A Naive Bayesian Multi-label Classification Algorithm With Application to Visualize Text Search Results

Search results visualization has emerged as an important research topic due to its application on search engine amelioration. From the perspective of machine learning, the text search results visualization task fits to the multi-label learning framework that a document is usually related to multiple category labels. In this paper, a Näıve Bayesian (NB) multi-label classification algorithm is proposed by incorporating a two-step feature selection strategy which aims to satisfy the assumption of conditional independency in NB classification theory. The experiments over public data set demonstrate that the proposed method has highly competitive performance with several well-established multi-label classification algorithms. We implement a prototype system named TJ-MLWC based on the proposed algorithm, which acts as an intermediate layer between users and a commercial Internet Search Engine, allowing the search results of a query displaying by one or multiple categories. Testing results indicate that our prototype improves search experience by adding the function of browsing search results by category.