DETECTING SARCASM BASED ON QUANTUM VECTORIZATION

Abstract

This article proposes a neural network approach for sarcasm detection based on the formation of quantum bag of words features from text. In the proposed method, words are first converted into quantum states using the Grover algorithm. Then, quantum bag of words features are generated and used for classification. The obtained results were compared with those of classical approaches. According to experimental results, the quantum computing-based approach achieved 80% accuracy, while the classical computing-based approach showed 78% accuracy. Additionally, the quantum bag of words method produced results 2% faster than the classical approach. This demonstrates that the quantum approach, through the properties of superposition and entanglement, achieves higher performance. The limitations include the restricted vocabulary size, insufficient datasets in many cases, and the lack of access to real quantum technologies. Nevertheless, the findings show that computers capable of simulating quantum behavior indicate the promising potential of quantum computing in natural language processing.