摘要 :
Spatial information exerts significant inaence on user behaviors in e-commerce. In this paper,we propose a user behavior model using namic Bayesian network (DBN) with mixed-memory Markov dependency, with consideration of spatial i...
展开
Spatial information exerts significant inaence on user behaviors in e-commerce. In this paper,we propose a user behavior model using namic Bayesian network (DBN) with mixed-memory Markov dependency, with consideration of spatial information (e.g., user'slocations). Our work distinguishes spatially related user features加m intrinsic preferences.We also suggest an expectation-maximization (EM)一based algorithm for our model.Some experiments are made to eval-
aate our model, using real-world data mane-commerce search portal.Experimental resalts show the proposed model is capable of user behavior prediction.
收起
