{"paper":{"title":"Context-Aware Web Attack Detection in Open-Source SIEM Systems via MITRE ATT&CK-Enriched Behavioral Profiling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Context features from prior events raise web attack detection F1 in open-source SIEM from 0.705 to 0.967.","cross_cats":["cs.LG"],"primary_cat":"cs.CR","authors_text":"Aref Shaheed, Assef Jafar, Badr Alboushy, Mohamad Aljnidi, Mohamad Bashar Disoki","submitted_at":"2026-05-13T10:54:36Z","abstract_excerpt":"Security Information and Event Management (SIEM) systems aggregate log data from heterogeneous sources to detect coordinated attacks. Traditional rule-based correlation engines struggle to classify multi-step web application attacks because they examine each event without reference to the behavioural history of the originating host.\n  We present Smart-SIEM, an AI module for the open-source Wazuh SIEM platform with two contributions: (1) a per-source-IP behavioural context vector encoding HTTP response-status distributions, peak rule activation counts, and MITRE ATT&CK technique frequencies fro"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Context features improve all tested gradient boosting algorithms from ~0.705 macro F1 to 0.947-0.967 (Stage 1) and 0.876-0.914 (Stage 2); the hybrid cascade achieves F1 of 0.967 (binary) and 0.914 (six-class). Wazuh's native rule engine detects 0% of Brute Force and Broken Authentication events; the AI module detects 100% and 98.3% respectively.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The purpose-built dataset of 46,454 Wazuh security events accurately reflects the distribution and behavioral patterns of real-world web attacks and normal traffic.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Smart-SIEM adds context-aware ML profiling to Wazuh SIEM, lifting binary attack detection F1 to 0.967 and six-class categorization to 0.914 while recovering from concept drift via retraining.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Context features from prior events raise web attack detection F1 in open-source SIEM from 0.705 to 0.967.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"5895749ffe3daee4560e966737f8c25a111d6172c9ac9a07f6352b18cd8529d4"},"source":{"id":"2605.13337","kind":"arxiv","version":1},"verdict":{"id":"08ce23df-0589-4857-a393-20de7c92fc8b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T18:29:41.191587Z","strongest_claim":"Context features improve all tested gradient boosting algorithms from ~0.705 macro F1 to 0.947-0.967 (Stage 1) and 0.876-0.914 (Stage 2); the hybrid cascade achieves F1 of 0.967 (binary) and 0.914 (six-class). Wazuh's native rule engine detects 0% of Brute Force and Broken Authentication events; the AI module detects 100% and 98.3% respectively.","one_line_summary":"Smart-SIEM adds context-aware ML profiling to Wazuh SIEM, lifting binary attack detection F1 to 0.967 and six-class categorization to 0.914 while recovering from concept drift via retraining.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The purpose-built dataset of 46,454 Wazuh security events accurately reflects the distribution and behavioral patterns of real-world web attacks and normal traffic.","pith_extraction_headline":"Context features from prior events raise web attack detection F1 in open-source SIEM from 0.705 to 0.967."},"references":{"count":55,"sample":[{"doi":"","year":2010,"title":"and Harris, Shon and Harper, Allen and VanDyke, Stephen and Blask, Chris , title =","work_id":"41960a47-bb82-461c-98ab-2bd793fa112c","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2012,"title":"and Schmidt, Kevin J","work_id":"9d8918c0-1ae4-4000-b86d-1e9c8dba9bfb","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2009,"title":"Event Correlation Engine , school =","work_id":"e9ff9605-0732-4b6b-a29f-2ffc667e5588","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"International Journal of Science and Research (IJSR) , volume =","work_id":"43d67098-8e62-4bd7-b254-475f20f1c5cc","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"Security Information and Event Management (","work_id":"cf5a6773-0e41-4b7a-9a9b-d146d295e9cf","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":55,"snapshot_sha256":"47c6b5422f064dfb3e2fb12157f2e44246086d8002cd24b7374c538fd96f7b87","internal_anchors":1},"formal_canon":{"evidence_count":1,"snapshot_sha256":"5413de02a8f8cc1a3356557079910c2b56a6cdbaeb9f2a299f06cbfa285157de"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}