Research on keyword extraction of Tibetan web news based on improved TEXT-RANK algorithm

摘要

This paper proposes an improved TEXT-RANK algorithm and applies it to keyword extraction in Tibetan web news. In this paper, we first apply the graph model algorithm to Tibetan keyword extraction, and improve the TEXT-RANK algorithm. In text preprocessing, according to the characteristics of new words and proper nouns in network news texts, the named entity recognition method is integrated by CRF and MaxEnt. In this paper, the voting mechanism of TEXT-RANK algorithm is improved by studying the grammatical features of Tibetan and the writing rules of Tibetan network news. Experiments show that the improved TEXT-RANK algorithm can effectively improve the accuracy of keyword extraction.