Two-Phase Decision Fusion Based On User Preference

In order to prevent information overload on the Web, people sometimes rely on critics’ evaluations to narrow down their search for favorite objects. However, due to the subjectivity of humans’ perception, the set of favorite objects varies from one person to the other. This is because each object can be reviewed using different criteria, and different people have different reliance on the same set of criteria and/or critics. Therefore, the task of fusing the decisions of different critics considering the users’ preferences is challenging. To address this challenge, we introduce a new concept, termed two-phase decision fusion, where both critics and criteria evaluations are considered in order to personalize a search. In each phase, we utilize various decision fusion operators and investigate their interplay. We conducted several empirical experiments within the context of two real-world application domains. Our experimental results indicate that the behaviors of the various decision fusion methods stay consistent across the two different application domains.