{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.29894", "kind": "paper", "url": "https://arxiv.org/abs/2606.29894"}], "authors": ["Nikolay Georgiev", "Maria Drencheva", "Kseniia Ibragimova", "Ivo Petrov", "Dimitar I. Dimitrov", "Martin Vechev"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.29894", "source_url": "https://arxiv.org/abs/2606.29894"}, "categories": ["Information retrieval", "Language & Knowledge", "Mathematics & Formal Science", "cs.IR"], "description": {"en": "SABER-Math evaluates information retrieval for mathematical queries, with about 283K problems and tasks for reranking based on fine-grained relevance. Scoring uses preference tournament ratings."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.29894", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.29894", "title": "SABER-Math"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.29894", "modality": null, "name": "SABER-Math", "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_saber-math_7bfa1c5b", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "9be896ae33208c4e475cf2f2f7bf15443f40bb144d46a8807e694cf5de408a26", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.29894"}, "publisher": null, "released": "2026-06-29", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.29894", "source_url": "https://arxiv.org/abs/2606.29894"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-29894", "source": "claire_radar", "source_benchmark_id": "2606.29894", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["Mathematics & Formal Science"], "area": "Language & Knowledge", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 0, "hfPaperUrl": "https://huggingface.co/papers/2606.29894", "observedAt": "2026-10-04T03:02:42.754940Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-10-04", "state": "stale"}}, "status": "stale"}, "capabilities": ["Information retrieval"], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-21T04:30:40.934319Z", "inputHash": "547f8647ce984c70f7184dd5f84253559a73559fe639920ee0986bfac8515a8f", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "SABER-Math evaluates information retrieval for mathematical queries, with about 283K problems and tasks for reranking based on fine-grained relevance. Scoring uses preference tournament ratings.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "We address this gap by introducing SABER-Math, the first fully automated benchmark for evaluating mathematical IR without expert annotation."}, "familyId": "bmf_7856a310f2fa", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_saber-math_7bfa1c5b", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.29894", "paper": "https://arxiv.org/abs/2606.29894", "pdf": "https://arxiv.org/pdf/2606.29894", "project": null, "report": "https://arxiv.org/abs/2606.29894"}, "metrics": [], "motivation": "As agentic AI systems tackle more complex mathematical tasks, they increasingly rely on information retrieval (IR) to search problem databases, theorem libraries, and educational resources.", "name": "SABER-Math", "oneLine": "SABER-Math evaluates information retrieval for mathematical queries, with about 283K problems and tasks for reranking based on fine-grained relevance. Scoring uses preference tournament ratings.", "paperTitle": "SABER-Math: Automated Benchmark for Information Retrieval Evaluation in Mathematics", "primaryDomain": "Mathematics & Formal Science", "publication": {"evidence": "Accepted at The 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP), Hungary 2026, 34 pages", "evidenceLevel": "author-claim", "evidenceUrl": "https://arxiv.org/abs/2606.29894", "source": "arxiv-comments", "status": "acceptance_claimed", "venue": "2026 Conference on Empirical Methods in Natural Language Processing (EMNLP), Hungary 2026, 34 pages", "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "arXiv", "organizationType": "community", "role": "benchmark-publisher", "sourceUrl": "https://arxiv.org/abs/2606.29894"}], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-29", "source": {"authors": ["Nikolay Georgiev and Maria Drencheva", "Kseniia Ibragimova", "Ivo Petrov", "Dimitar I. Dimitrov", "Martin Vechev"], "categories": ["cs.IR", "cs.AI", "cs.CL", "cs.LG"], "id": "2606.29894", "title": "SABER-Math: Automated Benchmark for Information Retrieval Evaluation in Mathematics", "type": "arxiv", "url": "https://arxiv.org/abs/2606.29894"}, "sourceUpdatedAt": "2026-06-29T07:32:50Z", "topics": ["cs.IR"], "venueAttempts": [{"decisionRaw": "Accepted at The 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP), Hungary 2026, 34 pages", "evidence": [{"level": "author-claim", "observedAt": "2026-10-14T01:46:58.058199Z", "rawValue": "Accepted at The 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP), Hungary 2026, 34 pages", "sourceType": "arxiv-comments", "sourceUrl": "https://arxiv.org/abs/2606.29894"}], "reviewStatus": "accepted", "venueName": "2026 Conference on Empirical Methods in Natural Language Processing (EMNLP), Hungary 2026, 34 pages"}], "whyItMatters": "Existing IR benchmarks fail to capture mathematical relevance, and MTEB doesn't predict math performance. SABER-Math provides a math-specific benchmark to guide retriever selection."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
