利用SAO方法進行專利前案檢索

Abstract

專利數量從1997年開始迅速上升。因此，專利辦公室，如：美國專利局需要雇用更多的審查員、人力來處理這些大幅上升的專利申請案。而與此同時，也代表需要發展一個更有效率的專利前案檢索技術。而過往的研究，大多都是著重在詞袋模型，然而詞袋模型難以展現出專利中的一些特徵、關鍵概念、詞與詞之間的關係，因此，我們提出一個基於SAO結構的專利前案檢索技術，最後結果也顯示，我們提出的方法，比過去研究提出的技術，確實有更好的表現。

The number of patent applications increases rapidly from 1997. At the same time, the number of patent applications increases rapidly that makes patent offices, such as United States Patent and Trademark Office (USPTO), have to recruit more patent examiners for dealing with the ever-increasing number of patent applications. Hence, an effective patent prior art retrieval from a huge patent repository becomes essential. In this research, we review several existing patent prior art retrieval approaches. These approaches typically involve extracting textual content by information retrieval technique called bag-of-words. However, bag-of-words cannot reflect the structural relationships among components. In order to solve this problem, we propose an SAO-based approach to conduct patent prior art retrieval. SAO structures can represent explicit relations among components used in a patent, and are considered to represent key concepts of the patent. Overall, the proposed SAO-based approach outperforms our performance benchmarks.