{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2018:MZHLXLH757Q47YBAGWPKTOCXX2","short_pith_number":"pith:MZHLXLH7","schema_version":"1.0","canonical_sha256":"664ebbacffefe1cfe020359ea9b857be95b098e2b169588b87b6e7ab04475fc8","source":{"kind":"arxiv","id":"1810.04882","version":7},"attestation_state":"computed","paper":{"title":"Towards Understanding Linear Word Analogies","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"David Duvenaud, Graeme Hirst, Kawin Ethayarajh","submitted_at":"2018-10-11T08:08:40Z","abstract_excerpt":"A surprising property of word vectors is that word analogies can often be solved with vector arithmetic. However, it is unclear why arithmetic operators correspond to non-linear embedding models such as skip-gram with negative sampling (SGNS). We provide a formal explanation of this phenomenon without making the strong assumptions that past theories have made about the vector space and word distribution. Our theory has several implications. Past work has conjectured that linear substructures exist in vector spaces because relations can be represented as ratios; we prove that this holds for SGN"},"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":"1810.04882","kind":"arxiv","version":7},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2018-10-11T08:08:40Z","cross_cats_sorted":[],"title_canon_sha256":"033630ce372d90477ed5365d4a497a6e6ff8043c866685ef54e8663870efc42d","abstract_canon_sha256":"ecbc3a7e193c71e73611ac0dcbf71abc73a75eefde63633c962a7a12aa6dd937"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-04T23:53:10.414535Z","signature_b64":"m+Z85GPQxWO6zT9HrmrvAb1ssq0jbHFzbnXv+1wdkPQdhlR8fPiyXJS7v4WD0/1pOJjhs0bQR1AehssmnVlqBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"664ebbacffefe1cfe020359ea9b857be95b098e2b169588b87b6e7ab04475fc8","last_reissued_at":"2026-07-04T23:53:10.414053Z","signature_status":"signed_v1","first_computed_at":"2026-07-04T23:53:10.414053Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards Understanding Linear Word Analogies","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"David Duvenaud, Graeme Hirst, Kawin Ethayarajh","submitted_at":"2018-10-11T08:08:40Z","abstract_excerpt":"A surprising property of word vectors is that word analogies can often be solved with vector arithmetic. However, it is unclear why arithmetic operators correspond to non-linear embedding models such as skip-gram with negative sampling (SGNS). We provide a formal explanation of this phenomenon without making the strong assumptions that past theories have made about the vector space and word distribution. Our theory has several implications. Past work has conjectured that linear substructures exist in vector spaces because relations can be represented as ratios; we prove that this holds for SGN"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1810.04882","kind":"arxiv","version":7},"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/1810.04882/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":"1810.04882","created_at":"2026-07-04T23:53:10.414110+00:00"},{"alias_kind":"arxiv_version","alias_value":"1810.04882v7","created_at":"2026-07-04T23:53:10.414110+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1810.04882","created_at":"2026-07-04T23:53:10.414110+00:00"},{"alias_kind":"pith_short_12","alias_value":"MZHLXLH757Q4","created_at":"2026-07-04T23:53:10.414110+00:00"},{"alias_kind":"pith_short_16","alias_value":"MZHLXLH757Q47YBA","created_at":"2026-07-04T23:53:10.414110+00:00"},{"alias_kind":"pith_short_8","alias_value":"MZHLXLH7","created_at":"2026-07-04T23:53:10.414110+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.07568","citing_title":"A Systematic Study of Behavioral Cloning for Scientific Data Annotation","ref_index":87,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MZHLXLH757Q47YBAGWPKTOCXX2","json":"https://pith.science/pith/MZHLXLH757Q47YBAGWPKTOCXX2.json","graph_json":"https://pith.science/api/pith-number/MZHLXLH757Q47YBAGWPKTOCXX2/graph.json","events_json":"https://pith.science/api/pith-number/MZHLXLH757Q47YBAGWPKTOCXX2/events.json","paper":"https://pith.science/paper/MZHLXLH7"},"agent_actions":{"view_html":"https://pith.science/pith/MZHLXLH757Q47YBAGWPKTOCXX2","download_json":"https://pith.science/pith/MZHLXLH757Q47YBAGWPKTOCXX2.json","view_paper":"https://pith.science/paper/MZHLXLH7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1810.04882&json=true","fetch_graph":"https://pith.science/api/pith-number/MZHLXLH757Q47YBAGWPKTOCXX2/graph.json","fetch_events":"https://pith.science/api/pith-number/MZHLXLH757Q47YBAGWPKTOCXX2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MZHLXLH757Q47YBAGWPKTOCXX2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MZHLXLH757Q47YBAGWPKTOCXX2/action/storage_attestation","attest_author":"https://pith.science/pith/MZHLXLH757Q47YBAGWPKTOCXX2/action/author_attestation","sign_citation":"https://pith.science/pith/MZHLXLH757Q47YBAGWPKTOCXX2/action/citation_signature","submit_replication":"https://pith.science/pith/MZHLXLH757Q47YBAGWPKTOCXX2/action/replication_record"}},"created_at":"2026-07-04T23:53:10.414110+00:00","updated_at":"2026-07-04T23:53:10.414110+00:00"}