{"query":"claims","field":"","total":3,"results":[{"name":"joel-chan-context24-contextualizing-scientific-claims-sdp-2024","title":"Context24: Contextualizing Scientific Figures and Tables Shared Task Dataset","native":false,"license":"CC-BY-4.0","keywords":["scholarly document processing","scientific claim verification","evidence retrieval","figure extraction","table extraction","natural language processing","contextualization","scientific text mining","nougat ocr"],"networks":["iosp-laptops"],"crate_cid":"bafybeihizzlrrwg7bljv6rhdmabio4vwqfgqoi4wzkagba45kdirrlctmq","indexed_at":"2026-10-12T10:50:33+00:00","source_url":"https://huggingface.co/datasets/joelchan/contextualizing-scientific-claims","total_size":292976671,"description":"This dataset provides training, development, and test data for the 2024 Workshop on Scholarly Document Processing (SDP 2024) Shared Task on Contextualizing Scientific Figures and Tables (Context24). The task focuses on connecting scientific claims drawn from working lab notebooks and research synthesis discussions to supporting evidence within scholarly publications across domains including cell biology, computer science, human-computer interaction, and the social sciences (from sub-collections akamatsulab, BIOL403, dg-social-media-polarization, and megacoglab).\n\nThe dataset covers two tasks:\n- Task 1 (Evidence Identification): Linking natural-language claims to relevant figures and tables within a cited paper that present the supporting findings.\n- Task 2 (Grounding Context Identification): Identifying textual snippets in the paper providing key methodological details grounding the claim.\n\nIncluded files encompass JSON datasets for claim annotations across training, development, and evaluation splits (task1-train-dev.json, task1-train-dev-2024-04-25-update.json, task1-test.json, task2-train-dev.json, task2-test.json), parsed full text of reference publications (full_texts.json, full_texts-2024-04-25-update.json, full_texts-test.json), segmented figure and table images, extracted caption texts parsed via Nougat OCR, and evaluation scripts.","total_files":7252,"research_field":"Computer and information sciences","spatial_coverage":"","temporal_coverage":""},{"name":"elife-pathways-mira-graphs","title":"eLife Pathways MIRA Knowledge Graphs and Claim Trees","native":false,"license":"CC-BY-4.0","keywords":["json-ld","knowledge graph","mira","discourse graphs","claims","evidence","ontology","semantic web"],"networks":["iosp-laptops"],"crate_cid":"bafybeibwyij4lrmfav3nkx5osau5gcbs4pqqrpnp3332d6maikwftztmhi","indexed_at":"2026-10-12T10:02:40+00:00","source_url":"Copied from our underlay repository, https://www.underlay.org/elifepathways/mira","total_size":92220,"description":"This dataset consists of a JSON-LD document containing knowledge graph structures and schema definitions for scientific claims and evidence, originated from an eLife pathways claim trees pipeline. The graph defines schema nodes, subclasses, and accounts aligned with vocabularies including MIRA, Discourse Graphs (dgb), Haak Claim Schema, OWL, RDF, Dublin Core Terms (dct), and PROV.","total_files":1,"research_field":"Computer and information sciences","spatial_coverage":"","temporal_coverage":"2026-03-30/2026-09-10"},{"name":"akamatsu-lab-open-discourse-graph-mira-2026-10-12","title":"Akamatsu lab open discourse graph (akamatsulab-open)","native":true,"license":"CC-BY-4.0","keywords":["discourse graph","MIRA","JSON-LD","evidence","endocytosis","actin","cryo-ET","open science","research attribution"],"networks":["iosp-laptops"],"crate_cid":"QmbLmBsQvX7nutpGxnzXcPdfB3mYBeZnLHvHe1CoNzGppj","indexed_at":"2026-10-12T12:32:36+00:00","source_url":"Handed off to me by the author","total_size":136666517,"description":"The publicly shareable slice of a working lab discourse graph (Akamatsu lab, University of Washington Biology: membrane biophysics, endocytosis, cryo-ET), serialized against the MIRA schema (mira.science): evidence from lab-authored and CC BY papers, the claims they support, and results flagged as published, each credited to the lab member who wrote it, with the source paper's DOI and licence. One JSON-LD file per record (nodes/, sources/), the figure snippets they observe (images/), the whole graph in one file, an index of record CIDs, and the schema snapshot needed to read it offline.","total_files":551,"research_field":"","spatial_coverage":"","temporal_coverage":""}]}