Clustering Analysis Of Feature Words In News Text Based On Co-Occurrence Matrix

In this paper, we use a new method to improve the context of a news. Especially, a new TF-IDF formula is used to calculate the weight of the feature word of News. We set a parameter W to the feature words, which takes into account the factors such as the parts of speech of feature words and inverse document frequency. The parameter is adjusted based on the K-core theory, and therefore to determine the range of feature words. Our work aims to analysis the co-occurrence strength of news feature words in a certain period of time, and obtains the cluster analysis of the co-occurrence intensity distance of news feature words. The simulation results of clustering analysis is significant and the quantitative research methods can be commonly used in bibliometrics in the news field to analyze the news content.