Personalized Learning Diagnosis Approach with Multi-Agent Collaboration and Four-Layer Anti-Hallucination Mechanism
Journal of Big Data and Computing · 2026-09 · 4 (3) · 155-165
DOI: 10.62517/jbdc.202601320 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Big Data and Computing, 2026-09. Verfasst von Zhengxuan Luo und 3 weiteren Autor:innen. Die Arbeit nennt 15 Literaturangaben.
Abstract (Originalsprache)
Aiming at uneven learning foundations, opaque teaching workflows and hallucination-reasoning defects of single-LLM solutions in university AI education, this paper presents ZhiCheng, a personalized learning-diagnosis system with the "1+7+8" multi-agent architecture. Built upon the dual-brain functional partition, seven core agents are scheduled by an Orchestrator, while eight extended nodes constitute a quality-control loop. A four-tier anti-hallucination pipeline cuts factual error rate from 15%-25% to below 1%. Under 1250 concurrent users, the system reaches 842 TPS, and index tuning reduces core-query P95 latency by 60%. Experiments, ablation studies and closed-loop tests validate the architecture superiority and system effectiveness.
Zitieren
APA
Luo, Z., Fan, H., Chen, W., & Gong, M. (2026). Personalized Learning Diagnosis Approach with Multi-Agent Collaboration and Four-Layer Anti-Hallucination Mechanism. Journal of Big Data and Computing, 4(3), 155-165. https://doi.org/10.62517/jbdc.202601320
Vancouver
Luo Z, Fan H, Chen W, Gong M. Personalized Learning Diagnosis Approach with Multi-Agent Collaboration and Four-Layer Anti-Hallucination Mechanism. Journal of Big Data and Computing. 2026;4(3):155-165. doi:10.62517/jbdc.202601320
BibTeX
@article{Luo2026,
author = {Luo, Zhengxuan and Fan, Huaiming and Chen, Wen and Gong, Mingming},
title = {Personalized Learning Diagnosis Approach with Multi-Agent Collaboration and Four-Layer Anti-Hallucination Mechanism},
journal = {Journal of Big Data and Computing},
year = {2026},
volume = {4},
number = {3},
pages = {155--165},
doi = {10.62517/jbdc.202601320}
}
EUVIMED