Research On Competitor Classification System Based On Patent Analysis

Continuously develop and improve the technology of today, technological position in the competition has become increasingly prominent, while the patent system to protect technological innovation as an important measure to be careful of their effective competition in the market research companies with an important meaning. As the modern business, the competitors determine boundaries become increasingly blurred relationship between enterprises may change at any time, how to integrate effective use of enterprise resources to enterprises and potential rival to track it is particularly important, so by patent analysis of competitors from the technical aspects of classification, so that rival companies to detect and correct to judge the skill level and characteristics of competitors, accurate evaluation of the technical strength of our competitors has important practical significance.This paper from the enterprise's practical point of view, in the full study at home and abroad on the basis of relevant information, proposed to build Competitors based on patent analysis of the idea category is hoped that by building such a system to enable enterprises to accurately evaluate the competitors skills, decision-making for the company to provide a reference standard, so enterprises can allocate limited resources.This study mainly in three aspects:(1) Construct a set of competitors meet the practical needs of enterprise patent classification system, which select the appropriate patent evaluation, the use of benchmarking methods and data mining technology for enterprise analysis of competitor's patent information, from the technical aspects classification of the enterprise's competitors.(2) In this paper, an improved K-means algorithm, through the establishment of centers and type of automatic consolidation and other methods to avoid a large extent K-means algorithm depends on the initial value, the number of clusters need to set and the outlier sensitive issue, so that it can rival the actual situation of patent analysis(3) Though the experiment, the K-means algorithm and the construction of the competitor classification system is effective.