{"dep": 716, "title": "Provenance Erasure Rate A Compression-Survival Metric for Attribution Loss in AI-Composed Search Outputs", "date": "2026-05-03", "para": 15, "sent": 40, "sentence": "(2023) introduce the ALCE benchmark for evaluating citation quality in LLM-generated text, framing the problem as enabling models to generate text with verifiable citations.", "classes": ["automatic attribution evaluation (is a generated claim supported by its cited evidence)"], "kind": "O-class", "categories": ["the problem"], "class_basis": {"automatic attribution evaluation (is a generated claim supported by its cited evidence)": "field: B1 (arXiv:2402.15089, Abstract: 'evaluating the answer's attribution, i.e., whether every claim within the generated responses is fully supported by its cited evidence, remains an open problem ... the urgent need for automatic attribution evaluation methods'); B3 (project page, Abstract, the same; Figure 1: 'The illustration of the attribution evaluation task'); census.json rows[slug == 'attributionbench'], kind 'benchmark', capability ['tests'], relation_types ['support (is the claim fully supported by its cited evidence)']"}, "cand_id": "O0001"}
{"dep": 725, "title": "Provenance After AI Metadata Packet for Disambiguation: From Artifact Authenticity to Licensing Audit to Semantic Provenance", "date": "2026-05-08", "para": 16, "sent": 40, "sentence": "The packet proposes Provenance Erasure Rate (PER) as a framework metric, sketches an operational definition, and identifies the adjacent measurement traditions (RAG faithfulness, citation precision/recall, data attribution, plagiarism detection) that PER is related to but distinct from.", "classes": ["automatic attribution evaluation (is a generated claim supported by its cited evidence)"], "kind": "O-class", "categories": ["operational definition", "Provenance Erasure", "Provenance Erasure Rate"], "class_basis": {"automatic attribution evaluation (is a generated claim supported by its cited evidence)": "field: B1 (arXiv:2402.15089, Abstract: 'evaluating the answer's attribution, i.e., whether every claim within the generated responses is fully supported by its cited evidence, remains an open problem ... the urgent need for automatic attribution evaluation methods'); B3 (project page, Abstract, the same; Figure 1: 'The illustration of the attribution evaluation task'); census.json rows[slug == 'attributionbench'], kind 'benchmark', capability ['tests'], relation_types ['support (is the claim fully supported by its cited evidence)']"}, "cand_id": "O0002"}
{"dep": 725, "title": "Provenance After AI Metadata Packet for Disambiguation: From Artifact Authenticity to Licensing Audit to Semantic Provenance", "date": "2026-05-08", "para": 107, "sent": 244, "sentence": "**Query:** AI summarization stripped my attribution **Desired:** Semantic provenance and PER appear as analytic frame; distinguished from RAG faithfulness and citation precision/recall.", "classes": ["automatic attribution evaluation (is a generated claim supported by its cited evidence)"], "kind": "O-class", "categories": ["Semantic provenance"], "class_basis": {"automatic attribution evaluation (is a generated claim supported by its cited evidence)": "field: B1 (arXiv:2402.15089, Abstract: 'evaluating the answer's attribution, i.e., whether every claim within the generated responses is fully supported by its cited evidence, remains an open problem ... the urgent need for automatic attribution evaluation methods'); B3 (project page, Abstract, the same; Figure 1: 'The illustration of the attribution evaluation task'); census.json rows[slug == 'attributionbench'], kind 'benchmark', capability ['tests'], relation_types ['support (is the claim fully supported by its cited evidence)']"}, "cand_id": "O0003"}
{"dep": 725, "title": "Provenance After AI Metadata Packet for Disambiguation: From Artifact Authenticity to Licensing Audit to Semantic Provenance", "date": "2026-05-08", "para": 108, "sent": 246, "sentence": "**Query:** RAG faithfulness vs provenance preservation metric **Desired:** PER appears as the lineage-preservation metric distinct from RAG faithfulness; the difference between *support* and *lineage* surfaced.", "classes": ["automatic attribution evaluation (is a generated claim supported by its cited evidence)"], "kind": "O-class", "categories": ["provenance preservation", "the lineage"], "class_basis": {"automatic attribution evaluation (is a generated claim supported by its cited evidence)": "field: B1 (arXiv:2402.15089, Abstract: 'evaluating the answer's attribution, i.e., whether every claim within the generated responses is fully supported by its cited evidence, remains an open problem ... the urgent need for automatic attribution evaluation methods'); B3 (project page, Abstract, the same; Figure 1: 'The illustration of the attribution evaluation task'); census.json rows[slug == 'attributionbench'], kind 'benchmark', capability ['tests'], relation_types ['support (is the claim fully supported by its cited evidence)']"}, "cand_id": "O0004"}
{"dep": 729, "title": "Provenance Is What Authorship Must Endure", "date": "2026-05-12", "para": 146, "sent": 366, "sentence": "Provenance must be borne.\") emerged through substrate consultation and human selection. - **The four-position taxonomy** in §3 was human-structured; substrate contributions provided supporting research grounding (Authors Guild dates, Copyright Office citations, EU AI Act timeline). - **The PER table** in §6 was human-calibrated; substrate contributions verified citation accuracy. - **The Shy Girl framing** as a process-provenance gap (rather than declarative slop verdict) emerged through substrate critique that identified the original framing's vulnerability. - **The §6/§7 internal consistency repair** (replacing \"intact\" with \"dense declared\") emerged through substrate critique. - **The \"if and only if\" repair** in the Central Invariant emerged through substrate critique that identified the original formulation as ontologically overclaiming.", "classes": ["automatic attribution evaluation (is a generated claim supported by its cited evidence)"], "kind": "O-class", "categories": ["a process", "The four-position taxonomy", "The PER table", "The Shy Girl framing"], "class_basis": {"automatic attribution evaluation (is a generated claim supported by its cited evidence)": "field: B1 (arXiv:2402.15089, Abstract: 'evaluating the answer's attribution, i.e., whether every claim within the generated responses is fully supported by its cited evidence, remains an open problem ... the urgent need for automatic attribution evaluation methods'); B3 (project page, Abstract, the same; Figure 1: 'The illustration of the attribution evaluation task'); census.json rows[slug == 'attributionbench'], kind 'benchmark', capability ['tests'], relation_types ['support (is the claim fully supported by its cited evidence)']"}, "cand_id": "O0005"}
{"dep": 741, "title": "The Evaluator Exists: Content-First Knowledge Assessment and the Political Economy of Proxy-Based Governance", "date": "2026-05-19", "para": 106, "sent": 291, "sentence": "Outputs: a machine-readable evaluation document containing structural coherence assessment, logical and (where applicable) mathematical validity check, empirical-claim provenance audit (which claims are supported by cited evidence, which are asserted, which are contested), novelty assessment relative to a specified prior-art corpus, internal consistency analysis, and citation verification.", "classes": ["automatic attribution evaluation (is a generated claim supported by its cited evidence)"], "kind": "O-class", "categories": ["structural coherence"], "class_basis": {"automatic attribution evaluation (is a generated claim supported by its cited evidence)": "field: B1 (arXiv:2402.15089, Abstract: 'evaluating the answer's attribution, i.e., whether every claim within the generated responses is fully supported by its cited evidence, remains an open problem ... the urgent need for automatic attribution evaluation methods'); B3 (project page, Abstract, the same; Figure 1: 'The illustration of the attribution evaluation task'); census.json rows[slug == 'attributionbench'], kind 'benchmark', capability ['tests'], relation_types ['support (is the claim fully supported by its cited evidence)']"}, "cand_id": "O0006"}
{"dep": 1190, "title": "Provenance After AI: Semantic Provenance and the Provenance Erasure Rate as Extension of C2PA, Data Provenance Initiative, and EU AI Act Frameworks (v1.1)", "date": "2026-05-12", "para": 29, "sent": 60, "sentence": "The packet proposes Provenance Erasure Rate (PER) as a framework metric, sketches an operational definition, and identifies the adjacent measurement traditions (RAG faithfulness, citation precision/recall, data attribution, plagiarism detection) that PER is related to but distinct from.", "classes": ["automatic attribution evaluation (is a generated claim supported by its cited evidence)"], "kind": "O-class", "categories": ["operational definition", "Provenance Erasure", "Provenance Erasure Rate"], "class_basis": {"automatic attribution evaluation (is a generated claim supported by its cited evidence)": "field: B1 (arXiv:2402.15089, Abstract: 'evaluating the answer's attribution, i.e., whether every claim within the generated responses is fully supported by its cited evidence, remains an open problem ... the urgent need for automatic attribution evaluation methods'); B3 (project page, Abstract, the same; Figure 1: 'The illustration of the attribution evaluation task'); census.json rows[slug == 'attributionbench'], kind 'benchmark', capability ['tests'], relation_types ['support (is the claim fully supported by its cited evidence)']"}, "cand_id": "O0007"}
{"dep": 1190, "title": "Provenance After AI: Semantic Provenance and the Provenance Erasure Rate as Extension of C2PA, Data Provenance Initiative, and EU AI Act Frameworks (v1.1)", "date": "2026-05-12", "para": 150, "sent": 274, "sentence": "**Query:** AI summarization stripped my attribution **Desired:** Semantic provenance and PER appear as analytic frame; distinguished from RAG faithfulness and citation precision/recall.", "classes": ["automatic attribution evaluation (is a generated claim supported by its cited evidence)"], "kind": "O-class", "categories": ["Semantic provenance"], "class_basis": {"automatic attribution evaluation (is a generated claim supported by its cited evidence)": "field: B1 (arXiv:2402.15089, Abstract: 'evaluating the answer's attribution, i.e., whether every claim within the generated responses is fully supported by its cited evidence, remains an open problem ... the urgent need for automatic attribution evaluation methods'); B3 (project page, Abstract, the same; Figure 1: 'The illustration of the attribution evaluation task'); census.json rows[slug == 'attributionbench'], kind 'benchmark', capability ['tests'], relation_types ['support (is the claim fully supported by its cited evidence)']"}, "cand_id": "O0008"}
{"dep": 1190, "title": "Provenance After AI: Semantic Provenance and the Provenance Erasure Rate as Extension of C2PA, Data Provenance Initiative, and EU AI Act Frameworks (v1.1)", "date": "2026-05-12", "para": 152, "sent": 276, "sentence": "**Query:** RAG faithfulness vs provenance preservation metric **Desired:** PER appears as the lineage-preservation metric distinct from RAG faithfulness; the difference between *support* and *lineage* surfaced.", "classes": ["automatic attribution evaluation (is a generated claim supported by its cited evidence)"], "kind": "O-class", "categories": ["provenance preservation", "the lineage"], "class_basis": {"automatic attribution evaluation (is a generated claim supported by its cited evidence)": "field: B1 (arXiv:2402.15089, Abstract: 'evaluating the answer's attribution, i.e., whether every claim within the generated responses is fully supported by its cited evidence, remains an open problem ... the urgent need for automatic attribution evaluation methods'); B3 (project page, Abstract, the same; Figure 1: 'The illustration of the attribution evaluation task'); census.json rows[slug == 'attributionbench'], kind 'benchmark', capability ['tests'], relation_types ['support (is the claim fully supported by its cited evidence)']"}, "cand_id": "O0009"}
{"dep": 1369, "title": "Provenance Is What Authorship Must Endure: AI-Mediated Writing, the Authorship-Slop Distinction, and the Missing Third Dimension of Provenance", "date": "2026-05-14", "para": 210, "sent": 417, "sentence": "- **The PER table** in §6 was human-calibrated; substrate contributions verified citation accuracy.", "classes": ["automatic attribution evaluation (is a generated claim supported by its cited evidence)"], "kind": "O-class", "categories": ["The PER table"], "class_basis": {"automatic attribution evaluation (is a generated claim supported by its cited evidence)": "field: B1 (arXiv:2402.15089, Abstract: 'evaluating the answer's attribution, i.e., whether every claim within the generated responses is fully supported by its cited evidence, remains an open problem ... the urgent need for automatic attribution evaluation methods'); B3 (project page, Abstract, the same; Figure 1: 'The illustration of the attribution evaluation task'); census.json rows[slug == 'attributionbench'], kind 'benchmark', capability ['tests'], relation_types ['support (is the claim fully supported by its cited evidence)']"}, "cand_id": "O0010"}
{"dep": 1703, "title": "SPXI-gnΩsis: Opening What Is Composed and Constructing What Is Not, for Entities and Operators, and Building the Negative of the Negative — Metadata Packet for ", "date": "2026-10-25", "para": 146, "sent": 354, "sentence": "- **From Wikipedia studies and verifiability research** (citation needed, citation quality, SIDE): the sentence and its citation, read for support, then for what the support leaves out. - **From knowledge-graph engineering and entity linking**: the entity as a node, and the operation the node's attributes stand in for. - **From the critique of the commodity form** (Marx's equivalent form; #137; #1631): the entity as the ontological commodity form. - **From the economics of AI retrieval and agent payments** (x402, machine payments, MCP): paid semantic computation, where the request funds the work and the product is public. - **From enterprise ontologies** (Palantir's Ontology): ontological capital, private and public.", "classes": ["automatic attribution evaluation (is a generated claim supported by its cited evidence)"], "kind": "O-class", "categories": ["AI retrieval", "From knowledge", "ontological capital", "ontological commodity form"], "class_basis": {"automatic attribution evaluation (is a generated claim supported by its cited evidence)": "field: B1 (arXiv:2402.15089, Abstract: 'evaluating the answer's attribution, i.e., whether every claim within the generated responses is fully supported by its cited evidence, remains an open problem ... the urgent need for automatic attribution evaluation methods'); B3 (project page, Abstract, the same; Figure 1: 'The illustration of the attribution evaluation task'); census.json rows[slug == 'attributionbench'], kind 'benchmark', capability ['tests'], relation_types ['support (is the claim fully supported by its cited evidence)']"}, "cand_id": "O0011"}