제품관점의 특허지능 시스템을 위한 특허의 대상제품 인지 방법

A patent has its intended target products, and thus, identifying target products of patents could become a basis for product-level pa-tent intelligence systems that analyze firm’s competititive intelligence and developmental stages for multiple products. As part of de-veloping technology opportunity discovery system of Korea Institute of Science and Technology Information, this study proposes models for patent-product connection recognition. Based on the occurrences of product names in patent text, this study identifies prod-uct-IPC distribution vectors and patent-IPC distribution vectors. Then, we define patent-product connection recognition models, using similarities between pairs of patent and product IPC distributions and applying some recognition rules. As a result of performance tests, our proposed models showed a dramatic improvement in recognizing patents’ target products, compared to an existing recogni-tion model. We expect that our approach, as the initial work to computationally recognize patents’ target products, will be a corner-stone for other later product-level patent intelligence systems and studies.