Open Access

ARTICLE

Mathematical Framework for Detecting Sentiments Analysis Using Quantum Computing

Anitya Kumar Gupta^*, Pankaj Vaidya, Anurag Rana

Yogananda School of AI Computers and Data Science, Solan, India

* Corresponding Author: Anitya Kumar Gupta. Email:

Journal on Artificial Intelligence 2026, 8, 377-402. https://doi.org/10.32604/jai.2026.074171

Received 04 October 2025; Accepted 28 January 2026; Issue published 22 July 2026

Abstract

Sentiment analysis aims at determining the stance or point of view of a topic, author, or speaker about a certain topic, document, or event. The given paper proposes a hybrid quantum-classical model of sentiment classification, which is known as Complex-Valued Quantum-Enhanced Recurrent Neural Network (CQRNN). The model combines Quantum Long Short-Term Memory (QLSTM) and Quantum Gated Recurrent Units (QGRU) with the use of Variational Quantum Circuits (VQCs) and complex-valued embeddings designed at the level of semantic information, both in the real and imaginary domains. Experiments on benchmark sentiment datasets show that CQRNN can be fine-tuned and reach an accuracy of 85% on noisy and ambiguous data, and this is a demonstration of its strength and better ability to discriminate between similar sentiment expressions like neutral and negative expressions. The performance of quantum-inspired architectures in sentiment classification compared with the traditional models, including SVM and Random Forest, establishes the potential. The article prepares the basis of benchmarking CQRNN to the transformer-based models (BERT, RoBERTa) and implementing it on actual quantum hardware.

---

Keywords

---