{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:GRWTILAKENOCUJSERLSQ5ASMG7","short_pith_number":"pith:GRWTILAK","schema_version":"1.0","canonical_sha256":"346d342c0a235c2a26448ae50e824c37f033e880face6bd236f1c1641a4bd89c","source":{"kind":"arxiv","id":"2310.03991","version":2},"attestation_state":"computed","paper":{"title":"SemStamp: A Semantic Watermark with Paraphrastic Robustness for Text Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Abe Bohan Hou, Benjamin Van Durme, Daniel Khashabi, Hongwei Wang, Jingyu Zhang, Lingfeng Shen, Tianxing He, Yichen Wang, Yulia Tsvetkov, Yung-Sung Chuang","submitted_at":"2023-10-06T03:33:42Z","abstract_excerpt":"Existing watermarking algorithms are vulnerable to paraphrase attacks because of their token-level design. To address this issue, we propose SemStamp, a robust sentence-level semantic watermarking algorithm based on locality-sensitive hashing (LSH), which partitions the semantic space of sentences. The algorithm encodes and LSH-hashes a candidate sentence generated by an LLM, and conducts sentence-level rejection sampling until the sampled sentence falls in watermarked partitions in the semantic embedding space. A margin-based constraint is used to enhance its robustness. To show the advantage"},"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":"2310.03991","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-10-06T03:33:42Z","cross_cats_sorted":[],"title_canon_sha256":"9e44f8756a633792623275dd7f6fd68fd81a72b1785e1fe64984a2a4c06a1c8f","abstract_canon_sha256":"e8ff80116e0bccc69bf88757757c99885508e46fdabf65ada048161ebc122c5e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:10:18.481182Z","signature_b64":"VtAvEkLpIhwEBXQdYm0L86XskOEuLv5jEcpZh4XlPiw9XxKfVriKFaM4QsISLJTQSMfMSsLJmpUu5QY9WbDlCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"346d342c0a235c2a26448ae50e824c37f033e880face6bd236f1c1641a4bd89c","last_reissued_at":"2026-07-05T08:10:18.480716Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:10:18.480716Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SemStamp: A Semantic Watermark with Paraphrastic Robustness for Text Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Abe Bohan Hou, Benjamin Van Durme, Daniel Khashabi, Hongwei Wang, Jingyu Zhang, Lingfeng Shen, Tianxing He, Yichen Wang, Yulia Tsvetkov, Yung-Sung Chuang","submitted_at":"2023-10-06T03:33:42Z","abstract_excerpt":"Existing watermarking algorithms are vulnerable to paraphrase attacks because of their token-level design. To address this issue, we propose SemStamp, a robust sentence-level semantic watermarking algorithm based on locality-sensitive hashing (LSH), which partitions the semantic space of sentences. The algorithm encodes and LSH-hashes a candidate sentence generated by an LLM, and conducts sentence-level rejection sampling until the sampled sentence falls in watermarked partitions in the semantic embedding space. A margin-based constraint is used to enhance its robustness. To show the advantage"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.03991","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/2310.03991/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":"2310.03991","created_at":"2026-07-05T08:10:18.480773+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.03991v2","created_at":"2026-07-05T08:10:18.480773+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.03991","created_at":"2026-07-05T08:10:18.480773+00:00"},{"alias_kind":"pith_short_12","alias_value":"GRWTILAKENOC","created_at":"2026-07-05T08:10:18.480773+00:00"},{"alias_kind":"pith_short_16","alias_value":"GRWTILAKENOCUJSE","created_at":"2026-07-05T08:10:18.480773+00:00"},{"alias_kind":"pith_short_8","alias_value":"GRWTILAK","created_at":"2026-07-05T08:10:18.480773+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":12,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.04486","citing_title":"Global Sketch-Based Watermarking for Diffusion Language Models","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2605.25796","citing_title":"SAMark: A Self-Anchored Text Watermarking with Paragraph-Level Paraphrase Robustness","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27921","citing_title":"Show, Don't TELL: Explainable AI-Generated Text Detection","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06315","citing_title":"LLM Self-Recognition: Steering and Retrieving Activation Signatures","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12456","citing_title":"TextSeal: A Localized LLM Watermark for Provenance & Distillation Protection","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2508.11548","citing_title":"Copyright Protection for Large Language Models: A Survey of Methods, Challenges, and Trends","ref_index":58,"is_internal_anchor":false},{"citing_arxiv_id":"2510.18333","citing_title":"Position: LLM Watermarking Should Align Stakeholders' Incentives for Practical Adoption","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2605.13095","citing_title":"Watermarking Should Be Treated as a Monitoring Primitive","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.13095","citing_title":"Watermarking Should Be Treated as a Monitoring Primitive","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12456","citing_title":"TextSeal: A Localized LLM Watermark for Provenance & Distillation Protection","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05443","citing_title":"SLAM: Structural Linguistic Activation Marking for Language Models","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04305","citing_title":"SWAN: Semantic Watermarking with Abstract Meaning Representation","ref_index":46,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GRWTILAKENOCUJSERLSQ5ASMG7","json":"https://pith.science/pith/GRWTILAKENOCUJSERLSQ5ASMG7.json","graph_json":"https://pith.science/api/pith-number/GRWTILAKENOCUJSERLSQ5ASMG7/graph.json","events_json":"https://pith.science/api/pith-number/GRWTILAKENOCUJSERLSQ5ASMG7/events.json","paper":"https://pith.science/paper/GRWTILAK"},"agent_actions":{"view_html":"https://pith.science/pith/GRWTILAKENOCUJSERLSQ5ASMG7","download_json":"https://pith.science/pith/GRWTILAKENOCUJSERLSQ5ASMG7.json","view_paper":"https://pith.science/paper/GRWTILAK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.03991&json=true","fetch_graph":"https://pith.science/api/pith-number/GRWTILAKENOCUJSERLSQ5ASMG7/graph.json","fetch_events":"https://pith.science/api/pith-number/GRWTILAKENOCUJSERLSQ5ASMG7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GRWTILAKENOCUJSERLSQ5ASMG7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GRWTILAKENOCUJSERLSQ5ASMG7/action/storage_attestation","attest_author":"https://pith.science/pith/GRWTILAKENOCUJSERLSQ5ASMG7/action/author_attestation","sign_citation":"https://pith.science/pith/GRWTILAKENOCUJSERLSQ5ASMG7/action/citation_signature","submit_replication":"https://pith.science/pith/GRWTILAKENOCUJSERLSQ5ASMG7/action/replication_record"}},"created_at":"2026-07-05T08:10:18.480773+00:00","updated_at":"2026-07-05T08:10:18.480773+00:00"}