TY - GEN
T1 - Sentiment Analysis of Lithuanian Texts Using Deep Learning Methods
AU - Kapočiūtė-Dzikienė, Jurgita
AU - Damaševičius, Robertas
AU - Woźniak, Marcin
N1 - Publisher Copyright:
© 2018, Springer Nature Switzerland AG.
PY - 2018
Y1 - 2018
N2 - We describe experiments in sentiment analysis of the Lithuanian texts using the deep learning methods: Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN). Methods used with pre-trained Lithuanian neural word embeddings are tested with different pre-processing techniques: emoticons restoration, stop words removal, diacritics restoration/elimination. Despite the selected pre-processing technique, CNN was always outperformed by LSTM. Better results (reaching an accuracy of 0.612) were achieved with the undiacritized texts and undiacritized word embeddings. However, these results are still worse if compared to the ones obtained using Support Vector Machines or Naive Bayes Multinomial and with the frequencies of words as features.
AB - We describe experiments in sentiment analysis of the Lithuanian texts using the deep learning methods: Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN). Methods used with pre-trained Lithuanian neural word embeddings are tested with different pre-processing techniques: emoticons restoration, stop words removal, diacritics restoration/elimination. Despite the selected pre-processing technique, CNN was always outperformed by LSTM. Better results (reaching an accuracy of 0.612) were achieved with the undiacritized texts and undiacritized word embeddings. However, these results are still worse if compared to the ones obtained using Support Vector Machines or Naive Bayes Multinomial and with the frequencies of words as features.
KW - LSTM and CNN methods
KW - Neural word embeddings
KW - Positive/negative/neutral sentiments
KW - The Lithuanian language
UR - https://www.scopus.com/pages/publications/85053597165
U2 - 10.1007/978-3-319-99972-2_43
DO - 10.1007/978-3-319-99972-2_43
M3 - Conference contribution
AN - SCOPUS:85053597165
SN - 9783319999715
T3 - Communications in Computer and Information Science
SP - 521
EP - 532
BT - Information and Software Technologies - 24th International Conference, ICIST 2018, Proceedings
A2 - Damasevicius, Robertas
A2 - Vasiljeviene, Giedre
PB - Springer Verlag
T2 - 24th International Conference on Information and Software Technologies, ICIST 2018
Y2 - 4 October 2018 through 6 October 2018
ER -