Efficient and Effective Expert Finding based on Community Search: A Demonstration

In recent years, with the vigorous development of social networks, various kinds of data in the network have shown an explosive growth trend. Among them, a large amount of data in the field of academics includes rich and diverse entity information such as high-quality academic papers, experts, venues, and topics that have intricate and complex relationships, constituting important heterogeneous academic networks, e.g., DBLP. Many expert finding systems have been investigated on the academic network. But most of them are using textual keyword matching techniques to support the systems. Different from the above systems, we designed and implemented an expert finding system to effectively and efficiently return desired experts not only based on textual keyword matching, but also on the experts' relationship achieved by community search. It contains three layers: data processing layer, core algorithm layer, and application layer. The data processing layer is responsible for data collection and processing to construct heterogeneous academic networks. The core algorithm layer includes the academic network community search algorithm and Top-n expert finding through multiple recalls based on the Threshold algorithm. The application layer receives data from the core algorithm layer to present to users at the front end. On this basis, our core algorithms can also be migrated to other applications, e.g., recommendation, biological data analysis, reviewer assignment, and public safety protection.