How to Build Trust-Aware Voting Advice Applications?

摘要

Voting advice applications (VAAs) are intelligent systems that provide personalized recommendations of political candidate/party to a voter with regard to her/his political attitudes. The existing gaps related to the improvements of the system design will be addressed in this paper. The paper presents a trust-aware Voting Advice Application which aims to improve the recommendation accuracy, in this way to facilitate voter's decision-making processes and enhance citizens' participation in eDemocracy. By establishing the candidates-to-voters communication via the forum channel, alternatively to traditional VAAs, the proposed system calculates the similarity of political profiles between voters and candidates including into the computation the candidates' reputation based on her/his behaviour within the forum. To test the proposed prototype, the evaluation framework has been designed, and a user study has been conducted. The results show that trust-aware VAAs give satisfactorily accurate and helpful recommendations. Furthermore, causal relationships of the evaluated constructs demonstrate that by participating in forum discussions, a user can improve the accuracy of her/his recommendation that is further helpful in her/his decision-making processes.