A Network Based Approach For Expertise Recommendation In Academic Contexts

Expertise management plays an important role in organizations.It aims to improve the performance of organizations by facilitate effectively utilization of expertise knowledge.Due to information overload and implicit nature of expertise, it is difficult to find experts with required expertise.Expertise recommendation as a subfield of expertise management aims to alleviate the information overloading problem and provide users with experts who might satisfy users'needs.Previous research on expertise recommendation is usually from two separate research streams. The first stream focuses on exploring the social connections of experts.The second stream focuses on investigating the content of the experts'knowledge. However, these two dimensions of expertise recommendation are seldom integrated. Besides, previous research usually used exact matching to find experts, and the semantic analysis of their expertises is lacked.To address these problems, this thesis presents an integrated network based approach which combines social relations and semantic analysis for expertise recommendation in academic contexts.Social network analysis is used to represent the communication relationships of researchers,and semantic analysis is used to capture the content of researchers'expertise.Based on the proposed network based approach, I investigate how the approach can be used to recommend researchers in research communities and recommend reviewers in peer review settings through case studies.In summary, this dissertation makes contributions to both academic and practical fields. For academia, it is a typical design science research contributing to expertise management by proposing two new artifacts to solve problems existing in actual worlds:find appropriate reviewers in a government funding agency and recommend researchers in a research community. With respect to practical contribution, the artifacts proposed in this research can be used in actual settings and has potential to integrate with existing systems to improve the performance of expertise management. In addition, the proposed approaches and models in this research have the potential to be adapted and adopted in other practical settings.