Author:
Luo Shihan
Affiliation:
KyuShu University, Japan
Keyword(s):
Network popular words, Word2vec, Sentiment analysis, Sentiment polarity
Abstract:
In today’s SNS, numerous network popular words have been creating. Due to most network popular words are only used in a short period of time, it is hardly to find a dictionary that includes network popular words. So, it is unpractical to translate network popular words by traditional machine translation method. Now, the research of extracting network popular words has obtained some outcomes to a certain extent. However, the existing research of paraphrasing network popular words, which uses distributed representation method to analyse network popular words, does not make much progress. The purpose of this paper is to improve the translation of network popular words by adding sentiment analysis into distributed representation method. By analysing network popular words in eight sentiment polarity (joy, sadness, trust, disgust, fear, anger, surprise, anticipation), the accuracy of translation of network popular words is obviously improved.