Coreference Resolution Method Integrating Textual Information and Semantic Assessment
Da Teng, Xiaochuan Zhang, Xinlai Xing, Panpan Chen, Changmeng Yang, Dafei Lin · 2023
The coreference resolution refers to the task of identifying and clustering different expressions that refer to the same entity. In natural language processing, it is a critical task that provides effective support for other NLP tasks. To address issues such as unclear referential information, weak feature vector correlations due to long distances between pronouns, this paper proposes an end-to-end coreference resolution model that integrates Bi-LSTM and multi-head self-attention mechanisms, combined with text similarity calculation to improve the coreference scoring function. The model was tested on the OntoNotes 5.0 dataset and compared with the c2f-coref+spanBERT model. The results show an Avg.F1 score improvement of 0.3% under the CONLL evaluation standard, with an Avg.F1 score of 79.7%, demonstrating the effectiveness of the proposed algorithm.