The Support Vector Machine Classification System for Patent Document Information Importance Analysis
Chih‐Hung Wu, Yun Ken, Tao Huang · 2008
This study proposed a novel two-stage process of integrating support vector machine with expert screening technique to develop an automatic patent categorization system with high accuracy and high validity. The approach is tested on a real world case-the search history involving 264 patent documents of semiconductor equipment components. A 100% patent classification accuracy via the description portion of the patent documents was achieved using our proposed two-stage approach. The results showed that the proposed approach performed well in the real-world case of patent classification. The description field of the patent document was more than adequate for patent classification.