An interactive approach to term relation extraction and term extraction

In the current research, the study of term extraction is usually separated from the study of term relation extraction. Seldom is the information of relations between two terms used in the term extraction. As for the term relation extraction, the normal extraction process is, firstly to annotate the terms in corpus, then to use various approaches to find the correlation between those terms. The problem existing in this method is, relation extraction precision is usually greatly influenced by the term extraction precision because the recognition mistake occurring in the term extraction will extend to the term relation extraction. This paper introduces an interactive approach integrating the two tasks. In the experiment, the approach is applied to two different fields. The experimental results show that this approach not only improves the performance of the relation extraction, but also enhances the term extraction. Copyright © 2010 Binary Information Press.