AVER

Abstract

Academic venues act as the main platform of communities in academia and the bridge of connecting researchers, which have rapidly developed in recent years. However, information overload in big scholarly data creates tremendous challenges for mining useful and effective information in order to recommend researchers to acknowledge high quality and fruitful academic venues, thereby enabling them to participate in relevant academic conferences as well as contributing to important/influential journals. In this work, we propose AVER, a novel random walk based Academic VEnue Recommendation model. AVER runs a random walk with restart model on a co-publication network which contains two kinds of associations, coauthor relations and author-venue relations. Moreover, we define a transfer matrix with bias to drive the random walk by exploiting three academic factors, co-publication frequency, weight of relations and researchers' academic level. AVER is inspired from the fact that researchers are more likely to contact those who have high co-publication frequency and similar academic levels. Additionally, in AVER, we consider the difference of weights between two kinds of associations. We conduct extensive experiments on DBLP data set in order to evaluate the performance of AVER. The results demonstrate that, in comparison to relevant baseline approaches, AVER performs better in terms of precision, recall and F1.