Research Of Relativity Of Web Based On Co-occurrence Graph Of Keywords

Extracting Keywords of text refers to quantifying the words in the text body which are able to deliver the theme of the article; it also refers to text information. As the middle representation of text, these keywords are used to analyze the similarity between the unknown texts in information searching and intelligent retrieval. The diversity and complexity of message that in Network text made it difficult to obtain a standard form description of text characterization. Word frequency calculation is to calculate the frequency of different words in the same sentence, so as to reveal the relevance of different paragraphs in the same text or different texts.In present study, vector is mainly used to represent various keywords and their characteristics in an article which called characterized vectors. Then calculate the Relativity base on it. This paper presents the new approach of article expression by using co-occurrence graph which represent relation of keywords through graph, so as to deliver the relation of different texts. There are two steps in this method; first is to express the text with graph. Next is to analyze the similarity of formed graph. This implements a plan to meet the current method with the use of text clustering processing system.This paper also presents a system which is able to handle the texts with co-occurrence graph. The system can automatically extract the text content as well as keywords by using the Chinese Word Segmentation, and then create the graph of keywords and co-occurrence relationship. The system also shows how to apply a keyword graph in text clustering. And finally evaluate it with Recall and Precision. And prove the validity and advantage of clustering through the experiment.