基于通用词与术语部件的专利术语抽取

Patent Term Extraction Based on Generic Words and Term Components

针对目前专利术语抽取中不能有效地过滤一些高频非术语词串和无法正确抽取低频术语的问题,本文提出基于通用词与术语部件的专利术语抽取方法。该方法首先使用通用词作为切分符选取候选术语;再利用与候选术语有相同术语部件的相似候选术语信息,评估候选术语成为术语的可能性。实验结果表明,与传统的方法相比,提出的方法能够有效地提高专利术语抽取的准确度。

Aiming at the problems that some high-frequency non-term strings cannot be effectively filtered and that low-frequency terms cannot be correctly extracted in patent term extraction, this paper proposes a patent term extraction method based on generic words and term components. The proposed method first takes advantage of generic words to select candidate terms. Then, candidate terms with the same term component as the target candidate term are used to evaluate the target candidate term. Experimental results show that the proposed method can effectively improve the accuracy of patent term extraction, when compared with the traditional methods.