Research on Information Retrieval Algorithm Based on TextRank

Abstract

Compared with people's urgent desire for information, the current information retrieval still has problems such as slow speed and low precision. In response to this problem, this paper proposes an information retrieval method based on NLP. Firstly, the semi-supervised learning algorithm is used to describe the natural language under the manifold condition, and the undirected graph is constructed for all the data. For the undirected graph after construction, we use the label propagation algorithm to simplify the undirected graph to reduce the computational complexity in the information retrieval process. Then the keywords will be extracted by TextRank algorithm to achieve information retrieval. Finally the experimental results show that the retrieval efficiency can be improved by using this algorithm.