{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:OU5Q6JBCNNHHG6LMFKX3LAG45J","short_pith_number":"pith:OU5Q6JBC","schema_version":"1.0","canonical_sha256":"753b0f24226b4e73796c2aafb580dcea5ec446ce6a16d81fdd4bc30892f79c64","source":{"kind":"arxiv","id":"2605.13034","version":1},"attestation_state":"computed","paper":{"title":"ViDR: Grounding Multimodal Deep Research Reports in Source Visual Evidence","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Treating source figures as verifiable evidence objects improves the quality and verifiability of multimodal deep research reports.","cross_cats":["cs.IR"],"primary_cat":"cs.CV","authors_text":"Baoqin Sun, Haiyang Shen, Peilun Jia, Sixiong Xie, Xiang Jing, Yun Ma, Zhuofan Shi","submitted_at":"2026-05-13T05:39:38Z","abstract_excerpt":"Recent deep research systems have improved the ability of large language models to produce long, grounded reports through iterative retrieval and reasoning. However, most text-centered systems rely mainly on textual evidence, while multimodal systems often retrieve images only weakly or generate charts themselves, leaving source figures underused as evidence. We present ViDR, a multimodal deep research framework that grounds long-form reports in source figures. ViDR treats source figures as retrievable, interpretable, routable, and verifiable evidence objects, while still generating analytical"},"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":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2605.13034","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T05:39:38Z","cross_cats_sorted":["cs.IR"],"title_canon_sha256":"9c917cbe3ab626a3c6082f975b679434e6e71edd7eca2c6dd2e85f1627553f49","abstract_canon_sha256":"5c2192600db12b1070f8f5e9612cf07396e4e555763c9704013b3dd30493def5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T03:08:59.668781Z","signature_b64":"3v8Qxvet45yKLZl5HYss+ksQ0IFHls+DJiNw3JUpjV2+Gp25L1fxDAPCjcZsNeYHr+sWQxE0JVYGSSXa/UE6Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"753b0f24226b4e73796c2aafb580dcea5ec446ce6a16d81fdd4bc30892f79c64","last_reissued_at":"2026-05-18T03:08:59.668216Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T03:08:59.668216Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ViDR: Grounding Multimodal Deep Research Reports in Source Visual Evidence","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Treating source figures as verifiable evidence objects improves the quality and verifiability of multimodal deep research reports.","cross_cats":["cs.IR"],"primary_cat":"cs.CV","authors_text":"Baoqin Sun, Haiyang Shen, Peilun Jia, Sixiong Xie, Xiang Jing, Yun Ma, Zhuofan Shi","submitted_at":"2026-05-13T05:39:38Z","abstract_excerpt":"Recent deep research systems have improved the ability of large language models to produce long, grounded reports through iterative retrieval and reasoning. However, most text-centered systems rely mainly on textual evidence, while multimodal systems often retrieve images only weakly or generate charts themselves, leaving source figures underused as evidence. We present ViDR, a multimodal deep research framework that grounds long-form reports in source figures. ViDR treats source figures as retrievable, interpretable, routable, and verifiable evidence objects, while still generating analytical"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experiments show that ViDR improves overall report quality, source-figure integration, and verifiability over strong commercial and open-source baselines.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That context-aware filtering, outline-aware reranking, and VLM-based visual analysis can reliably turn noisy web images into accurate, non-hallucinated evidence atoms without introducing new errors that affect report claims.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"ViDR treats source figures as retrievable and verifiable evidence objects in multimodal deep research reports and introduces MMR Bench+ to measure improvements in visual integration and verifiability.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Treating source figures as verifiable evidence objects improves the quality and verifiability of multimodal deep research reports.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"37c97e4ec82ec3d1c004641d8a493fa9dd5388ae63af924e73133a00b96039b6"},"source":{"id":"2605.13034","kind":"arxiv","version":1},"verdict":{"id":"7b067368-07aa-4317-8a46-d0b3f01a293c","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:15:11.787518Z","strongest_claim":"Experiments show that ViDR improves overall report quality, source-figure integration, and verifiability over strong commercial and open-source baselines.","one_line_summary":"ViDR treats source figures as retrievable and verifiable evidence objects in multimodal deep research reports and introduces MMR Bench+ to measure improvements in visual integration and verifiability.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That context-aware filtering, outline-aware reranking, and VLM-based visual analysis can reliably turn noisy web images into accurate, non-hallucinated evidence atoms without introducing new errors that affect report claims.","pith_extraction_headline":"Treating source figures as verifiable evidence objects improves the quality and verifiability of multimodal deep research reports."},"references":{"count":36,"sample":[{"doi":"","year":2024,"title":"Try deep research and our new experimental model in gemini, your ai assistant","work_id":"942e9dd3-f690-4939-a3d1-2163373696f8","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Paladugu, Pranav Setlur, Jiahe Jin, James P","work_id":"2f6b8cd2-51ef-4670-8361-26e737e919cf","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"DeepResearch Bench: A Comprehensive Benchmark for Deep Research Agents","work_id":"449edaa7-8f9d-4170-952d-4ab0d740572c","ref_index":3,"cited_arxiv_id":"2506.11763","is_internal_anchor":true},{"doi":"","year":2025,"title":"gpt-researcher","work_id":"537bce4c-bf4b-4dbb-a350-0a4566ccb2c4","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"WebWatcher: Breaking New Frontier of Vision-Language Deep Research Agent","work_id":"b734fd16-d7cf-4129-ac4d-129f31964890","ref_index":5,"cited_arxiv_id":"2508.05748","is_internal_anchor":true}],"resolved_work":36,"snapshot_sha256":"7b180e11774dfcdd6a3c082a5079c29725bc30831ab5e214611568ee440162ed","internal_anchors":7},"formal_canon":{"evidence_count":2,"snapshot_sha256":"4031afe3658214c0f334248249b98857643e87468042d0595f794506fadbb6cf"},"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":"2605.13034","created_at":"2026-05-18T03:08:59.668309+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.13034v1","created_at":"2026-05-18T03:08:59.668309+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13034","created_at":"2026-05-18T03:08:59.668309+00:00"},{"alias_kind":"pith_short_12","alias_value":"OU5Q6JBCNNHH","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"OU5Q6JBCNNHHG6LM","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"OU5Q6JBC","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OU5Q6JBCNNHHG6LMFKX3LAG45J","json":"https://pith.science/pith/OU5Q6JBCNNHHG6LMFKX3LAG45J.json","graph_json":"https://pith.science/api/pith-number/OU5Q6JBCNNHHG6LMFKX3LAG45J/graph.json","events_json":"https://pith.science/api/pith-number/OU5Q6JBCNNHHG6LMFKX3LAG45J/events.json","paper":"https://pith.science/paper/OU5Q6JBC"},"agent_actions":{"view_html":"https://pith.science/pith/OU5Q6JBCNNHHG6LMFKX3LAG45J","download_json":"https://pith.science/pith/OU5Q6JBCNNHHG6LMFKX3LAG45J.json","view_paper":"https://pith.science/paper/OU5Q6JBC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.13034&json=true","fetch_graph":"https://pith.science/api/pith-number/OU5Q6JBCNNHHG6LMFKX3LAG45J/graph.json","fetch_events":"https://pith.science/api/pith-number/OU5Q6JBCNNHHG6LMFKX3LAG45J/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OU5Q6JBCNNHHG6LMFKX3LAG45J/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OU5Q6JBCNNHHG6LMFKX3LAG45J/action/storage_attestation","attest_author":"https://pith.science/pith/OU5Q6JBCNNHHG6LMFKX3LAG45J/action/author_attestation","sign_citation":"https://pith.science/pith/OU5Q6JBCNNHHG6LMFKX3LAG45J/action/citation_signature","submit_replication":"https://pith.science/pith/OU5Q6JBCNNHHG6LMFKX3LAG45J/action/replication_record"}},"created_at":"2026-05-18T03:08:59.668309+00:00","updated_at":"2026-05-18T03:08:59.668309+00:00"}