Attention-Based Multi-level Network for Text Matching with Feature Fusion
Yixiao Yang, Chongyang Zhang · 2021 4th International Conference on Algorithms, Computing and Artificial Intelligence · 2021
Text matching is a basic and common task in natural language processing. Recently, deep learning has achieved excellent performance in text matching tasks. The major process of the existing model is to pass two sentences through shallow encoder and interact with each other, then only the last layer of feature representation is utilized to conduct the final matching, which lacks sufficient semantic feature extraction and sentence interaction. To address the above limitation, we design an Attention-Based Multi-level Network(ABMN) for text matching, which utilizes the multi-level interaction layer with feature fusion to obtain more refined text information across all levels. We evaluate our proposed architecture on the three public real-world datasets:SNLI, Quora, and LCQMC. Experimental results show that the proposed model achieves the state-of-the-art performance.