Artificial Intelligence Breakthrough: Enhancing Metaphor Understanding via Cognitive Linguistic Model

Strategic Insights -

Researchers from the Zhongyuan University of Technology in Zhengzhou, People's Republic of China, have made significant strides in enhancing the ability of computers to understand and generate metaphors, a crucial aspect of natural language processing. The breakthrough comes from a novel approach that combines a Convolutional Neural Network (CNN) with a Support Vector Machine (SVM) to recognize metaphors. In experiments with English and Chinese texts, the integrated model achieved remarkable results, including an accuracy of 85% and an F1 score of 85.5% in English verb metaphor recognition tasks.

Key Takeaways:

  • The research proposed a metaphor recognition algorithm that combines a Convolutional Neural Network (CNN) with a Support Vector Machine (SVM) to achieve optimal metaphor recognition.
  • The model achieved an accuracy of 85% and an F1 score of 85.5% in English verb metaphor recognition tasks.
  • The integration of the SVM classifier significantly improved performance in Chinese metaphor recognition experiments, yielding an F1 score of 81.5%, an accuracy of 81%, and a recall of 82%.
  • The proposed model effectively integrates the CNN's powerful feature extraction capabilities with the SVM's superior classification performance.
  • The research concluded that the integrated approach enables more accurate identification of complex textual semantics, particularly in interpreting metaphorical language that requires deeper understanding.
  • Dongmei Zhu and her team demonstrated the effectiveness of the cognitive linguistic model in enhancing metaphor understanding via hierarchical classification and artificial intelligence SVM.
  • The study highlights the importance of integrating traditional machine learning techniques with deep learning methods to improve natural language processing capabilities.
  • The breakthrough has significant implications for applications in fields such as language translation, text summarization, and sentiment analysis.

Statistics:

  • 85% accuracy in English verb metaphor recognition tasks
  • 85.5% F1 score in English verb metaphor recognition tasks
  • 86% recall in English verb metaphor recognition tasks
  • 81.5% F1 score in Chinese metaphor recognition experiments
  • 81% accuracy in Chinese metaphor recognition experiments
  • 82% recall in Chinese metaphor recognition experiments

Sources:

  • Improvement of metaphor understanding via a cognitive linguistic model based on hierarchical classification and artificial intelligence SVM. Scientific Reports, 2025;15(1):18947.
  • Zhongyuan University of Technology Reports Findings in Artificial Intelligence (Improvement of metaphor understanding via a cognitive linguistic model based on hierarchical classification and artificial intelligence SVM). Journal of Engineering. June 9, 2025; p 5158.