Link Prediction in Drug-drug Interaction Network with Syncretic Naive Bayes Method
Runfang Wang, Zengqiang Chen, Zhongxin Liu · 2019 IEEE 8th Data Driven Control and Learning Systems Conference (DDCLS) · 2019
As is known to all, the pharmacological effects of drugs may change, and even generate unpredictable consequences when several drugs are taken simultaneously. However, the connection information of complex network with drug as the node is incomplete and inaccurate, and link prediction can be used as a favorable auxiliary tool to solve such problems. In this paper, a Syncretic Naive Bayes (SNB) method is developed by utilizing the Local Naive Bayes (LNB) method and node degree to quantify the influence of common neighbors and node pairs themselves, respectively. Subsequently, the proposed method is extended with the idea of Common Neighbors (CN), Adamic-Adar (AA) and Resource Allocation (RA) methods. Experimental results on drug-drug interaction network demonstrate the effectiveness and applicability of the proposed method, especially when the observed network is very sparse.