Predicting employee relations in enterprises by bi-directional link mining

By using link mining method, we can understand associations among persons in enterprises. In this paper, we propose a link mining model called a bi-directional link mining, which is capable of predicting edges that do not exist but may appear in the future in dynamic multi-dimensional networks. Firstly, we construct a dynamic multi-dimensional networks from an E-mail network and a telephone network, which makes us correctly describe a Group´s staff relationship. Then we present some algorithms to build a reduction dimensional network and a dynamic network. After that, we give bi-directional link mining methods in a dynamic multi-dimensional weighted network and apply these methods in the enterprise´s employee relational network. Experimental results show that these methods can effectively predict associations among persons in a large scale employee networks.