Technology matching of the patent documents using clustering algorithms

This paper analyzes the accuracy of different clustering algorithms to handle different parts of the patent documents. The algorithms are part of the software package which is used as a tool for business intelligence purposes. The tool assembles patent data from publicly available data bases, collects and analyzes patents bibliographic parameters and performs text mining. Performances of clustering algorithms: k-means, the neural-gas; fuzzy c-means and ronn algorithm are examined when run on different parts of the patent document, such as abstract, claim, international patent code description and detailed patent description, but applied on the same patent data set. Patent data set was previously classified in technology groups by the experts and obtained results are compared with the purpose of selection of the most suitable algorithm and patent document part.