Sentiment Sentence Construction Algorithm of Newly-Coined Words and Emoticons Dictionary for Social Data Opinion Analysis

Abstract

With the development of social networks, Social Network Services (SNS) has been populated. And people create newly-coined words and emoticons. Accordingly, the newly-coined word and emoticon of SNS show the social phenomenon of modern society. For social opinion analysis with SNS newly-coined words and emoticons, additional and continued sentiment proactive manual measures are required with a lot of time and money. It is necessary to make an objective decision when classifying the sentiment dictionary. This paper proposes a method for automatically constructing a newly-coined word and emoticon sentiment dictionary by extracting newly-coined words and emoticons from SNS reviews. In addition, the sentiment sentence is determined from the collected SNS reviews to automatically determine the polarity and intensity of newly-coined words and emoticons. Sentiment sentences contain newly-coined words and emoticons with strong polarity intensity, and mean sentences with strong polarity. The automatically constructed newly-coined word and emoticon sentiment dictionary proposed in this paper showed similar analysis accuracy compared to the existing manual construction method. As a result, this study can extract newly-coined words and emoticons in real time from trend-sensitive SNS and apply them to the sentiment dictionary to improve analysis accuracy.