Abstract: 【Purpose/significance】Extracting accurate keywords from massive microblog information, in order to provide valuable refer⁃
ence for government and enterprises to analyze public opinion. 【Method/process】Through the analysis of the characteristics of tradition⁃
al microblog keywords extraction method and the insufficiency, proposed microblog keywords extraction method based on the semantic
concept and word co-occurrence, the method uses text expansion strategy to expand microblog from short text to long text, by means of
semantic dictionary to do semantic concept extenseion for microblog words, combining with the characteristics of microblog to distrib⁃
ute structure weight of vocabulary, and considering the degree of co-occurrence words to extract microblog keywords.【Result/conclu⁃
sion】The experimental results show that the microblog subject word extraction algorithm proposed in this paper is superior to tradition⁃
al methods. It can effectively improve the performance of microblog subject word extraction.【Innovation/limitation】It is a new explora⁃
tion to use semantic concepts combined with the idea of word co-occurrence to extract microblog subject words. Since the word seg⁃
mentation method in the algorithm may not be appropriate for the segmentation of some new network words, there is a slight impact on
the accuracy of keyword extraction.