GraphSkillAA：用归因引导技能图更新，支持定向验证与回滚

View PDF HTML (experimental)

Abstract:External skills provide domain knowledge and procedures without updating model parameters, but flat text makes their applicability, dependencies, and composition difficult to express. Existing skill graphs organize rules into nodes and relations mainly for retrieval and composition. They give less attention to what should change after a failure, which examples should be retested, and which edits should be reverted. We introduce GraphSkillAA, short for Graph-Skill Abductive Attribution, which connects rule use, error localization, and update validation through the same skill graph. Nodes record applicability, procedures, and exclusion conditions, while edges describe prerequisite and enhancement relations. The student records the nodes and edges used during execution. The teacher contrasts related successful and failed trajectories to identify the fields or relations that need repair, keeping changes as small as possible. Execution lapses or insufficient evidence leave the graph unchanged. A Local Gate retests affected examples, while a Big Gate checks the merged graph on the complete update pool. Rejected edits are rolled back, and model parameters remain frozen throughout. With GPT-5.6-sol, GraphSkillAA reaches 81.5%, 66.7%, and 91.2% on SearchQA, LiveMath, and DocVQA, respectively, and achieves the highest mean performance in every main setting. These results show that graph structure supports skill selection and composition while also providing explicit targets for repair, retesting, and rollback.

Subjects:	Artificial Intelligence (cs.AI)
Cite as:	arXiv:2609.20455 [cs.AI]
	(or arXiv:2609.20455v3 [cs.AI] for this version)
	https://doi.org/10.48550/arXiv.2609.20455 arXiv-issued DOI via DataCite

Submission history

From: Ziqiao Shang [view email]
[v1] Thu, 17 Sep 2026 14:20:45 UTC (6,009 KB)
[v2] Wed, 23 Sep 2026 18:12:44 UTC (7,130 KB)
[v3] Thu, 12 Nov 2026 10:40:34 UTC (792 KB)