{
 "name": "Provenance meta-aggregator census",
 "compiled": "2026-10-29",
 "schema_version": "1.0",
 "schema_note": "MMRS census fields extended for provenance: lineage, capability (expresses / records / tests), provenance_object, relation_types, provenance_carrier, transformations_tested, retention_results, authority_dependencies, version_and_time. Every quote is verbatim from a saved source (sources/manifest.json). Untested capability reads 'not tested'.",
 "organizing_question": "What provenance does this system carry, through which operation, on whose assertion, with what evidence of survival?",
 "rows": [
  {
   "slug": "iptc-photo-metadata-tests",
   "name": "IPTC photo metadata tests",
   "aka": [
    "IPTC Social Media Sites Photo Metadata Test Results",
    "IPTC Photo Metadata Research"
   ],
   "url": "https://iptc.org/standards/photo-metadata/research/",
   "operator": "IPTC (International Press Telecommunications Council), Photo Metadata Working Group",
   "lineage": "media",
   "kind": "test series",
   "capability": [
    "expresses",
    "tests"
   ],
   "provenance_object": "the photograph as a file: embedded Exif and IPTC metadata (IIM and XMP)",
   "relation_types": [
    "creation (creator)",
    "credit",
    "rights (copyright notice)",
    "description (caption)"
   ],
   "provenance_carrier": "embedded metadata inside the image file",
   "unit": "a site or system, tested by upload and by saving or downloading the uploaded image",
   "transformations_tested": [
    "upload to a sharing site or system",
    "display to a viewer",
    "Save As from the browser",
    "download through the system"
   ],
   "retention_results": "2019: of eleven image-sharing sites, Pexels, pixabay and Unsplash stripped metadata from saved and downloaded files; Flickr replaced the embedded creator with the account owner; among file and message systems Twitter stripped it on Save As, Facebook kept a few rights fields in IIM and Exif",
   "cadence": "periodic: 2013, 2016, Spring 2019 (the 2019 page); a software test published August 2014",
   "count": {
    "value": "15 social media sites tested in 2016 and 2013 (research page); about a dozen checked regularly (2019 page)",
    "as_of": "2019"
   },
   "lifecycle": {
    "state": "historical",
    "detail": "dated results; the 2019 table is not a 2026 platform measurement. Succeeded in method by Metawatch (2026)"
   },
   "version_and_time": {
    "results": "Spring 2019",
    "earlier": "2013, 2016",
    "software_study": "August 2014",
    "read": "2026-10-29"
   },
   "negative_tags": [
    "NOT a current platform audit: each table carries its test year",
    "NOT a test of C2PA credentials"
   ],
   "evidence": {
    "self_description": {
     "src": "iptc-photo-metadata-tests--social-2019",
     "basis": "operator",
     "quote": "IPTC promotes the proper use of IPTC Photo Metadata and claims “embedded metadata should not be removed” from images shared on a photo platform or a system for sharing files or messages with images."
    },
    "cadence": {
     "src": "iptc-photo-metadata-tests--social-2019",
     "basis": "operator",
     "quote": "IPTC checks about a dozen sites regularly: in 2013, in 2016 and now again in Spring 2019"
    },
    "count": {
     "src": "iptc-photo-metadata-tests--research",
     "basis": "operator",
     "quote": "The IPTC Photo Metadata Working Group tested 15 social media sites in 2016 and 2013"
    },
    "transformations_tested": {
     "src": "iptc-photo-metadata-tests--social-2019",
     "basis": "operator",
     "quote": "Check if the image can be saved to a local device by a Save As function of the web browser."
    },
    "provenance_carrier": {
     "src": "iptc-photo-metadata-tests--social-2019",
     "basis": "operator",
     "quote": "are the embedded Exif and IPTC metadata (in the formats IIM and XMP) of the uploaded image still embedded?"
    },
    "relation_types": {
     "src": "iptc-photo-metadata-tests--social-2019",
     "basis": "operator",
     "quote": "Caption/description, Creator, Copyright Notice, Credit Line (= the 4Cs )"
    },
    "retention_results": {
     "src": "iptc-photo-metadata-tests--social-2019",
     "basis": "operator",
     "quote": "Embedded Creator replaced by account owner."
    },
    "retention_results_2": {
     "src": "iptc-photo-metadata-tests--social-2019",
     "basis": "operator",
     "quote": "No embedded metadata displayed, metadata is stripped off in SavedAs or downloaded files."
    },
    "finding": {
     "src": "iptc-photo-metadata-tests--research",
     "basis": "operator",
     "quote": "The results show that it is not impossible to preserve metadata in a workflow."
    }
   },
   "relations": [
    {
     "rel": "has_part",
     "target": "iptc-metawatch"
    },
    {
     "rel": "neighbour",
     "target": "c2pa-conformance"
    }
   ],
   "unchecked": [
    "the 2016 and 2013 result tables",
    "the August 2014 software study",
    "the interoperability test procedures in detail"
   ],
   "sources": {
    "iptc-photo-metadata-tests--social-2019": {
     "url": "https://iptc.org/standards/photo-metadata/social-media-sites-photo-metadata-test-results-2019/",
     "read": "2026-10-29",
     "sha256": "1b4794817413523cb403e2da7fe33943e97b40a7d42752bcc1f071c11ba14f66",
     "http": "200"
    },
    "iptc-photo-metadata-tests--research": {
     "url": "https://iptc.org/standards/photo-metadata/research/",
     "read": "2026-10-29",
     "sha256": "4d8b44f1ba520320e92060e092a8931e4a40742c313427791bc7373b463616db",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  },
  {
   "slug": "iptc-metawatch",
   "name": "IPTC Metawatch",
   "aka": [
    "Metawatch"
   ],
   "url": "https://metawatch.iptc.org/",
   "operator": "IPTC",
   "lineage": "media",
   "kind": "monitor",
   "capability": [
    "tests"
   ],
   "provenance_object": "the lead photograph of a news article, as served to the reader",
   "relation_types": [
    "creation (creator)",
    "credit",
    "rights (copyright)",
    "description (caption)",
    "data-mining preference (PLUS Data Mining)",
    "C2PA Content Credentials"
   ],
   "provenance_carrier": "embedded metadata (Exif, IPTC IIM and XMP) and C2PA manifests",
   "unit": "a news publisher, sampled each month through up to 20 recent articles",
   "transformations_tested": [
    "publication through a news site's pipeline (CMS, CDN, image optimisation)"
   ],
   "retention_results": "October 2026: 7.2% of editorial images carried a credit or copyright notice (IMATAG's 2018 test: 8%); 321 of 425 crawlable publishers kept none of the scored IPTC fields; 27 of 6,580 images carried a C2PA manifest",
   "cadence": "monthly, on the first of each month",
   "count": {
    "value": "508 news publishers in 122 countries",
    "as_of": "2026-10"
   },
   "data_access": "every run published as open data",
   "license": "CC BY 4.0 (each run's data)",
   "lifecycle": {
    "state": "active",
    "detail": "announced 5 October 2026"
   },
   "founded": "announced 5 October 2026",
   "version_and_time": {
    "announced": "2026-10-05",
    "run_reported": "October 2026",
    "read": "2026-10-29"
   },
   "negative_tags": [
    "NOT the 2019 IPTC social-media test: it measures news publishers monthly",
    "NOT a C2PA verifier"
   ],
   "evidence": {
    "self_description": {
     "src": "iptc-photo-metadata-tests--metawatch",
     "basis": "operator",
     "quote": "a free, open monitor that checks every month whether more than 500 news publishers in 122 countries keep the credits, captions and rights information embedded in their photographs"
    },
    "count": {
     "src": "iptc-photo-metadata-tests--metawatch",
     "basis": "operator",
     "quote": "On the first of each month our well-behaved crawler visits 508 news publishers in 122 countries, takes up to 20 recent articles from each, and reads the lead photograph of every article."
    },
    "license": {
     "src": "iptc-photo-metadata-tests--metawatch",
     "basis": "operator",
     "quote": "every run is published as open data under CC BY 4.0"
    },
    "retention_results": {
     "src": "iptc-photo-metadata-tests--metawatch",
     "basis": "operator",
     "quote": "In 2018 IMATAG found that 8% of editorial images carried a credit or copyright notice. Applying the same test to our October run gives 7.2%."
    },
    "retention_results_2": {
     "src": "iptc-photo-metadata-tests--metawatch",
     "basis": "operator",
     "quote": "In the October run, 321 of the 425 publishers we could crawl, three out of four kept none of the IPTC fields we score on any image we sampled."
    },
    "transformations_tested": {
     "src": "iptc-photo-metadata-tests--metawatch",
     "basis": "operator",
     "quote": "Stripping appears to happen in the publishing pipeline, which means that publishers can fix it."
    },
    "c2pa": {
     "src": "iptc-photo-metadata-tests--metawatch",
     "basis": "operator",
     "quote": "Leaving out IPTC’s own site, 27 of the 6,580 images we analysed carried a C2PA manifest"
    },
    "lineage_note": {
     "src": "iptc-photo-metadata-tests--metawatch",
     "basis": "operator",
     "quote": "Metawatch builds on IMATAG’s 2018 State of Image Metadata study and IPTC’s own prototype crawler from 2019"
    },
    "access_limit": {
     "src": "iptc-photo-metadata-tests--metawatch",
     "basis": "operator",
     "quote": "AP and USA Today began blocking automated access this month. It is a reminder that bot-blocking aimed at AI also shuts out research."
    }
   },
   "relations": [
    {
     "rel": "part_of",
     "target": "iptc-photo-metadata-tests"
    },
    {
     "rel": "neighbour",
     "target": "c2pa-conformance"
    }
   ],
   "unchecked": [
    "the Metawatch site itself and its published run data",
    "IMATAG's 2018 study"
   ],
   "sources": {
    "iptc-photo-metadata-tests--metawatch": {
     "url": "https://iptc.org/news/announcing-iptc-metawatch/",
     "read": "2026-10-29",
     "sha256": "0a7fea35f38f83bbc3a4f2807ea580ed054e661b774a5055703b693632f5da3d",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  },
  {
   "slug": "nanopublications",
   "name": "Nanopublications",
   "aka": [
    "nanopub",
    "Nanopub Registry",
    "Nanopub Query",
    "Nanodash"
   ],
   "url": "https://nanopub.net/",
   "operator": "the nanopublication community (nanopub.net)",
   "lineage": "scholarly record",
   "kind": "publication model and service ecosystem",
   "capability": [
    "expresses",
    "records"
   ],
   "provenance_object": "an assertion: a small knowledge-graph snippet published on its own",
   "relation_types": [
    "assertion provenance (how the assertion came to be)",
    "publication info (who created the nanopublication, when, licence)"
   ],
   "provenance_carrier": "RDF graphs: assertion, provenance and publication-info parts of one nanopublication",
   "unit": "a nanopublication",
   "transformations_tested": [],
   "retention_results": "not tested: the model carries provenance; it does not measure its survival",
   "data_access": "decentralized publishing and SPARQL query services",
   "lifecycle": {
    "state": "active",
    "detail": "second-generation Registry and Query services; the legacy nanopub-server keeps old unsigned nanopublications"
   },
   "version_and_time": {
    "architecture_page_updated": "2026-10-28",
    "read": "2026-10-29"
   },
   "negative_tags": [
    "NOT a test of whether provenance survives reuse",
    "publication and signing do not settle an assertion's truth"
   ],
   "evidence": {
    "self_description": {
     "src": "nanopublications--home",
     "basis": "operator",
     "quote": "A nanopublication is a small knowledge graph snippet with metadata that is treated as an independent (scientific) publication."
    },
    "relation_types": {
     "src": "nanopublications--home",
     "basis": "operator",
     "quote": "Provenance : This part describes how the assertion above came to be."
    },
    "relation_types_2": {
     "src": "nanopublications--home",
     "basis": "operator",
     "quote": "Publication Info : This part contains metadata about the nanopublication as a whole, such as when and by whom it was created and the license terms for its reuse."
    },
    "provenance_carrier": {
     "src": "nanopublications--home",
     "basis": "operator",
     "quote": "Nanopublications are implemented in the language RDF"
    },
    "lifecycle": {
     "src": "nanopublications--architecture",
     "basis": "operator",
     "quote": "The legacy nanopub-server ( instance , code ) is still used to keep the old (unsigned) nanopublications, but has otherwise no longer a function in the latest generation of the ecosystem."
    },
    "data_access": {
     "src": "nanopublications--architecture",
     "basis": "operator",
     "quote": "They allow for decentrally querying the published nanopublications via a number of SPARQL endpoints"
    }
   },
   "relations": [
    {
     "rel": "neighbour",
     "target": "provstore"
    }
   ],
   "unchecked": [
    "the guidelines and the published articles",
    "record counts in the Registry",
    "the nanopub Agent Skill"
   ],
   "sources": {
    "nanopublications--home": {
     "url": "https://nanopub.net/",
     "read": "2026-10-29",
     "sha256": "b2d9af40755a66ed0efd1b37d10933140d8c0860425721d41075dbc2679cbf5d",
     "http": "200"
    },
    "nanopublications--architecture": {
     "url": "https://nanopub.net/docs/architecture/",
     "read": "2026-10-29",
     "sha256": "0e21d7ae559d83011d3897162f9518043240fd9cd5d901086d980577f54e5e3f",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  },
  {
   "slug": "c2pa-conformance",
   "name": "C2PA Conformance Program and Explorer",
   "aka": [
    "C2PA Conforming Products List",
    "C2PA Trust List",
    "Content Credentials conformance"
   ],
   "url": "https://c2pa.org/conformance/",
   "operator": "Coalition for Content Provenance and Authenticity (C2PA), Conformance Task Force",
   "lineage": "media",
   "kind": "registry",
   "capability": [
    "expresses",
    "records"
   ],
   "provenance_object": "a digital asset with its Content Credential (C2PA manifest); the registry's own unit is the product or signer",
   "relation_types": [
    "signing",
    "creation and edit history (ingredients)",
    "trust-list membership"
   ],
   "provenance_carrier": "signed C2PA manifests; the conformance lists as JSON",
   "unit": "a conforming generator or validator product, or a trust-list entry",
   "transformations_tested": [],
   "retention_results": "not tested here: conformance concerns producing and validating C2PA data correctly",
   "count": {
    "value": "286 records in the conforming-products JSON list",
    "as_of": "2026-10-29"
   },
   "data_access": "Conformance Explorer; JSON lists on GitHub",
   "lifecycle": {
    "state": "active",
    "detail": "program opened June 2025, now v0.2"
   },
   "founded": "June 2025 (the program)",
   "authority_dependencies": [
    "C2PA Steering Committee",
    "C2PA Trust List",
    "C2PA TSA Trust List",
    "the Certificate Policy"
   ],
   "version_and_time": {
    "program": "v0.2",
    "explainer_read": "2.4",
    "read": "2026-10-29"
   },
   "negative_tags": [
    "Content Credentials do not judge whether provenance data is 'true'",
    "NOT IPTC embedded metadata",
    "C2PA does not address human or organizational identity"
   ],
   "evidence": {
    "self_description": {
     "src": "c2pa-conformance--conformance",
     "basis": "operator",
     "quote": "The C2PA Conformance Program provides assurance that products adhere to the Content Credentials specification, and fulfill a set of security requirements to ensure they are producing and validating C2PA data correctly."
    },
    "lifecycle": {
     "src": "c2pa-conformance--conformance",
     "basis": "operator",
     "quote": "After opening the program in June 2025, we have upgraded our conformance program to v0.2."
    },
    "unit": {
     "src": "c2pa-conformance--conformance",
     "basis": "operator",
     "quote": "Conforming products are placed on a publicly-accessible list."
    },
    "limit": {
     "src": "c2pa-conformance--explainer-2.4",
     "basis": "operator",
     "quote": "It is important to highlight that Content Credentials do not provide value judgments about whether a given set of provenance data is 'true', but instead merely whether the provenance information is well-formed and free from tampering, valid and trusted"
    },
    "limit_2": {
     "src": "c2pa-conformance--explainer-2.4",
     "basis": "operator",
     "quote": "provenance information alone cannot tell you whether the digital content is true, accurate or factual."
    },
    "identity": {
     "src": "c2pa-conformance--explainer-2.4",
     "basis": "operator",
     "quote": "The C2PA specification does not directly address the topic of human or organizational identity"
    },
    "ingredients": {
     "src": "c2pa-conformance--explainer-2.4",
     "basis": "operator",
     "quote": "the ingredient’s provenance cannot be verified in the same way as the provenance data of the asset in which it is contained."
    },
    "count": {
     "src": "c2pa-conformance--products-json",
     "basis": "observed",
     "quote": null,
     "note": "list length counted in the saved JSON"
    }
   },
   "relations": [
    {
     "rel": "neighbour",
     "target": "iptc-metawatch",
     "note": "Metawatch found 27 C2PA manifests in 6,580 news images, October 2026"
    }
   ],
   "unchecked": [
    "the C2PA Trust List and TSA Trust List entries",
    "the latest specification version",
    "the Content Authenticity Initiative's tools"
   ],
   "sources": {
    "c2pa-conformance--conformance": {
     "url": "https://c2pa.org/conformance/",
     "read": "2026-10-29",
     "sha256": "2478cf522501d050fc2ea29118109f6931c5325d1e7745462b69869e0ba5ad54",
     "http": "200"
    },
    "c2pa-conformance--explainer-2.4": {
     "url": "https://spec.c2pa.org/specifications/specifications/2.4/explainer/Explainer.html",
     "read": "2026-10-29",
     "sha256": "e1872450e1ae7f6af771812a94d0388322c9b7064f2cdcaa12f0447043dabc7f",
     "http": "200"
    },
    "c2pa-conformance--products-json": {
     "url": "https://raw.githubusercontent.com/c2pa-org/conformance-public/main/conforming-products/conforming-products-list.json",
     "read": "2026-10-29",
     "sha256": "e2fb3928fbb2ba3d019ac02c5ccd40d5e7c974697d9308b7efd966108cd6f4a6",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  },
  {
   "slug": "data-provenance-initiative",
   "name": "Data Provenance Initiative",
   "aka": [
    "Data Provenance Explorer",
    "Data Provenance Collection",
    "Data Provenance Card"
   ],
   "url": "https://www.dataprovenance.org/",
   "operator": "the Data Provenance Initiative (volunteer effort; paper by Shayne Longpre and 16 other authors)",
   "lineage": "training data",
   "kind": "audit collection and explorer",
   "capability": [
    "records"
   ],
   "provenance_object": "a fine-tuning dataset, traced through its collection to its sources",
   "relation_types": [
    "derivation (source and lineage)",
    "creators",
    "licences",
    "subsequent use"
   ],
   "provenance_carrier": "audit metadata per dataset; a Data Provenance Card generated for a chosen subset",
   "unit": "a dataset within a collection",
   "count": {
    "value": "44 data collections, 1800+ fine-tuning text-to-text datasets (first release)",
    "as_of": "the repository README"
   },
   "transformations_tested": [
    "dataset repackaging and hosting (licence carried or lost on hosting sites)"
   ],
   "retention_results": "the founding paper reports licence omission of 70%+ and error rates of 50%+ on widely used hosting sites",
   "data_access": "repository scripts to filter and download; the Explorer (a JavaScript application, unreadable to a plain fetch)",
   "lifecycle": {
    "state": "active",
    "detail": null
   },
   "founded": "2023 (arXiv:2310.16787, v1 25 October 2023)",
   "version_and_time": {
    "paper_v1": "2023-10-25",
    "paper_v3": "2023-11-04",
    "read": "2026-10-29"
   },
   "negative_tags": [
    "NOT legal advice (its own notice)",
    "describing a collection does not establish which model used it",
    "initiative, explorer and collection are distinct records"
   ],
   "evidence": {
    "self_description": {
     "src": "data-provenance-initiative--repo-readme",
     "basis": "operator",
     "quote": "The Data Provenance Initiative is a multi-disciplinary volunteer effort to improve transparency, documentation, and responsible use of training datasets for AI."
    },
    "count": {
     "src": "data-provenance-initiative--repo-readme",
     "basis": "operator",
     "quote": "Through a large scale audit of 44 data collections, spanning 1800+ finetuning text-to-text datasets"
    },
    "provenance_carrier": {
     "src": "data-provenance-initiative--repo-readme",
     "basis": "operator",
     "quote": "Generating a subset of data from this repository will create a `Data Provenance Card`---symbolic attribution for all constituent datasets"
    },
    "relation_types": {
     "src": "data-provenance-initiative--paper",
     "basis": "operator",
     "quote": "We develop tools and standards to trace the lineage of these datasets, from their source, creators, series of license conditions, properties, and subsequent use."
    },
    "retention_results": {
     "src": "data-provenance-initiative--paper",
     "basis": "operator",
     "quote": "We also observe frequent miscategorization of licenses on widely used dataset hosting sites, with license omission of 70%+ and error rates of 50%+."
    },
    "finding": {
     "src": "data-provenance-initiative--paper",
     "basis": "operator",
     "quote": "This points to a crisis in misattribution and informed use of the most popular datasets"
    },
    "negative": {
     "src": "data-provenance-initiative--repo-readme",
     "basis": "operator",
     "quote": "**This repository does NOT constitute legal advice.**"
    }
   },
   "relations": [
    {
     "rel": "sibling",
     "url": "https://www.machinemediation.org/aggregators/data-provenance-explorer/",
     "label": "its MMRS meta-aggregator record",
     "note": "the same entity, read for reception there and for provenance carriage here"
    }
   ],
   "unchecked": [
    "the Explorer itself (JavaScript)",
    "the multimodal extension (arXiv:2412.17847)",
    "data licence of the audit"
   ],
   "sources": {
    "data-provenance-initiative--repo-readme": {
     "url": "https://raw.githubusercontent.com/Data-Provenance-Initiative/Data-Provenance-Collection/main/README.md",
     "read": "2026-10-29",
     "sha256": "db4c35a5f4f2b17a14680fc1c6854a6f36f82a930c6dfffa4f0297b5ea7dd4f1",
     "http": "200"
    },
    "data-provenance-initiative--paper": {
     "url": "https://arxiv.org/abs/2310.16787",
     "read": "2026-10-29",
     "sha256": "d01552052066b391aeed8b50adde61897cb194ea14727553cbf65971af51bb53",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  },
  {
   "slug": "attributionbench",
   "name": "AttributionBench",
   "aka": [
    "AttributionBench (OSU)"
   ],
   "url": "https://osu-nlp-group.github.io/AttributionBench/",
   "operator": "OSU NLP Group (The Ohio State University)",
   "lineage": "answer engines",
   "kind": "benchmark",
   "capability": [
    "tests"
   ],
   "provenance_object": "a generated claim and the evidence it cites",
   "relation_types": [
    "support (is the claim fully supported by its cited evidence)"
   ],
   "provenance_carrier": "citations in a generated answer",
   "unit": "a claim–evidence pair with a binary attribution label",
   "transformations_tested": [
    "generation with citations by search-augmented models"
   ],
   "retention_results": "a fine-tuned GPT-3.5 reaches about 80% macro-F1 on the binary task (project abstract)",
   "data_access": "Hugging Face dataset",
   "license": "apache-2.0 (dataset card)",
   "lifecycle": {
    "state": "published",
    "detail": "arXiv:2402.15089"
   },
   "negative_tags": [
    "NOT AttriBench: it asks whether evidence supports a claim, not who wrote a quote",
    "evidence support is not credit to the original author"
   ],
   "evidence": {
    "self_description": {
     "src": "attributionbench--project",
     "basis": "operator",
     "quote": "we present AttributionBench, a comprehensive benchmark compiled from various existing attribution datasets."
    },
    "relation_types": {
     "src": "attributionbench--project",
     "basis": "operator",
     "quote": "whether every claim within the generated responses is fully supported by its cited evidence"
    },
    "retention_results": {
     "src": "attributionbench--project",
     "basis": "operator",
     "quote": "even a fine-tuned GPT-3.5 only achieves around 80% macro-F1 under a binary classification formulation."
    },
    "license": {
     "src": "attributionbench--dataset-card",
     "basis": "operator",
     "quote": "license: apache-2.0"
    },
    "aggregation": {
     "src": "attributionbench--dataset-card",
     "basis": "operator",
     "quote": "We constructed this dataset from multiple existing data sources in a unified format"
    }
   },
   "disambiguation": [
    {
     "neighbour": "attribench",
     "context": "quote attribution, author, suppression, Berman et al., 2026"
    }
   ],
   "relations": [
    {
     "rel": "neighbour",
     "target": "attribench"
    }
   ],
   "unchecked": [
    "the constituent datasets one by one (ExpertQA, BEGIN, HAGRID, AttrScore and others named on the card)"
   ],
   "sources": {
    "attributionbench--project": {
     "url": "https://osu-nlp-group.github.io/AttributionBench/",
     "read": "2026-10-29",
     "sha256": "0bb85b9388499eb582d61161b89e6426162c02ca2e9ee73d9dc080a0f29c5d1e",
     "http": "200"
    },
    "attributionbench--dataset-card": {
     "url": "https://huggingface.co/datasets/osunlp/AttributionBench/raw/main/README.md",
     "read": "2026-10-29",
     "sha256": "b90e7e73959c0b91966cb75cb330d03f16600e51f9c5a10c525787aa4ab31c6c",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  },
  {
   "slug": "attribench",
   "name": "AttriBench",
   "aka": [
    "Attribution Bias in Large Language Models"
   ],
   "url": "https://huggingface.co/datasets/bermaneh/AttriBench",
   "operator": "Eliza Berman, Bella Chang, Daniel B. Neill, Emily Black",
   "lineage": "answer engines",
   "kind": "benchmark",
   "capability": [
    "tests"
   ],
   "provenance_object": "a quote and its author",
   "relation_types": [
    "authorship (who wrote the quote)",
    "suppression (attribution omitted though authorship information is available)"
   ],
   "provenance_carrier": "the model's answer naming, or not naming, an author",
   "unit": "a quote of 5–30 words with its canonical author",
   "count": {
    "value": "15,620 quotes from 6,292 unique authors across two splits",
    "as_of": "dataset card"
   },
   "transformations_tested": [
    "quote attribution by 11 widely used LLMs across prompt settings"
   ],
   "retention_results": "large and systematic disparities in attribution accuracy between race, gender and intersectional groups; suppression widespread and unevenly distributed",
   "license": "CC BY 4.0",
   "lifecycle": {
    "state": "published",
    "detail": "accepted at COLM 2026"
   },
   "version_and_time": {
    "v1": "2026-04-06",
    "v2": "2026-09-24",
    "read": "2026-10-29"
   },
   "negative_tags": [
    "NOT OSU's AttributionBench",
    "fame is proxied by Google Search hit counts"
   ],
   "evidence": {
    "self_description": {
     "src": "attribench--paper",
     "basis": "operator",
     "quote": "we introduce AttriBench, the first fame- and demographically-balanced quote attribution benchmark dataset."
    },
    "relation_types": {
     "src": "attribench--paper",
     "basis": "operator",
     "quote": "We further introduce and investigate suppression, a distinct failure mode in which models omit attribution entirely, even when the model has access to authorship information."
    },
    "retention_results": {
     "src": "attribench--paper",
     "basis": "operator",
     "quote": "We find that suppression is widespread and unevenly distributed across demographic groups"
    },
    "transformations_tested": {
     "src": "attribench--paper",
     "basis": "operator",
     "quote": "we evaluate 11 widely used LLMs across different prompt settings"
    },
    "count": {
     "src": "attribench--dataset-card",
     "basis": "operator",
     "quote": "**15,620 quotes** from **6,292 unique authors** across two splits"
    },
    "version_and_time": {
     "src": "attribench--paper",
     "basis": "operator",
     "quote": "[v1]\nMon, 6 Apr 2026 22:40:03 UTC (2,212 KB)"
    },
    "lifecycle": {
     "src": "attribench--paper",
     "basis": "operator",
     "quote": "Accepted at COLM 2026. 25 pages"
    },
    "fame_proxy": {
     "src": "attribench--dataset-card",
     "basis": "operator",
     "quote": "Google Search hit count, used as a fame proxy."
    }
   },
   "disambiguation": [
    {
     "neighbour": "attributionbench",
     "context": "claim support from cited evidence, OSU NLP Group, macro-F1"
    }
   ],
   "relations": [
    {
     "rel": "neighbour",
     "target": "attributionbench"
    }
   ],
   "unchecked": [
    "the paper's suppression definitions and prompts (full text)",
    "a return to the prediction-witness audit, version by version"
   ],
   "sources": {
    "attribench--paper": {
     "url": "https://arxiv.org/abs/2604.05224",
     "read": "2026-10-29",
     "sha256": "4cd45d850843388a115f9a978cfcfc0312ed6f955cc82b9df07a53ce4cf39894",
     "http": "200"
    },
    "attribench--dataset-card": {
     "url": "https://huggingface.co/datasets/bermaneh/AttriBench/raw/main/README.md",
     "read": "2026-10-29",
     "sha256": "3a243bf599199af1e8cf91b8ac8f0c26dd80993bf8d0b4018febfe836f3ab7df",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  },
  {
   "slug": "provstore",
   "name": "ProvStore",
   "aka": [
    "OpenProvenance ProvStore"
   ],
   "url": "https://openprovenance.org/",
   "operator": "OpenProvenance (the PROV tool suite; contact at kcl.ac.uk); maintainer Dong Huynh",
   "lineage": "computational lineage",
   "kind": "repository (retired)",
   "capability": [
    "records"
   ],
   "provenance_object": "a W3C PROV document",
   "relation_types": [
    "the PROV model: entities, activities, agents, derivations"
   ],
   "provenance_carrier": "PROV documents (PROV-N, PROV-JSON, PROV-XML and others)",
   "unit": "a provenance document",
   "transformations_tested": [],
   "retention_results": "not applicable",
   "lifecycle": {
    "state": "ended",
    "detail": "retired December 2024, per the maintainer's page updated 12 September 2026; the OpenProvenance landing page still says it is coming back soon"
   },
   "version_and_time": {
    "retired": "2024-12",
    "maintainer_update": "2026-09-12",
    "landing_page_read": "2026-10-29"
   },
   "negative_tags": [
    "a live landing page is not a live repository"
   ],
   "evidence": {
    "self_description": {
     "src": "provstore--openprovenance",
     "basis": "operator",
     "quote": "A provenance repository that allows storing, browsing, and managing provenance documents via a Web interface or a REST API."
    },
    "landing": {
     "src": "provstore--openprovenance",
     "basis": "operator",
     "quote": "ProvStore is coming back soon."
    },
    "lifecycle": {
     "src": "provstore--maintainer",
     "basis": "operator",
     "quote": "Updated 12-Sep-2026 : ProvStore was retired in December 2024."
    },
    "neighbours": {
     "src": "provstore--openprovenance",
     "basis": "operator",
     "quote": "A RESTful web service that validates PROV descriptions against the PROV Constraints specification."
    }
   },
   "relations": [
    {
     "rel": "neighbour",
     "target": "nanopublications"
    }
   ],
   "unchecked": [
    "whether the public ProvStore documents were archived anywhere"
   ],
   "sources": {
    "provstore--openprovenance": {
     "url": "https://openprovenance.org/",
     "read": "2026-10-29",
     "sha256": "c86df4e417dc47c87678e8bc1ea4edc598017bd6d2991ba5f050f845c411ae4f",
     "http": "200"
    },
    "provstore--maintainer": {
     "url": "https://trungdong.github.io/getting-started-with-provenance.html",
     "read": "2026-10-29",
     "sha256": "4ce13be88b0e487025200d9ce164ea6ded9b57b082c94311184457c11a6b9c9f",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  },
  {
   "slug": "wikiwho",
   "name": "WikiWho",
   "aka": [
    "WhoColor",
    "Who Wrote That?"
   ],
   "url": "https://wikiwho.wmcloud.org/",
   "operator": "first developed by Karlsruhe Institute of Technology and GESIS; since August 2021 maintained by Wikimedia Community Tech and the Wiki Education Foundation",
   "lineage": "authorship and reuse",
   "kind": "algorithm and API",
   "capability": [
    "records",
    "tests"
   ],
   "provenance_object": "a single token (word) of a Wikipedia article",
   "relation_types": [
    "origin (the revision and editor that introduced the token)",
    "deletion",
    "reintroduction"
   ],
   "provenance_carrier": "the article's full revision history, re-parsed",
   "unit": "a token in an article revision",
   "transformations_tested": [
    "every edit in an article's revision history"
   ],
   "retention_results": "per token: when it entered, who wrote it, every removal and reinsertion; method evaluated at 95% on revision-rich articles",
   "data_access": "the WikiWho and WhoColor APIs",
   "lifecycle": {
    "state": "active",
    "detail": "moved to Wikimedia Cloud Services in August 2021; the GESIS host is to give way to Wikimedia"
   },
   "version_and_time": {
    "algorithm": "WWW 2014",
    "moved": "2021-08",
    "read": "2026-10-29"
   },
   "negative_tags": [
    "the Meta research page still describes a beta for four languages: an earlier state of the project"
   ],
   "evidence": {
    "self_description": {
     "src": "wikiwho--api-readme",
     "basis": "operator",
     "quote": "The WikiWho API gives access to the provenance/changes of words (tokens) in Wikipedia articles."
    },
    "relation_types": {
     "src": "wikiwho--mediawiki",
     "basis": "operator",
     "quote": "find out who wrote and/or removed and/or reinserted which exact text at token level at what revision ."
    },
    "retention_results": {
     "src": "wikiwho--mediawiki",
     "basis": "operator",
     "quote": "The original algorithm working behind the scenes is described in a WWW 2014 paper , along with an extensive evaluation resulting in 95% accuracy on fairly revision-rich articles."
    },
    "operator": {
     "src": "wikiwho--mediawiki",
     "basis": "operator",
     "quote": "In August 2021, it was moved to Wikimedia Cloud Services infrastructure and is now maintained and under further development by Community Tech and the Wiki Education Foundation ."
    },
    "lifecycle": {
     "src": "wikiwho--api-readme",
     "basis": "operator",
     "quote": "In the future, only Wikimedia will continue its support."
    },
    "earlier_state": {
     "src": "wikiwho--meta-research",
     "basis": "operator",
     "quote": "We currently offer the service in English, German, Turkish and Basque, with more languages planned."
    },
    "precision_note": {
     "src": "wikiwho--meta-research",
     "basis": "operator",
     "quote": "the task of identifying the \"correct\" original author of a piece of text in a WP article is not trivial."
    }
   },
   "relations": [],
   "unchecked": [
    "the WWW 2014 paper itself",
    "the list of supported wikis today"
   ],
   "sources": {
    "wikiwho--api-readme": {
     "url": "https://raw.githubusercontent.com/gesiscss/wikiwho_api_public/master/README.md",
     "read": "2026-10-29",
     "sha256": "b93dc1cea1081954e853d52bf2d46a5b5c03b8c0831636b2d75bb402ac943d2b",
     "http": "200"
    },
    "wikiwho--mediawiki": {
     "url": "https://www.mediawiki.org/wiki/WikiWho",
     "read": "2026-10-29",
     "sha256": "97cd049ccef3a05b63e29d2b47b87f6cda92755f1822563f8485865e180acb69",
     "http": "200"
    },
    "wikiwho--meta-research": {
     "url": "https://meta.wikimedia.org/wiki/Research:Wikiwho_Provenance_Api",
     "read": "2026-10-29",
     "sha256": "2055e9f1886d338599645276a0f98612cc712bbdac30bc640ef9ed125a30be34",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  },
  {
   "slug": "mapping-manuscript-migrations",
   "name": "Mapping Manuscript Migrations",
   "aka": [
    "MMM",
    "MMM Semantic Portal"
   ],
   "url": "https://mappingmanuscriptmigrations.org/",
   "operator": "University of Oxford, University of Pennsylvania (Schoenberg Institute for Manuscript Studies), IRHT-CNRS, Aalto University",
   "lineage": "manuscripts and books",
   "kind": "aggregator (linked open data)",
   "capability": [
    "records"
   ],
   "provenance_object": "a medieval or Renaissance manuscript and the events of its history",
   "relation_types": [
    "ownership",
    "sale and transfer events",
    "place",
    "works contained"
   ],
   "provenance_carrier": "RDF mapped to a unified model drawing on CIDOC CRM and FRBRoo",
   "unit": "a manuscript, reconciled across three source databases",
   "count": {
    "value": "217,700 manuscripts and other entities in the portal; 262,000 source records combined",
    "as_of": "the white paper"
   },
   "data_access": "SPARQL endpoint on Linked Data Finland; a Sampo-UI semantic portal; a copy on Zenodo",
   "license": "CC BY-NC 4.0 (the MMM data)",
   "lifecycle": {
    "state": "project ended, data maintained",
    "detail": "funded 2017–2020 by the Digging into Data Challenge; the portal home is a JavaScript application, unreadable to a plain fetch"
   },
   "founded": "2017 (funding period 2017–2020)",
   "version_and_time": {
    "project": "2017–2020",
    "read": "2026-10-29"
   },
   "negative_tags": [
    "NOT the Schoenberg Database itself: it aggregates SDBM with Bibale and the Bodleian's catalogue"
   ],
   "evidence": {
    "self_description": {
     "src": "mapping-manuscript-migrations--white-paper",
     "basis": "operator",
     "quote": "Its main goal is to combine data from several disparate sources about medieval and Renaissance manuscripts, and to use the aggregated data to explore a range of research questions about their history and provenance."
    },
    "sources": {
     "src": "mapping-manuscript-migrations--white-paper",
     "basis": "operator",
     "quote": "The project took the data from three existing databases (the Schoenberg Database of Manuscripts, Medieval Manuscripts in Oxford Libraries, and Bibale) and turned them into Linked Open Data (LOD)."
    },
    "count": {
     "src": "mapping-manuscript-migrations--white-paper",
     "basis": "operator",
     "quote": "the 217,700 manuscripts and other entities can be searched, browsed, and analyzed"
    },
    "license": {
     "src": "mapping-manuscript-migrations--white-paper",
     "basis": "operator",
     "quote": "The MMM data are made available for reuse under a CC BY-NC 4.0 license"
    },
    "units_compared": {
     "src": "mapping-manuscript-migrations--white-paper",
     "basis": "operator",
     "quote": "The former takes a single manuscript as the basic unit, and attaches the evidence of its history to that record."
    },
    "lifecycle": {
     "src": "mapping-manuscript-migrations--white-paper",
     "basis": "operator",
     "quote": "is a project funded between 2017 and 2020 by the Digging into Data Challenge."
    },
    "data_access": {
     "src": "mapping-manuscript-migrations--white-paper",
     "basis": "operator",
     "quote": "A copy of the dataset was published through the Zenodo repository."
    }
   },
   "relations": [
    {
     "rel": "has_part",
     "target": "schoenberg-database"
    }
   ],
   "unchecked": [
    "the live portal (JavaScript)",
    "the Zenodo copy and its record",
    "Bibale's present state"
   ],
   "sources": {
    "mapping-manuscript-migrations--white-paper": {
     "url": "https://seco.cs.aalto.fi/publications/2020/burrows-et-al-mmm-white-paper.pdf",
     "read": "2026-10-29",
     "sha256": "1e28a0193aa6342b7e4c81cd29d0bcf86c3047ea82ff2b34b77235bb0528f38c",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  },
  {
   "slug": "schoenberg-database",
   "name": "Schoenberg Database of Manuscripts",
   "aka": [
    "SDBM"
   ],
   "url": "https://sdbm.library.upenn.edu/",
   "operator": "Schoenberg Institute for Manuscript Studies, University of Pennsylvania",
   "lineage": "manuscripts and books",
   "kind": "database",
   "capability": [
    "records"
   ],
   "provenance_object": "an observation of a pre-modern manuscript in a catalogue, inventory or other source",
   "relation_types": [
    "sale",
    "location",
    "ownership"
   ],
   "provenance_carrier": "observation records with a VIAF-based name authority",
   "unit": "an observation (an entry in a source that documents a manuscript)",
   "count": {
    "value": "over 14,000 auction and sales catalogs, inventories and other sources; 240,000 records (MMM white paper)",
    "as_of": "2026-10-29 (home page); white paper"
   },
   "intake": "continuous; registered users contribute observations and batch uploads",
   "data_access": "search, download, SPARQL endpoint",
   "license": "CC BY-SA 4.0 for downloaded search results",
   "lifecycle": {
    "state": "active",
    "detail": null
   },
   "evidence": {
    "self_description": {
     "src": "mapping-manuscript-migrations--sdbm",
     "basis": "operator",
     "quote": "The SDBM continuously aggregates and updates historic observations of pre-modern manuscripts drawn from over 14,000 auction and sales catalogs, inventories, catalogs from institutional and private collections, and other sources that document the sales and locations of these books from around the world."
    },
    "license": {
     "src": "mapping-manuscript-migrations--sdbm",
     "basis": "operator",
     "quote": "Download Search Results for your own use under a Creative Commons Attribution-ShareAlike 4.0 Unported License"
    },
    "unit": {
     "src": "mapping-manuscript-migrations--white-paper",
     "basis": "operator",
     "quote": "The latter focuses on an “observation” of a manuscript in a sales or auction catalogue"
    },
    "count": {
     "src": "mapping-manuscript-migrations--white-paper",
     "basis": "operator",
     "quote": "Schoenberg Database of Manuscripts (Schoenberg Institute for Manuscript Studies, 240,000 records)"
    }
   },
   "relations": [
    {
     "rel": "part_of",
     "target": "mapping-manuscript-migrations"
    }
   ],
   "unchecked": [
    "the current record count on the database itself"
   ],
   "sources": {
    "mapping-manuscript-migrations--sdbm": {
     "url": "https://sdbm.library.upenn.edu/",
     "read": "2026-10-29",
     "sha256": "9a0fdf8805bd970aba7371e6ca91d6bef572b5f6be01d9ae2bf7c6ab4df31bb8",
     "http": "200"
    },
    "mapping-manuscript-migrations--white-paper": {
     "url": "https://seco.cs.aalto.fi/publications/2020/burrows-et-al-mmm-white-paper.pdf",
     "read": "2026-10-29",
     "sha256": "1e28a0193aa6342b7e4c81cd29d0bcf86c3047ea82ff2b34b77235bb0528f38c",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  },
  {
   "slug": "vroniplag-wiki",
   "name": "VroniPlag Wiki",
   "aka": [
    "VroniPlag"
   ],
   "url": "https://vroniplag.fandom.com/de",
   "operator": "a volunteer wiki on Fandom (formerly Wikia); German Wikipedia names Martin Heidingsfelder as founder",
   "lineage": "authorship and reuse",
   "kind": "documentation wiki",
   "capability": [
    "records",
    "tests"
   ],
   "provenance_object": "a passage of a doctoral thesis and the source it reproduces without attribution",
   "relation_types": [
    "unattributed reuse (plagiarism fragment)",
    "source text"
   ],
   "provenance_carrier": "text-parallel documentation: page-by-page fragments, each classified and checked by a second person",
   "unit": "a thesis, documented as fragments",
   "count": {
    "value": "201 dissertations and 13 habilitation theses named (German Wikipedia, as of 3 April 2023); at least 90 doctorates revoked after a documentation (as of 30 April 2024)",
    "as_of": "2023-04-03 / 2024-04-30"
   },
   "retention_results": "the share of a thesis's pages found to contain plagiarism, per thesis",
   "lifecycle": {
    "state": "active",
    "detail": "online since 28 March 2011; the wiki returned 403 to a plain fetch on 2026-10-29, so its own figures were not read"
   },
   "founded": "28 March 2011",
   "license": "CC BY-SA 3.0 (English Wikipedia's infobox)",
   "negative_tags": [
    "NOT GuttenPlag Wiki, from which it grew",
    "publishes a case only where serious plagiarism was found, after notifying the university"
   ],
   "evidence": {
    "self_description": {
     "src": "vroniplag-wiki--wikipedia",
     "basis": "third-party",
     "quote": "is a [[wiki]] started 28 March 2011 at [[Wikia]] that examines and documents the extent of [[plagiarism]] in German doctoral [[theses]]."
    },
    "method": {
     "src": "vroniplag-wiki--wikipedia",
     "basis": "third-party",
     "quote": "Each fragment is classified as to the kind of plagiarism and must be verified by a second person before it is acceptable."
    },
    "policy": {
     "src": "vroniplag-wiki--wikipedia",
     "basis": "third-party",
     "quote": "Before publication, the office of academic integrity at the university in question is formally notified, as any revocation of title must be done by the university."
    },
    "count": {
     "src": "vroniplag-wiki--dewiki",
     "basis": "third-party",
     "quote": "erhöhte sich die Zahl der namentlich genannten untersuchten Arbeiten auf 201 Dissertationen und 13 [[Habilitationsschrift]]en"
    },
    "consequence": {
     "src": "vroniplag-wiki--dewiki",
     "basis": "third-party",
     "quote": "mindestens 90 Personen im Nachgang einer Dokumentation auf ''VroniPlag Wiki'' der Doktorgrad"
    },
    "founder": {
     "src": "vroniplag-wiki--dewiki",
     "basis": "third-party",
     "quote": "| Gründer         = [[Martin Heidingsfelder]]"
    }
   },
   "relations": [],
   "unchecked": [
    "the wiki's own case list and statistics (403 on a plain fetch)",
    "the current licence on Fandom"
   ],
   "sources": {
    "vroniplag-wiki--wikipedia": {
     "url": "https://en.wikipedia.org/w/index.php?title=VroniPlag_Wiki&action=raw",
     "read": "2026-10-29",
     "sha256": "493709c4110c66189bcd94112e18f6236bc6abc0427921e904ad48d76657528b",
     "http": "200"
    },
    "vroniplag-wiki--dewiki": {
     "url": "https://de.wikipedia.org/w/index.php?title=VroniPlag_Wiki&action=raw",
     "read": "2026-10-29",
     "sha256": "fffeef99bb5626eded74f096bc05c1ad04052232323b36a0fa1bd5d61f5a7f98",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  },
  {
   "slug": "getty-provenance-index",
   "name": "Getty Provenance Index",
   "aka": [
    "GPI",
    "Getty Provenance Index®"
   ],
   "url": "https://www.getty.edu/databases-tools-and-technologies/provenance/",
   "operator": "Getty Research Institute (J. Paul Getty Trust)",
   "lineage": "objects and art",
   "kind": "database (linked open data)",
   "capability": [
    "records"
   ],
   "provenance_object": "a work of art and the documents of its ownership and sale",
   "relation_types": [
    "ownership",
    "sale",
    "dealer stock",
    "inventory",
    "payment to artists"
   ],
   "provenance_carrier": "transcribed primary documents: archival inventories, auction and dealer sales catalogs, dealer stock books",
   "unit": "a resource drawn from a primary document",
   "count": {
    "value": "over 12 million digitized, open-access resources",
    "as_of": "2026-10-29"
   },
   "data_access": "the Remodeled Index (Arches, linked open data) and the Legacy Index during the transition",
   "lifecycle": {
    "state": "active",
    "detail": "transitioning to linked open data; archival inventories to follow"
   },
   "founded": "1974 by Burton Fredericksen (home page); the initiative page places its beginning in the early 1980s",
   "version_and_time": {
    "digital": "1984",
    "cd_rom": "1994",
    "online": "1998",
    "read": "2026-10-29"
   },
   "negative_tags": [
    "coverage varies by region, period and type of document"
   ],
   "evidence": {
    "self_description": {
     "src": "getty-provenance-index--home",
     "basis": "operator",
     "quote": "The Getty Provenance Index® (GPI) contains over 12 million digitized, open-access resources that trace the origins, movements, and lineages of art and cultural artifacts."
    },
    "founded": {
     "src": "getty-provenance-index--home",
     "basis": "operator",
     "quote": "Founded in 1974 by Burton Fredericksen"
    },
    "founded_2": {
     "src": "getty-provenance-index--initiative",
     "basis": "operator",
     "quote": "The GPI began as an initiative within the Getty Museum’s Paintings department in the early 1980s led by curator Burton B. Fredericksen."
    },
    "lifecycle": {
     "src": "getty-provenance-index--home",
     "basis": "operator",
     "quote": "Archival Inventory data will also transition to this new format soon. While the Index transitions, the original datasets will remain accessible in the Legacy Index platform."
    },
    "version_and_time": {
     "src": "getty-provenance-index--home",
     "basis": "operator",
     "quote": "First compiled digitally in 1984, published on CD-ROM in 1994, and launched online in 1998"
    },
    "coverage": {
     "src": "getty-provenance-index--whats-covered",
     "basis": "operator",
     "quote": "The quantity and scope of research material available in the Getty Provenance Index® varies by region, period, and type of document."
    },
    "units": {
     "src": "getty-provenance-index--whats-covered",
     "basis": "operator",
     "quote": "The Provenance Index encompasses over 12 million resources extracted from more than 22,000 catalogs"
    }
   },
   "relations": [],
   "unchecked": [
    "the data licence of the Remodeled Index",
    "the Public Collections and Collectors Files databases in detail"
   ],
   "sources": {
    "getty-provenance-index--home": {
     "url": "https://www.getty.edu/databases-tools-and-technologies/provenance/",
     "read": "2026-10-29",
     "sha256": "b52cee21703553b27011adff4e8e099f1053ddded486ce0a82c20285c7e21a57",
     "http": "200"
    },
    "getty-provenance-index--initiative": {
     "url": "https://www.getty.edu/projects/getty-provenance-index-initiative",
     "read": "2026-10-29",
     "sha256": "760348e6cb7a40fc1e0e9e373c989b70b1d98d216364e3e623e97e6bc16ab58e",
     "http": "200"
    },
    "getty-provenance-index--whats-covered": {
     "url": "https://www.getty.edu/databases-tools-and-technologies/provenance/whats-covered/",
     "read": "2026-10-29",
     "sha256": "e5cafa0ab34a095ddb773e05e7754bbf000d79f50b353d0ac322dcaceb3d5613",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  },
  {
   "slug": "olmotrace",
   "name": "OLMoTrace",
   "aka": [],
   "url": "https://playground.allenai.org",
   "operator": "Allen Institute for AI (Ai2), with the University of Washington, UC Berkeley and Stanford",
   "lineage": "training data",
   "kind": "tracing tool",
   "capability": [
    "records"
   ],
   "provenance_object": "a span of a language model's output and the training documents that contain it verbatim",
   "relation_types": [
    "verbatim match between output and training text"
   ],
   "provenance_carrier": "the model's full training corpus, indexed with an extended infini-gram",
   "unit": "a maximal matching span of model output",
   "count": {
    "value": "3.2 billion documents and 4.6 trillion tokens (OLMo-2-32B-Instruct's training data)",
    "as_of": "the paper, v2"
   },
   "transformations_tested": [
    "model generation, traced back to training text"
   ],
   "retention_results": "returns matching documents within a few seconds; the authors state the matches are not causal and not citations",
   "data_access": "Ai2 Playground; open source",
   "lifecycle": {
    "state": "active",
    "detail": "ACL 2025 demo track"
   },
   "version_and_time": {
    "v1": "2025-04-09",
    "v2": "2025-07-08",
    "read": "2026-10-29"
   },
   "negative_tags": [
    "the retrieved documents are not causes of the output",
    "NOT a citation system",
    "traces only models whose training data is open"
   ],
   "evidence": {
    "self_description": {
     "src": "olmotrace--abs",
     "basis": "operator",
     "quote": "We present OLMoTrace, the first system that traces the outputs of language models back to their full, multi-trillion-token training data in real time."
    },
    "relation_types": {
     "src": "olmotrace--abs",
     "basis": "operator",
     "quote": "OLMoTrace finds and shows verbatim matches between segments of language model output and documents in the training text corpora."
    },
    "limit": {
     "src": "olmotrace--paper",
     "basis": "operator",
     "quote": "The retrieved documents should not be interpreted as having a causal effect on the LM output, or as supporting evidence or citations for the LM output."
    },
    "count": {
     "src": "olmotrace--paper",
     "basis": "operator",
     "quote": "which totals 3.2 billion documents and 4.6 trillion (Llama-2) tokens"
    },
    "data_access": {
     "src": "olmotrace--abs",
     "basis": "operator",
     "quote": "OLMoTrace is publicly available and fully open-source."
    }
   },
   "relations": [
    {
     "rel": "neighbour",
     "target": "data-provenance-initiative"
    }
   ],
   "unchecked": [
    "the Playground itself",
    "infini-gram as a separate record"
   ],
   "sources": {
    "olmotrace--abs": {
     "url": "https://arxiv.org/abs/2504.07096",
     "read": "2026-10-29",
     "sha256": "cd982680a1ca4c02a28ebd0cb3c7f7c1b7ed60cde75787fd3918c12f99154c22",
     "http": "200"
    },
    "olmotrace--paper": {
     "url": "https://arxiv.org/html/2504.07096v2",
     "read": "2026-10-29",
     "sha256": "b1b16ba583eeb0dbfdb07615ebda0ce56623b190b14c63115ff04a8a374e0bdd",
     "http": "200"
    }
   },
   "observed": {
    "on": "2026-10-29",
    "fetch": "read"
   }
  }
 ]
}