Topic-Level Random Walk through Probabilistic Model

In this paper, we study the problem of topic-level random walk, which concerns the random walk at the topic level. Previously, several related works such as topic sensitive page rank have been conducted. However, topics in these methods were predefined, which makes the methods inapplicable to different domains. In this paper, we propose a four-step approach for topic-level random walk. We employ a probabilistic topic model to automatically extract topics from documents. Then we perform the random walk at the topic level. We also propose an approach to model topics of the query and then combine the random walk ranking score with the relevance score based on the modeling results. Experimental results on a real-world data set show that our proposed approach can significantly outperform the baseline methods of using language model and that of using traditional PageRank.