Comprehensive analysis and classification of natural language questions based on Bi-LSTM-CRF
Jiajun Li, Huazhu Song, Jun Li, Kaituo Mi · 2022 IEEE International Conference on Advances in Electrical Engineering and Computer Applications (AEECA) · 2022
This paper aims to study the reasoning and understanding of natural language questions. First, for the knowledge of the learning process, we convert the reasoning and understanding of questions into named entity extraction and relation classification of questions. Thus, a named entity recognition model based on Bi-LSTM-CRF is proposed. Due to the specificity of natural language questions and in view of the fact that contextual information plays a decisive role in interrogative relation, we propose a relation classification model based on Bi-LSTM+Attention mechanism to capture the contextual information. Then, using the already established knowledge graph, the corresponding comparative experiments are designed. The F1 value of the named entity recognition experiment based on Bi-LSTM-CRF could reach 84.63%, and the F1-score of the relationship classification experiment based on Bi-LSTM+Attention mechanism could reach 71.94%. Both of them are better than the traditional model.