Online rating system is a popular feature of Web 2.0 applications. It typically involves a set of reviewers assigning rating scores (based on various evaluation criteria) to a set of objects. We identify two objectives for research on online rating data, namely achieving effective evaluation of objects and learning behaviors of reviewers/objects. These two objectives have conventionally been pursued separately. We argue that the future research direction should focus on the integration of these two objectives, as well as the integration between rating data and other types of data.

This paper extends previous work from the authors' research group in their Summarizing review scores of "unequal" reviewers paper. It provides theoretical information of how and why online rating systems should combine the evaluation of rating data and behavioural learning (non-rating data that help evaluate and infer assumptions based on reviewers behaviour) to calculate a fair quality rating of an object. As future work, the authors have suggested the possibility of combining rating data with social network data (i.e. trust / reputation data). I also feel that rating data can be utilised as a mechanism for assessing user contributions.

• Evaluation of rating data
• Behavioural learning
• Clustering
• Data mining techniques
• Deviation - Bias and controversy