Keyword extraction method based on WS small world model

The invention discloses a keyword extraction method based on a WS small world model. Obtaining comment text data, and performing data cleaning on the comment text data to obtain a target statement of a text; preprocessing the target statement to obtain candidate words; secondly, taking candidate words in the candidate word set of the target statement as nodes in the WS model, constructing a statement word network graph G, and obtaining WS feature parameters of words; obtaining part-of-speech, position and TF-IDF feature parameters of the candidate words; and finally, according to the obtained WS feature parameters and other feature parameters, calculating the final weights of the candidate words, and setting the keywords as the first N candidate words of which the final weights are ranked from high to low. Compared with the prior art, the method has the advantages that the WS small-world model is fused with the characteristics of part of speech, position, TF-IDF and the like, not only can the correlation amon