A Study About Academic Network Community's Cluster Based On K-means Algorithm

With the development of the Internet,there has been more and more virtual community and people tends to communicate by it.Academic network community as part of virtual community,it has become the scholars,enthusiasts and researchers first choose to discuss and get knowledge,the platform not only provides access to approach knowledge also provides a better convenient communication platform.In the actual application,the users may only search the community for a certain problem,and the academic network community mainly classifies according to the subject category.The other features of the academic network community have not been well dug,and its characteristics have not been subdivided,which can affect the time,cost and accuracy of knowledge acquisition.The related concepts of academic network community and related theories of data mining technology are given,and studied the basic principle of K-means algorithm in clustering analysis.On this basis,First,preprocessed the data then cleaned the data,word frequency statistics,word matrix construction and cosine similarity evaluation are achieved through the acquisition of keywords that display the characteristics of academic network community.The distance between hot words is obtained by using cosine similarity,and K-means clustering is used for the above numerical value.The main tool is SPSS.According to the clustering results,the academic network community can be divided into:Industry learning research academic network community,Professional academic network community,Question answering academic network community and Comprehensive academic network community.From the characteristics of each class,the development proposal of this kind of academic network community is proposed.Clustering the academic network community can improve the use efficiency of the academic network community,and also be beneficial to the clear future development goal of the academic network community.