Multi-class Sentiment Analysis on Twitter
| Main Authors: | Venkatesh, Nagaraju Y., Sheema Sultana, Mamthaj A., Priyadarshini R., Kavya S. |
|---|---|
| Format: | Article |
| Bahasa: | eng |
| Terbitan: |
, 2020
|
| Subjects: | |
| Online Access: |
https://zenodo.org/record/4130218 |
Daftar Isi:
- Abstract—Twitter, Online Social Networks (OSN) moved toward becoming truly the stage for individuals express their contemplations, share their thoughts, opinion on real-time events and up coming occasions. Sentiment analysis is procedure of gathering, conglomeration, and classification of data gathered in social networks into a different emotion classes. Sentiment analysis and opinion mining in social networks are current research topics. The state of the art works focused on the binary and ternary classifications. These binary and ternary classifications hindrance the actual sentiments present in the tweets of user. However, existence of multiple meanings that might have different sentiment polarity for the same word, the binary classification fails to identify meaning and the polarity of slang words. To analyse opinion of user, it is necessary to go deeper in the classification and detect the sentiment hidden behind user post. Multi-class sentiment classification has always been a challenging task due to complexity of natural languages and the difficulty of understanding and mathematically quantifying how humans express their feelings. In this paper, we proposed model that performs the task of multi-class classification of online posts of Twitter users. It demonstrates how far it is possible to go with the classification, and the limitations of handling the slang words, hashtags used in the tweets. Our implementation demonstrates that proposed approach achieves an accuracy about 45.19% on the multi-class classification for the dataset (text emotion). For dataset (Consumer complaint) the proposed approach achieves accuracy of 82.07%. The approach achieve better accuracy due to the number of features extracted from the individual datasets.