Trust Discovery in Online Communities

This research aims to discover interpersonal trust in online communities. Two novel trust models are built to explain interpersonal trust in online communities drawing theories and models from multiple relevant areas, including organizational trust models, trust in virtual settings, speech act theory, identity theory, and common bond theory. In addition, the detection of trust in online communities is automated by leveraging natural language processing techniques. Online communities continue to grow on the internet and vary from grass roots organizations to communities facilitated by large corporations. Examples of increased use of social networks include seeking healthcare, financial, and technical advice. Topics such as these stress the importance of trust between individuals in online communities. Although trust has been widely studied in the literature, the question of how trust evolves in online communities remains as a research gap. This research seeks to model the evolution of trust in online communities to address this gap. Establishing practical trust models provides opportunities for new algorithms for discovering trust relationships in online communities. Today trust is typically measured through the use of psychometric surveys that do not scale with the growth of online communities. Alternatively, the creation of automated trust discovery tools would provide benefit to online community managers in moderating communities. First the research extends organizational trust theories to online communities. Specifically, the Calculus-Based Trust (CBT) and Knowledge-Based Trust (KBT) theories showed high correlation to trust relationships in various online communities. Moreover, in view of the evolvement of trust relationships, CBT was found to precede KBT. The extension of CBT and KBT was validated through empirical survey using active participants in online communities such as financial investing, healthcare, shopping, and technology communities. To help operationalize the theory, a formal trust model was proposed using speech act theory. The model was tested in a financial investing community, and discussion threads were discovered that matched this model. The formal trust model sets a foundation for applying natural language processing techniques to text in discussion threads, allowing the development of new tools for online community managers. Next, an identity-based trust model was developed using the artifacts of virtual co-presence, deep profiling, and self-presentation to predict CBT and KBT. This finding resulted from an empirical study using the same online community participants that validated CBT and KBT in online communities. Algorithms for discovering likely trustees in online communities can be facilitated by knowing that artifacts provide potential indicators of individuals serving as trustees. Lastly, a two-part trust discovery algorithm is proposed to automatically find trust relationships in online communities. The first part of the algorithm consists of a speech act classifier to categorize each sentence in a discussion thread as one of four speech acts that are relevant to the trust model in this dissertation. The second part of the algorithm involves applying similarity measures to rank speech act pairs and then using the ranking score with additional features to find trustors in a discussion thread. &#8195