A Quantum-like Approach to Semantic Text Classification

Area of Science:

• Natural Language Processing
• Computational Linguistics
• Quantum-Inspired Computing

Background:

• Traditional machine learning (ML) methods face challenges in text classification and analysis.
• Existing models often overlook complex semantic relationships within language.
• A need exists for alternative text representation models that capture nuanced linguistic structures.

Purpose of the Study:

• To explore a quantum-like (wave-based) model as an alternative to ML for sentiment analysis.
• To investigate the impact of semantic interference on text classification accuracy.
• To develop computationally efficient algorithms for wave-based text representation.

Main Methods:

• Sentiment analysis of English-language reviews using a quantum-like wave model.
• Exploration of text segmentation algorithms influenced by language structure.
• Comparison of quantum-like model results with classical probabilistic methods.
• Development of optimization techniques to reduce computational complexity.

Main Results:

• The quantum-like model improved classification accuracy by approximately 15% compared to classical methods.
• The model achieved precision and recall scores around 0.8 for classification tasks.
• A proposed optimization reduced the algorithm's computational complexity from O(n^2) to O(n).

Conclusions:

• The quantum-like wave model is a viable alternative or complement to traditional ML approaches for text analysis.
• Accounting for quantum-like semantic interference enhances classification accuracy.
• The developed model offers significant computational efficiency improvements.