Evaluation Method of Patent Scope Based on Semantic Information of Words and Dependency Structure of Patent Claims

Abstract

In patent documents, the "claim" is the most important section because the patent scope is interpreted based on the description of claims. However, because claims are expressed as sentences, the scope and boundaries of patents tend to be ambiguous. Prior studies have not been able to visualize the scope of a claim because they did not include the semantic range of words. Therefore, we develop a method to visualize the description of claims and the patent scope using natural language processing. Specifically, in order to visualize patent claims, we introduce the semantic range of words using the hierarchical structure of words provided by WordNet. Then, we apply the proposed method to two patents included in "PageRank" technology. The results show that parent patents are wider than child patents.