Multi-Agent Joint Learning from Argumentation

摘要

Joint learning from argumentation is the idea that groups of agents with different individual knowledge take part in argumentation to communicate with each other to improve their learning ability. This paper focuses on association rule, and presents MALA, a model for argumentation based multi-agent joint learning which integrates ideas from machine learning, data mining and argumentation. We introduce the argumentation model Arena as a communication platform with which the agents can communicate their individual knowledge mined from their own datasets. We experimentally show that MALA can get a shared and agreed knowledge base and improve the performance of association rule mining.