TY - GEN
T1 - Polarity Classification of Tweets Considering the Poster’s Emotional Change by a Combination of Naive Bayes and LSTM
AU - Tago, Kiichi
AU - Takagi, Kosuke
AU - Jin, Qun
PY - 2019/1/1
Y1 - 2019/1/1
N2 - Twitter, as a popular social networking service, is used all over the world, with which users post tweets for various purposes. When users post tweets, an emotion may be behind the messages. As the emotion changes over time, we should better consider their emotional changes and states when analyzing the tweets. In this study, we improve polarity classification by considering the poster’s emotional state. Firstly, we analyze the sentence structure of a tweet and calculate emotion scores for each category by Naive Bayes. Then, the poster’s emotion state is estimated by the emotion scores, and a prediction model of emotional state is created by Long Short Term Memory (LSTM). Based on the predicted emotional state, weights are added to the scores. Finally, polarity classification is performed based on the weighted emotion scores for each category. In our experiments, our approach showed better accuracy than other related studies.
AB - Twitter, as a popular social networking service, is used all over the world, with which users post tweets for various purposes. When users post tweets, an emotion may be behind the messages. As the emotion changes over time, we should better consider their emotional changes and states when analyzing the tweets. In this study, we improve polarity classification by considering the poster’s emotional state. Firstly, we analyze the sentence structure of a tweet and calculate emotion scores for each category by Naive Bayes. Then, the poster’s emotion state is estimated by the emotion scores, and a prediction model of emotional state is created by Long Short Term Memory (LSTM). Based on the predicted emotional state, weights are added to the scores. Finally, polarity classification is performed based on the weighted emotion scores for each category. In our experiments, our approach showed better accuracy than other related studies.
KW - Deep learning
KW - Naive Bayes
KW - Polarity classification
KW - Twitter
UR - http://www.scopus.com/inward/record.url?scp=85069158016&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85069158016&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-24289-3_43
DO - 10.1007/978-3-030-24289-3_43
M3 - Conference contribution
AN - SCOPUS:85069158016
SN - 9783030242886
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 579
EP - 588
BT - Computational Science and Its Applications – ICCSA 2019 - 19th International Conference, 2019, Proceedings
A2 - Misra, Sanjay
A2 - Stankova, Elena
A2 - Korkhov, Vladimir
A2 - Torre, Carmelo
A2 - Tarantino, Eufemia
A2 - Rocha, Ana Maria A.C.
A2 - Taniar, David
A2 - Gervasi, Osvaldo
A2 - Apduhan, Bernady O.
A2 - Murgante, Beniamino
PB - Springer-Verlag
T2 - 19th International Conference on Computational Science and Its Applications, ICCSA 2019
Y2 - 1 July 2019 through 4 July 2019
ER -