Abstract

This research investigates how synergies between the Web and social networks can enhance the process of obtaining relevant and trustworthy information. A review of literature on personalised search, social search, recommender systems, social networks and trust propagation reveals limitations of existing technology in areas such as relevance, collaboration, task-adaptivity and trust. In response to these limitations I present a Web-based approach to information-seeking using social networks. This approach takes a source-centric perspective on the information-seeking process, aiming to identify trustworthy sources of relevant information from within the user's social network. An empirical study of source-selection decisions in information- and recommendationseeking identified five factors that influence the choice of source, and its perceived trustworthiness. The priority given to each of these factors was found to vary according to the criticality and subjectivity of the task. A series of algorithms have been developed that operationalise three of these factors (expertise, experience, affinity) and generate from various data sources a number of trust metrics for use in social network-based information seeking. The most significant of these data sources is Revyu.com, a reviewing and rating Web site implemented as part of this research, that takes input from regular users and makes it available on the Semantic Web for easy re-use by the implemented algorithms. Output of the algorithms is used in Hoonoh.com, a Semantic Web-based system that has been developed to support users in identifying relevant and trustworthy information sources within their social networks. Evaluation of this system's ability to predict source selections showed more promising results for the experience factor than for expertise or affinity. This may be attributed to the greater demands these two factors place in terms of input data. Limitations of the work and opportunities for future research are discussed.

Citations