Title: Sentiment classification using hierarchical attention convolutional network with Twitter data

Authors: Yedlla Satyam; Kannan Rajchandar

Addresses: School of Computer Science and Artificial Intelligence, SR University, Ananthasagar, Hasanparthy, Hanumakonda 506371, India ' School of Computer Science and Artificial Intelligence, SR University, Ananthasagar, Hasanparthy, Hanumakonda 506371, India

Abstract: This paper proposes a hierarchical attention convolutional network (HAConv-Net) for sentiment analysis of Twitter data. Firstly, Twitter data is acquired from a specified dataset, which is passed through the tokenisation stage, where bidirectional encoder representation from transformer (BERT) is exploited to convert the input text to tokens. Afterwards, feature extraction is processed, where the features, including hashtag, all-caps, elongated units, emoticon, TF-IDF, and N-gram features are excerpted, along with word2Vec vector encoding. Lastly, sentiment analysis of Twitter data is performed using HAConv-Net, which is newly devised by incorporating hierarchical attention networks (HAN) and Siamese convolutional neural network (SCNN). Further, the examination of HAConv-Net is accomplished concerning the recall, f-measure, and precision and is found to record higher values of 92.885%, 92.702%, and 92.520%, respectively.

Keywords: natural language processing; NLP; sentiment classification; deep learning; DL; Twitter; hierarchical attention convolutional network; HAConv-Net.

DOI: 10.1504/IJWET.2026.155090

International Journal of Web Engineering and Technology, 2026 Vol.21 No.2, pp.214 - 244

Received: 18 Oct 2024
Accepted: 29 Sep 2025
Published online: 27 Jul 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article