숫자로 표상된 의미: 딥러닝 시대의 의미론

Abstract

The advent of some practical Artificial Intelligence (AI) applications and the wide availability of Deep Learning algorithms seem to have shaken most aspects of everyday life. In particular the arrival of the vector space modelling based on word embedding, and the availability of the tools like Word2vec signalled the era of high quality word vectors and literally have changed the world of Natural Language Processing. In this paper we discuss the nature of the vector space model as an alternative in linguistic semantics. We also discuss some of its characteristics and limitations, and some possible related linguistic issues based on results gained from applying Word2vec to two of the well known Korean corpora, the Sejong Semantically Annotated Corpus and part of the Trend21 corpora.