Z","cross_cats_sorted":["cs.IR","cs.LG"],"title_canon_sha256":"f4efc24724698a024ab5203ae573f5092fdb079a9758bfb5541283d1e96878f5","abstract_canon_sha256":"ed5ad7214a7687c3e1eac45678c827896f1228cb2a9c4883354e351ac817c064"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:15:22.276698Z","signature_b64":"eNigZIPLN+9nm5k0q0dZy49aOnpY2MVZfzYRHcuNA6U2v6EQy2JKGtQ+h9ZcPZR9IvCMCXmQ2xAtufVdIvSDBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1ba205e6098e7595f3b4ad7f40063747763c12811f33a9a2eea940afb0b68b04","last_reissued_at":"2026-07-05T05:15:22.276248Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:15:22.276248Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Salient Phrase Aware Dense Retrieval: Can a Dense Retriever Imitate a Sparse One?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Anchit Gupta, Barlas O\\u{g}uz, Kushal Lakhotia, Patrick Lewis, Sonal Gupta, Stan Peshterliev, Wen-tau Yih, Xilun Chen, Yashar Mehdad","submitted_at":"2021-10-13T17:56:19Z","abstract_excerpt":"Despite their recent popularity and well-known advantages, dense retrievers still lag behind sparse methods such as BM25 in their ability to reliably match salient phrases and rare entities in the query and to generalize to out-of-domain data. It has been argued that this is an inherent limitation of dense models. We rebut this claim by introducing the Salient Phrase Aware Retriever (SPAR), a dense retriever with the lexical matching capacity of a sparse model. We show that a dense Lexical Model {\\Lambda} can be trained to imitate a sparse one, and SPAR is built by augmenting a standard dense "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2110.06918","kind":"arxiv","version":3},"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/2110.06918/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":"2110.06918","created_at":"2026-07-05T05:15:22.276302+00:00"},{"alias_kind":"arxiv_version","alias_value":"2110.06918v3","created_at":"2026-07-05T05:15:22.276302+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2110.06918","created_at":"2026-07-05T05:15:22.276302+00:00"},{"alias_kind":"pith_short_12","alias_value":"DORALZQJRZ2Z","created_at":"2026-07-05T05:15:22.276302+00:00"},{"alias_kind":"pith_short_16","alias_value":"DORALZQJRZ2ZL45U","created_at":"2026-07-05T05:15:22.276302+00:00"},{"alias_kind":"pith_short_8","alias_value":"DORALZQJ","created_at":"2026-07-05T05:15:22.276302+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2112.09118","citing_title":"Unsupervised Dense Information Retrieval with Contrastive Learning","ref_index":123,"is_internal_anchor":false},{"citing_arxiv_id":"2112.09118","citing_title":"Unsupervised Dense Information Retrieval with Contrastive Learning","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2212.03533","citing_title":"Text Embeddings by Weakly-Supervised Contrastive Pre-training","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DORALZQJRZ2ZL45UVV7UABRXI5","json":"https://pith.science/pith/DORALZQJRZ2ZL45UVV7UABRXI5.json","graph_json":"https://pith.science/api/pith-number/DORALZQJRZ2ZL45UVV7UABRXI5/graph.json","events_json":"https://pith.science/api/pith-number/DORALZQJRZ2ZL45UVV7UABRXI5/events.json","paper":"https://pith.science/paper/DORALZQJ"},"agent_actions":{"view_html":"https://pith.science/pith/DORALZQJRZ2ZL45UVV7UABRXI5","download_json":"https://pith.science/pith/DORALZQJRZ2ZL45UVV7UABRXI5.json","view_paper":"https://pith.science/paper/DORALZQJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2110.06918&json=true","fetch_graph":"https://pith.science/api/pith-number/DORALZQJRZ2ZL45UVV7UABRXI5/graph.json","fetch_events":"https://pith.science/api/pith-number/DORALZQJRZ2ZL45UVV7UABRXI5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DORALZQJRZ2ZL45UVV7UABRXI5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DORALZQJRZ2ZL45UVV7UABRXI5/action/storage_attestation","attest_author":"https://pith.science/pith/DORALZQJRZ2ZL45UVV7UABRXI5/action/author_attestation","sign_citation":"https://pith.science/pith/DORALZQJRZ2ZL45UVV7UABRXI5/action/citation_signature","submit_replication":"https://pith.science/pith/DORALZQJRZ2ZL45UVV7UABRXI5/action/replication_record"}},"created_at":"2026-07-05T05:15:22.276302+00:00","updated_at":"2026-07-05T05:15:22.276302+00:00"}