TY - GEN
T1 - The effect of using Emoji and Hashtag in sentiment analysis on Twitter case study
T2 - 7th International Conference on Science and Technology, ICST 2021
AU - Amrullah, Muhammad Shofwan
AU - Budi, Indra
AU - Santoso, Aris Budi
AU - Putra, Prabu Kresna
N1 - Publisher Copyright:
© 2023 Author(s).
PY - 2023/2/14
Y1 - 2023/2/14
N2 - Twitter is one of the world's largest social media, reaching 330 million active users every month. Therefore, conducting sentiment analysis on Twitter should give a representative picture related to public perception. Twitter allows its users to use Hashtags and Emoji while posting a Tweet, and those become common on Twitter. Emojis and Hashtags can sometimes be decisive information for a Tweet's sentiment. Therefore, a study was conducted to find out how far Emoji's and Hashtag's influence if included as information when conducting sentiment analysis. The results showed that when doing sentiment analysis using the Naïve Bayes model, the one that includes Emoji got slightly better overall performance than other variants, with accuracy 77.55%, precision 0.819, recall 0.887, and F1-score 0.852. We conclude that in our case, Emoji and Hashtag do not significantly improve the performance but are worth considering in future research.
AB - Twitter is one of the world's largest social media, reaching 330 million active users every month. Therefore, conducting sentiment analysis on Twitter should give a representative picture related to public perception. Twitter allows its users to use Hashtags and Emoji while posting a Tweet, and those become common on Twitter. Emojis and Hashtags can sometimes be decisive information for a Tweet's sentiment. Therefore, a study was conducted to find out how far Emoji's and Hashtag's influence if included as information when conducting sentiment analysis. The results showed that when doing sentiment analysis using the Naïve Bayes model, the one that includes Emoji got slightly better overall performance than other variants, with accuracy 77.55%, precision 0.819, recall 0.887, and F1-score 0.852. We conclude that in our case, Emoji and Hashtag do not significantly improve the performance but are worth considering in future research.
UR - http://www.scopus.com/inward/record.url?scp=85149842043&partnerID=8YFLogxK
U2 - 10.1063/5.0118228
DO - 10.1063/5.0118228
M3 - Conference contribution
AN - SCOPUS:85149842043
T3 - AIP Conference Proceedings
BT - Proceedings of the 7th International Conference on Science and Technology
A2 - Yokozeki, Tomohiro
A2 - Santos, Gil Nonato C.
A2 - Omar, Rohayu Che
A2 - Widodo, null
A2 - Tristan, Abraham Cardenas
A2 - Putri, Ratih Fitria
A2 - Mustika, I. Wayan
PB - American Institute of Physics Inc.
Y2 - 7 September 2021 through 8 September 2021
ER -