Research on Automatically Extraction of Term Relation Extraction

摘要

Term relation extraction is the basis of automatically building the repository of terms. The current research about term relation extraction focuses on how to obtain effective characteristics of describing the relationship. This paper presents a novel method of feature extraction based on latent relation analysis, which automatically obtains the relation characteristics from patent corpus as the features of support vector machine. Experiments show that the F-score of this system yields 75%.