Integrating Trust in Argumentation Based Recommender Systems

With the enormous growth of the Internet, trust has become an increasingly important issue in Agent-based E-commerce. Argumentation technologies are needed for autonomous agents to come to mutually acceptable agreements, on behalf of humans. Agents can argue over each other's beliefs, desires and planning. It is also important for these agents to be able to compute their trust in other agents. Especially, in an argumentation-based recommendation system, the arguments uttered to persuade a customer over a product are not the result of an isolated analysis, but of an integral view of the preferences, goals and options available. In our opinion, trust and argumentation together can improve the recommendation process. This paper describes our work on using argumentation to handle and update trust in the agent-based recommender systems and vice versa. This paper proposes integration of fuzzy trust with an argumentation framework to enable the agents in reasoning about the beliefs, desires and plans with trust. This integration allows the user to take well-reasoned decisions based on trustworthy recommendations. As a result, trust in an agent increases when it generates more of acceptable arguments thereby reducing the number of messages passed and time consumed by the agents. This improves performance of the agents in terms of the communication overhead caused and time taken for decision making. The same was established by the results obtained from experiments conducted for a Book Recommender System (RS).