On the use of Intelligent Information Retrieval in Patent Prior-Art Search

Patents play an important role in Intellectual Property protection. So, in recent years a considerable attention has been paid to patent and prior-art search. In process of patent application filing, searching in the previous patent database is of great importance. Patent examiners search in a huge database of patents to find if there exists any similarity between applicant’s claim and the previous registered patents. This process that called patent invalidity run, is one of the important stages of patent registration. Because of legal aspects of this process, the searcher should not leave any relevant patent document. So patent searching is essentially a recall-oriented issue in information retrieval applications. In recent years, the use of intelligent information retrieval in this search process has been investigated by many researchers. In this paper we investigate various methods of information retrieval that have been proven to be effective in retrieving relevant results. The survey also has focused on query formulation and how to transform a query patent into a search query. So we have explored different factors of a successful transformation, such as how many query words should be used, where to extract query words, how to weight them and whether to use noun-phrases instead of individual words. Furthermore, the survey covers researches that combine different features and has been proven to make a significant improvement in retrieval performance.