A New Document Retrieval Model Using Dempster-Shafer Theory of Evidence

Vector space model is a widely used conventional document retrieval model. It judges the relevance between a query and a document by calculating the similarity between them. However, vector space model is not satisfactory in recall and precision. In this paper, we propose a new document retrieval model using Dempster-Shafer theory of evidence. Documents are segmented into sentences firstly. Then similarities between a query and every sentence are calculated. As our model is within Dempster-Shafer theory of evidence, the combination rule of the theory is utilized to calculate the similarity between a query and a document. To evaluate our model, experiments are carried out and the results show that the model outperforms vector space model in recall and precision.