{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2607.24707", "kind": "paper", "url": "https://arxiv.org/abs/2607.24707"}, {"id": "gh:salinaria/erunderstand", "kind": "repo", "url": "https://github.com/salinaria/ERUnderstand"}], "authors": ["Ali Ansari", "Yasmin Mohammadi", "Farnoush Nili", "Parsa Esmaeilkhani", "Longin Jan Latecki", "Eduard Dragut"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2607.24707", "source_url": "https://arxiv.org/abs/2607.24707"}, "categories": ["General AI", "Multimodal"], "description": {"en": "ERUnderstand evaluates vision-language models on structured understanding of entity-relationship diagrams. The benchmark contains 2,960 diagrams across curated educational sources, real-world schemas, and synthetic data, with standardized machine-readable JSON annotations. Scoring uses F1, BLEU, and graph edit distance to measure how accurately models recover schema elements such as entities, relationships, attributes, and extended ER constructs."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2607.24707", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2607.24707", "title": "ERUnderstand"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Temple University"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2607.24707", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2607.24707", "source_url": "https://openalex.org/W7171526760"}, "key": "claire-radar:2607.24707", "modality": null, "name": "ERUnderstand", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "unknown", "origin_record_id": "bm_erunderstand_657f59fd", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "286d26288bbe89f62135f25a4ffd63feb58f87b14bab06a205ffb6bef1923344", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2607.24707"}, "publisher": null, "released": "2026-07-27", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2607.24707", "source_url": "https://arxiv.org/abs/2607.24707"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2607-24707", "source": "claire_radar", "source_benchmark_id": "2607.24707", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Machine Generated", "applicationDomains": ["General AI"], "area": "Multimodal", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/salinaria/ERUnderstand", "githubScope": "benchmark_repo", "githubStars": 4, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 2, "hfPaperUrl": "https://huggingface.co/papers/2607.24707", "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-09-28", "state": "stale"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}}, "status": "fresh"}, "capabilities": [], "confidence": "High", "construction": "Original Synthetic", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "90f18217e551a6266cad1ded9eeda84755e1af9dbf872ef664ec3fb6683db4e5", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "ERUnderstand evaluates vision-language models on structured understanding of entity-relationship diagrams. The benchmark contains 2,960 diagrams across curated educational sources, real-world schemas, and synthetic data, with standardized machine-readable JSON annotations. Scoring uses F1, BLEU, and graph edit distance to measure how accurately models recover schema elements such as entities, relationships, attributes, and extended ER constructs.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "public artifact URL"], "snippet": "We introduce ERUnderstand, the first large-scale benchmark for structured understanding of ER diagrams, comprising 2,960 diagrams collected from curated educational sources, real-world schemas, and synthetically generated examples spanning diverse domains, notations, complexity levels, and Extended Entity-Relationship (EER) constructs."}, "familyId": "bmf_961dfc017065", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 1, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "candidates"}, "heat": null, "id": "bm_erunderstand_657f59fd", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/salinaria/ERUnderstand", "data": null, "hfPaper": "https://huggingface.co/papers/2607.24707", "paper": "https://arxiv.org/abs/2607.24707", "pdf": "https://arxiv.org/pdf/2607.24707", "project": null, "report": "https://arxiv.org/abs/2607.24707"}, "metrics": [], "motivation": "Entity-Relationship Diagrams (ERDs) are central to conceptual database design, yet they are typically available only as rendered images rather than machine-readable schemas, limiting AI-assisted database engineering.", "name": "ERUnderstand", "oneLine": "ERUnderstand evaluates vision-language models on structured understanding of entity-relationship diagrams. The benchmark contains 2,960 diagrams across curated educational sources, real-world schemas, and synthetic data, with standardized machine-readable JSON annotations. Scoring uses F1, BLEU, and graph edit distance to measure how accurately models recover schema elements such as entities, relationships, attributes, and extended ER constructs.", "paperTitle": "ERUnderstand: Evaluating Vision-Language Models on Structured ER Diagrams", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2607.24707", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {"90d": {"confidence": "Medium", "coverage": 0.7, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": 0.21483102985913266, "value": 4}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.19411886932144953, "value": 2}}, "confidence": "Medium", "coverage": 0.7, "rank": 279, "score": 30}, "method": "attention-ranking-v13", "rank": 279, "score": 30}}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-07-27", "source": {"authors": ["Ali Ansari", "Yasmin Mohammadi", "Farnoush Nili", "Parsa Esmaeilkhani", "Longin Jan Latecki", "Eduard Dragut"], "categories": ["cs.AI", "cs.CV", "cs.DB"], "id": "2607.24707", "title": "ERUnderstand: Evaluating Vision-Language Models on Structured ER Diagrams", "type": "arxiv", "url": "https://arxiv.org/abs/2607.24707"}, "sourceUpdatedAt": "2026-07-27T17:46:43Z", "topics": ["Multimodal"], "venueAttempts": [], "whyItMatters": "ER diagrams are central to database design but their image-based nature impedes automated processing. Existing VLM benchmarks do not focus on structured schema extraction from ERDs, and no public benchmark provides a standardized protocol for this task. ERUnderstand fills that gap with a reusable dataset and evaluation toolkit, enabling reproducible comparison of VLM performance on conceptual database schemas."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
