Agent Evaluation via Social Network Analysis in Educational E-CARGO Framework

Artificial intelligence technology is being increasingly applied in-depth in education. Smart education is a typical collaboration among agents that includes the negotiation, evaluation, allocation, and execution of agents in different roles. The social network of agents and environmental elements has impact on such collaboration. This paper first describes the process as the Role Based Collaboration (RBC) workflow and a real world scenario. Then the E-CARGO model for education is defined, accompanied with a social network structure definition. As the core of this paper, Social Network Analysis (SNA) approach is investigated on its application to the agent evaluation in edu-RBC context. A series of algorithms are defined and evaluated on a real world Teacher-Course-Student (TCS) data. The experimental result reveals that the path-based SNA algorithms have great potential to outperform existing baseline methods, especially when the data is accumulated over time.