Abstract: There are usually nodes of different types in heterogeneous networks.In order to satisfy requests of information retrieval for different types from users,different types of nodes need raking.In order to solve the above problem and improve the effectiveness at the same time,this paper proposed a random walk based node ranking algorithm in he-terogeneous networks.Firstly,we formally described a heterogeneous network including nodes of users,images and text.Secondly,we defined the similarity between nodes,and proposed a modified similarity measure that includes neighbors of different types.Thirdly,we proposed an algorithm for computing importance of node based on random walk,and analyzed the selection of bias vector in the random walk model.Finally,according to massive experiments on real dataset,we validated the effectiveness of the proposed approach while handling users’ request in information retrieval.

Key words: Heterogeneous networks,Image retrieval,Random walk,Ranking algorithm