워드 임베딩과 단어 네트워크 분석을 활용한비지도학습 기반의 문서 다중 범주 가중치 산출 : 휴대폰 리뷰 사례를 중심으로

Abstract

Due to the increased amounts of online documents, there is a growing demand for text categorization thatcategorizes documents into predefined categories. Many approaches to this problem are based on supervisedmachine learning which couldn’t be applied to unlabeled data. However, large number of documents, such asonline cell phone reviews, have no category information and key categories are not predefined. To solve theseproblems, we propose unsupervised document multi-labeling method based on word embedding and wordnetwork analysis. After embedding words in a lower dimensional space using Word2Vec technique, we generatea weight matrix by calculating similarities between words. We create a word network using this matrix andextract the key categories from this network. With key category-weight matrix and co-occurrence matrix, wegenerate a document-category score matrix. To verify our proposed method, we collect 298,206 cell phonereviews from four review websites. Then, we compared the results of the proposed method with labeleddocuments from human cognitive perspective.