Sentiment analysis requires a sentiment dictionary that maps words to sentiments. Further, sentiment weight is an important subtopic in the measurement of the strength of sentiments. A sentiment is the emotional response of an individual toward an external stimulus; therefore, the sentiment valence and sentiment weight vary among different persons. Hence, the definition and expression of a sentiment as a single state is a challenging task. In this study, we address the challenges in building a sentiment dictionary and analyzing sentiment weight. We construct a sentiment dictionary and propose a method to analyze word sentiments. We use the proposed method to analyze the general sentiments in social media. In our experiments, we used Flickr as the social media application and collected user responses to a sample post in order to utilize collective intelligence. We made four observations about this approach: (1) approximately 30 % of the words used in communication on social media signify a sentiment; (2) in addition to verbs and adjectives, nouns can be used for sentiment analysis; (3) 98.25 % of the seed words and words classified for sentiments matched; (4) the sentiment weight distribution was more concentrated for SO-NPMI than for SO-PMI.
Bibliographical noteFunding Information:
This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (2010-0022973).
© 2016, Springer Science+Business Media New York.
- Collective intelligence
- Sentiment analysis
- Social media
- Word polarity
ASJC Scopus subject areas
- Computer Science Applications
- Electrical and Electronic Engineering