Patent Citation Recommendation for Examiners

There is a consensus that U.S. patent examiners, who are responsible for identifying prior art relevant to adjudication of patentability of patent applications, often lack the time, resources and/or experience necessary to conduct adequate prior art search. This study aims to build an automatic and effective system of patent citation recommendation for patent examiners. In addition to focusing on content and bibliographic information, our proposed system considers another important piece of information that is known by patent examiners, namely, applicant citations. We integrate applicant citations and bibliographic information of patents into a heterogeneous citation-bibliographic network. Based on this network, we explore metapaths based relationships between a query patent application and a candidate prior patent and classify them into two categories: (1) Bibliographic meta-paths; (2) Applicant Bibliographic metapaths. We propose a framework based on a two-phase ranking approach: the first phase involves selection of a candidate subset from the whole U.S. patent data; and the second phase uses supervised learning models to rank prior patents in the candidate subset. The results show that both bibliographic information and applicant citation information are very useful for examiner citation recommendation, and that our approach significantly outperforms a search engine.