An algorithm for maximal bicliques searching in dynamic relationship graph
Sihan Zhang, Mingxue Liao, Qingdu Xiao, Xia Hou, Pin Lv · 2018
Many tasks of complex data relation mining in social networks, bioinformatics and cognitive radios usually involve a process of enumerating maximal bicliques from a relationship graph which is generally represented by relation matrix. Currently most of researchers focus on the enumerating maximal bicliques from a large static graph which keep unchanged through the process of enumeration in big data mining. But in most situations, the graph will be dynamically changed since either vertices or edges of the graph will be removed or updated to reflect corresponding change in a real world. Therefore, it is impossible to employ those traditional techniques for static graphs to dynamically changed graph mining. For this reason, we propose an efficient algorithm to enumerate maximal bicliques in a dynamically changed graph. In this paper we assume that the edges of the graph will be dynamically reduced. For this scenario, our algorithm need not to repeat the enumeration on the whole graph, but only to iteratively search the new maximal bicliques based on the result before the graph is changed. In paper, we show the rightness of our methods and perform simulations which demonstrate high efficiency.