{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:4FM2KF5NQ7ELDCYWVSARSHSBZP","short_pith_number":"pith:4FM2KF5N","schema_version":"1.0","canonical_sha256":"e159a517ad87c8b18b16ac81191e41cbd49091be86ee1b25afeea554dba33e22","source":{"kind":"arxiv","id":"2507.22603","version":1},"attestation_state":"computed","paper":{"title":"BALSAM: A Platform for Benchmarking Arabic Large Language Models","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Abdelrahman El-Sheikh, Abdullah Alfaifi, Abdulmohsen Al Thubaity, Abdulrahman AlOsaimy, Afrah Altamimi, Ahmed Abdelali, Amal Almazrua, Areej Alshaqarawi, Asma Al Wazrah, Atikah Alzeghayer, Bashar Alhafni, Emad Alghamdi, Eman alBilali, Fatima Haouari, Firoj Alam, Go Inoue, Hamdy Mubarak, Haonan Li, Injy Hamed, Kareem Darwish, Khalid Almubarak, Khalid Elmadani, Mais Alheraki, Maram Hasanain, Maryam Alshihri, Mohamed Anwar, Muhammad Elmallah, Muneera Alhoshan, Nizar Habash, Nora Altwairesh, Norah Alzahrani, Ola Aljarrah, Ossama Obeid, Preslav Nakov, Raghad Al-Rasheed, Rawan Al-Matham, Safa Alajlan, Saied Alshahrani, Shady Shehata, Sultana Alghurabi, Tamer Elsayed, Waad Alshammari, Zaid Alyafeai","submitted_at":"2025-07-30T12:16:39Z","abstract_excerpt":"The impressive advancement of Large Language Models (LLMs) in English has not been matched across all languages. In particular, LLM performance in Arabic lags behind, due to data scarcity, linguistic diversity of Arabic and its dialects, morphological complexity, etc. Progress is further hindered by the quality of Arabic benchmarks, which typically rely on static, publicly available data, lack comprehensive task coverage, or do not provide dedicated platforms with blind test sets. This makes it challenging to measure actual progress and to mitigate data contamination. Here, we aim to bridge th"},"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":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2507.22603","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-30T12:16:39Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"898cd7fca0e84a8c1c412a06e940a0de2760b8bef1576648cfe2c1f385804073","abstract_canon_sha256":"f3cdec5c12775a829c5ef24e9a9914a22ffa43a93266fea8702a56ecfdd0c06c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:45:45.306679Z","signature_b64":"OjtxvA+XrXSG6sYPby7+Jhc/xtCiFRH66sQ1ptBS3+iNGHlgN/cd6sllyS9waXdJvYkUsaFNAslfrrFmqLGRAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e159a517ad87c8b18b16ac81191e41cbd49091be86ee1b25afeea554dba33e22","last_reissued_at":"2026-07-05T11:45:45.306174Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:45:45.306174Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"BALSAM: A Platform for Benchmarking Arabic Large Language Models","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Abdelrahman El-Sheikh, Abdullah Alfaifi, Abdulmohsen Al Thubaity, Abdulrahman AlOsaimy, Afrah Altamimi, Ahmed Abdelali, Amal Almazrua, Areej Alshaqarawi, Asma Al Wazrah, Atikah Alzeghayer, Bashar Alhafni, Emad Alghamdi, Eman alBilali, Fatima Haouari, Firoj Alam, Go Inoue, Hamdy Mubarak, Haonan Li, Injy Hamed, Kareem Darwish, Khalid Almubarak, Khalid Elmadani, Mais Alheraki, Maram Hasanain, Maryam Alshihri, Mohamed Anwar, Muhammad Elmallah, Muneera Alhoshan, Nizar Habash, Nora Altwairesh, Norah Alzahrani, Ola Aljarrah, Ossama Obeid, Preslav Nakov, Raghad Al-Rasheed, Rawan Al-Matham, Safa Alajlan, Saied Alshahrani, Shady Shehata, Sultana Alghurabi, Tamer Elsayed, Waad Alshammari, Zaid Alyafeai","submitted_at":"2025-07-30T12:16:39Z","abstract_excerpt":"The impressive advancement of Large Language Models (LLMs) in English has not been matched across all languages. In particular, LLM performance in Arabic lags behind, due to data scarcity, linguistic diversity of Arabic and its dialects, morphological complexity, etc. Progress is further hindered by the quality of Arabic benchmarks, which typically rely on static, publicly available data, lack comprehensive task coverage, or do not provide dedicated platforms with blind test sets. This makes it challenging to measure actual progress and to mitigate data contamination. Here, we aim to bridge th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.22603","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2507.22603/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2507.22603","created_at":"2026-07-05T11:45:45.306235+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.22603v1","created_at":"2026-07-05T11:45:45.306235+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.22603","created_at":"2026-07-05T11:45:45.306235+00:00"},{"alias_kind":"pith_short_12","alias_value":"4FM2KF5NQ7EL","created_at":"2026-07-05T11:45:45.306235+00:00"},{"alias_kind":"pith_short_16","alias_value":"4FM2KF5NQ7ELDCYW","created_at":"2026-07-05T11:45:45.306235+00:00"},{"alias_kind":"pith_short_8","alias_value":"4FM2KF5N","created_at":"2026-07-05T11:45:45.306235+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.06456","citing_title":"Context-Aware Dialectal Arabic Machine Translation with Interactive Region and Register Selection","ref_index":4,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4FM2KF5NQ7ELDCYWVSARSHSBZP","json":"https://pith.science/pith/4FM2KF5NQ7ELDCYWVSARSHSBZP.json","graph_json":"https://pith.science/api/pith-number/4FM2KF5NQ7ELDCYWVSARSHSBZP/graph.json","events_json":"https://pith.science/api/pith-number/4FM2KF5NQ7ELDCYWVSARSHSBZP/events.json","paper":"https://pith.science/paper/4FM2KF5N"},"agent_actions":{"view_html":"https://pith.science/pith/4FM2KF5NQ7ELDCYWVSARSHSBZP","download_json":"https://pith.science/pith/4FM2KF5NQ7ELDCYWVSARSHSBZP.json","view_paper":"https://pith.science/paper/4FM2KF5N","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.22603&json=true","fetch_graph":"https://pith.science/api/pith-number/4FM2KF5NQ7ELDCYWVSARSHSBZP/graph.json","fetch_events":"https://pith.science/api/pith-number/4FM2KF5NQ7ELDCYWVSARSHSBZP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4FM2KF5NQ7ELDCYWVSARSHSBZP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4FM2KF5NQ7ELDCYWVSARSHSBZP/action/storage_attestation","attest_author":"https://pith.science/pith/4FM2KF5NQ7ELDCYWVSARSHSBZP/action/author_attestation","sign_citation":"https://pith.science/pith/4FM2KF5NQ7ELDCYWVSARSHSBZP/action/citation_signature","submit_replication":"https://pith.science/pith/4FM2KF5NQ7ELDCYWVSARSHSBZP/action/replication_record"}},"created_at":"2026-07-05T11:45:45.306235+00:00","updated_at":"2026-07-05T11:45:45.306235+00:00"}