Enhancing Metaphor Recognition of Literary Works in Applied Artificial Intelligence: A Multi-Level Approach with Bi-LSTM and CNN Fusion

Understanding metaphorical language is essential for AI to interpret and communicate with humans accurately. However, current methods often struggle with the complexity of metaphors, making it difficult for AI systems to understand human language fully. Recognizing metaphors is challenging because they are frequently ambiguous and depend on context. In this study, we propose a new approach using a combination of Bi-directional Long Short-Term Memory (Bi-LSTM) networks, Convolutional Neural Networks (CNN), and uni-directional LSTM components to create a multi-level model for recognizing metaphors. Our model uses various features, including dependency, semantics, and part-of-speech, to improve its learning ability. Additionally, we introduce a new method for recognizing the emotional context of metaphors using a random walk model to determine the emotional tone of words. Our results show that this model improves performance in recognizing metaphors, enhancing AI's ability to understand them.