Assoc. Prof. Sun Pengfei Wins Grand Prize at 2nd Sichuan AI Teaching Competition

Time : Oct 23, 2026 Browse the number :

On October 19, the Sichuan Provincial Department of Education publicized the list of proposed award-winning works for the 2nd Sichuan Provincial Teachers' Artificial Intelligence Application Ability Competition.

The project ‌Wenshu NexusRail: A Research Literature Working Platform Based on Evidence Chain‌, jointly completed by teachers Sun Pengfei, Wang Qingyuan, Xiao Zhuang, Huang Deqing and Qin Na from our college, stood out from numerous entries across the province with its outstanding innovation and practicality, and won the Special Prize in the direction of "AI Educational Agent Design and Application".

It is reported that this competition is hosted by the Sichuan Provincial Department of Education and undertaken by ‌Chengdu Normal University‌, with the theme of "AI Transforms-Smart Education for the Future". It collects AI education application achievements from teachers across the province, aiming to promote teachers to explore innovative applications of artificial intelligence technology based on actual teaching practice, and improve teachers' digital literacy and AI application capabilities. After recommendations from cities and universities, a total of 1917 works from the whole province participated in the provincial final, featuring a large scale of participation and fierce competition.

Our college attaches great importance to the work of AI empowering education and teaching, actively mobilizes teachers to participate in the competition, and selects and recommends excellent works within the college to participate in the provincial final. The faculty team's work stood out and won this honor on behalf of our college.

The project "Wenshu NexusRail" is a self-developed multi-agent research literature work platform tailored for postgraduate research competence cultivation. With the core concept of "scaffolding-style AI assistance", it builds an integrated research workflow covering "retrieval → parsing → reading → review → citation → verification" around the principles of "problem-driven, evidence-supported, process-tracked, knowledge-precipitated".

The platform's core innovations are reflected in three aspects: interpretable retrieval and LCP-structured in-depth parsing, which mandatorily identifies limitations, potential biases and implicit assumptions of papers; evidence-driven review generation via five-agent collaboration, with fully traceable evidence across the whole process; and an integrated disciplinary knowledge base that breaks the silos of research group achievements and enables the reuse of knowledge assets. In terms of application effects, the processing efficiency of a single piece of literature has been greatly improved, and the accuracy of large-scale literature downloading and citation remains at a high level. Controlled experiments show that students assisted by this platform significantly outperform the general AI group in dimensions of argumentation logic and metacognitive reflection. Relevant achievements have been compiled into 2 EI papers, published at 2026 45th Chinese Control Conference‌ (CCC).






【关闭】