Academic Venues Recommendation Based On Scholarly Information Network

Scholarly information network have risen beyond the imagination for the rapid development of information technology. It is necessary for researchers to acknowledge high quality and fruitful academic venues, participate popular and advanced conference, commit manuscript to journals or conferences with opportunity to public. However, the information overload problem in big scholarly data creates tremendous challenges for mining these venues and relevant information. In this paper, we work on metrology analysis on big scholarly data and proposed the personalized academic venues recommendation based on big scholarly data.Firstly, we did some metrology analysis on academic venues based on scholarly data. We researched the group structure of academic social, the association between numbers of venues researcher participated and personal academic level, academic venues' influence on researchers. We measured the influence of academic venues on researchers'collaboration and bacon number. The work prove that, academic venues recommendation is necessary and important because of the contribution on promoting academic communication, enhance academic collaboration and improving the academic level.Secondly, we focus on academic venues recommendation and proposed a personalized academic venues recommendation based on scholarly information network (PAVE). PAVE model conducts the scholarly information networks with two kinds of associations, coauthor relations and author-venue relations. A random walk with restart model runs on the scholarly information network drove by a transfer matrix with bias. We define the transfer matrix with bias by exploiting three academic factors, co-publication frequency, relation weight and researchers'academic level.Finally, we conducted extensive experiments on DBLP data set to evaluate the performance of PAVE model and other three baseline models. The results demonstrate that. PAVE model performs best on precision, recall, F1 score and Ave-Quality. The recommending efficiency of PAVE shows preferable. Moreover, comparing with other model, PAVE is better at making academic venue recommendation for junior researchers who have few publications. In addition, our experiments found the best settings of damping coefficient in PAVE model.