A METHOD OF POSITIONING UNKNOWN WORDS IN AN EXISTING THESAURUS BASED ON AN ASSOCIATION MECHANISM

摘要

Our research aims to develop a robot which can make smooth conversation with human beings. Such robot needs to have abilities to understand and interpret words. Currently, a technique based on a large-scale language dictionary or a corpus is predominantly mostly used in the field of the language processing. As quite a lot of costs, resources and time are necessary to create such linguistic capital, automatic construction technique is also being researched. However, the knowledge in a category of common sense is inherent to human and difficult to construct automatically although it is indispensable knowledge for robot to realize conversation with human beings without sense of unease. In this paper, we propose a technique which contributes to semiautomatic construction of a large-scale language dictionary and a corpus. Concretely, a system using the proposed technique indicates a position of an unknown word to be registered in an existing thesaurus dictionary. The proposed technique was able to improve approximately 20% in accuracy compared with the traditional techniques.