Chinese patent mining based on sememe statistics and key-phrase extraction

Abstract

Recently, key-phrase extraction from patent document has received considerable attention. However, the current statistical approaches of Chinese key-phrase extraction did not realize the semantic comprehension, thereby resulting in inaccurate and partial extraction. In this study, a Chinese patent mining approach based on sememe statistics and key-phrase extraction has been proposed to extract key-phrases from patent document. The key-phrase extraction algorithm is based on semantic knowledge structure of HowNet, and statistical approach is adopted to calculate the chosen value of the phrase in the patent document. With an experimental data set, the results showed that the proposed algorithm had improvements in recall from 62% to 73% and in precision from 72% to 81% compared with term frequency statistics algorithm. © Springer-Verlag Berlin Heidelberg 2007.