{"paper":{"title":"DataClawBench: An Agent Benchmark for Exploratory Real-World Financial Data Analysis","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"DataClaw is a new benchmark that evaluates data-analysis agents on noisy real-world records by tracking their step-by-step progress through 492 annotated tasks.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Bowen Deng, BoYuan Li, Chuan Chen, Jialong Chen, Jianhao Lin, Qiaohong Zhang, Weihao Ye, Wei-Shi Zheng, Yi Luo, Zibin Zheng","submitted_at":"2026-05-04T11:57:09Z","abstract_excerpt":"Autonomous data analysis agents are increasingly expected to conduct exploratory analysis over underexplored data environments. This burden is especially salient in complex financial analytics, where relevant evidence is rarely pre-specified. However, existing benchmarks typically evaluate such agents in prior-guided settings, providing selected data sources, explicit data schemas, or cleaned data, thereby understating the exploratory burden. We introduce DataClawBench, a benchmark for exploratory real-world financial data analysis under limited prior guidance. DataClawBench contains approxima"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experiments with eight advanced LLMs show that current agents remain far from reliable in this setting, with seven models achieving below 50% overall accuracy.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the 492 cross-domain tasks and their intermediate milestone annotations faithfully represent the process and difficulty of real-world exploratory data analysis in underexplored environments.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"DataClaw supplies a process-oriented benchmark of real-world noisy data and milestone-annotated tasks that shows seven of eight tested LLMs achieve below 50% accuracy on exploratory analysis.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"DataClaw is a new benchmark that evaluates data-analysis agents on noisy real-world records by tracking their step-by-step progress through 492 annotated tasks.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"388d2dbc7f236389b11922f571f9d73666cf1f8dd807a44ef81a616431d92315"},"source":{"id":"2605.02503","kind":"arxiv","version":2},"verdict":{"id":"e48c4361-3e8b-4b1b-9bc5-74be690e0db3","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-08T18:20:13.877925Z","strongest_claim":"Experiments with eight advanced LLMs show that current agents remain far from reliable in this setting, with seven models achieving below 50% overall accuracy.","one_line_summary":"DataClaw supplies a process-oriented benchmark of real-world noisy data and milestone-annotated tasks that shows seven of eight tested LLMs achieve below 50% accuracy on exploratory analysis.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the 492 cross-domain tasks and their intermediate milestone annotations faithfully represent the process and difficulty of real-world exploratory data analysis in underexplored environments.","pith_extraction_headline":"DataClaw is a new benchmark that evaluates data-analysis agents on noisy real-world records by tracking their step-by-step progress through 492 annotated tasks."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.02503/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_compliance","ran_at":"2026-05-19T16:20:13.844192Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"5b6c16b0532279806b0f9209ade9bf009423c020ea04c8c36a91583636e839b5"},"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"}