Multi-Granularity Information Expression Application on Patent Text Clustering

With the increasement of the patents amount, people pay more attention on the patent researcher. Having analyzed the research on the patent, this paper finds that the patent cluster is almost the basis of some other works. The main contents of the patent cluster consist of patent information selection, the information expression form and the cluster algorithm. The paper makes patent cluster improvements on information selection and information representation. Extract information from abstract text, the attribute-value pairs of abstract and the patent title. With the distributed representation, this paper obtains information expression and information fusion. The result shows that this algorithm enhances the experimental effect. This method can be applied to the patent clustering research.