Scholar Recommendation Model in Large Scale Academic Social Networking Platform

摘要

A scholar-recommended model based on community division is established due to the characteristics of social intercourse of academic social network. The model was developed by GraphChi, the single version of large-scale graphic computing system which was launched by GraphLab, to find the core network in parallel on network topology map. In the established network, using self-adaptive label transmission to create labels and then according to the number of labels on the nodes to get final results of community division. Calculation is done within the community for expert recommendation services. The experiment of data-set on academic social networking platform, SCHOLAT, suggests, models not only can create community quickly, but also can gain good recommendation results by all the three personalized methods, i.e. Community Weight Recommended (CWR), Community Random Recommended (CRR) and Acquaintance Community Recommended (ACR).