Abstract—Deep learning has made remarkable advances in
many application domains, demonstrating its robustness in
various data mining tasks. However, it is often overlooked that
external information sources can explicitly be included in deep
learning models to expand their feature space for improved
performance, especially when large datasets are not available.
This paper presents a neural network architecture for
multi-class sentiment analysis, incorporating semantic
information from a sentiment lexicon. The model was
evaluated on a small dataset of Japanese hotel reviews, with
results indicating that integrating sentiment polarities into
neural networks can increase classification accuracy.
Index Terms—Sentiment analysis, sentiment lexicon, deep learning, text mining, customer reviews.
Mate Kovacs is with the Graduate School of Information Science and Engineering, Ritsumeikan University, 525-8577 Kusatsu, Nojihigashi 1-1-1, Japan (e-mail: firstname.lastname@example.org).
Victor V. Kryssanov is with the Collage of Information Science and Engineering, Ritsumeikan University, 525-8577 Kusatsu, Nojihigashi 1-1-1, Japan (e-mail: email@example.com).
Cite: Mate Kovacs and Victor V. Kryssanov, "Expanding the Feature Space of Deep Neural Networks for Sentiment Classification," International Journal of Machine Learning and Computing vol. 10, no. 2, pp. 271-276, 2020.Copyright © 2020 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).