Abstract
In this paper, we propose a method resolving polysemy of terminology to select appropriate translation. Terminology polysemy in academic texts must be resolved for retrieving in or translating to other languages, since the polysemy may depend on the research domains, though it is not an easy task for non-experts of the domain. We focused on that the Japanese and English abstracts included in a paper is a cross-lingual corpus, and built a model of LDA latent topic space from a large set of the corpora, which can be used to resolve terminology polysemy.
