Selection of Core Words from Textual Patent Data with DEA based on Citation

Abstract

The web includes enormous data such as patents. The purpose of this research finds the rule of textual patent data and creates new model. Hence, we suggest new weighted method using DEA to handle unstructured data like patent. Our proposed method is advantageous because this considers the value of the patent compared with TF-IDF and other weighted methods. Using suggested method, we probe new text-mining in the field of patent.