Computer Engineering and Applications ›› 2023, Vol. 59 ›› Issue (18): 98-104.DOI: 10.3778/j.issn.1002-8331.2210-0497

• Pattern Recognition and Artificial Intelligence • Previous Articles     Next Articles

Research on Personality Identification Method by Fusing BERT and Syntactic Dependency

ZHANG Zhonglin, YUAN Chenyu, CHEN Liping, WU Yilin   

  1. 1.School of Electronic and Information Engineering, Lanzhou Jiaotong University, Lanzhou 730070, China
    2.State Key Laboratory of Multimodal Artificial Intelligence System, Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China
    3.School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing 100049, China
  • Online:2023-09-15 Published:2023-09-15



  1. 1.兰州交通大学 电子与信息工程学院,兰州 730070
    2.中国科学院 自动化研究所 多模态人工智能系统全国重点实验室,北京 100190
    3.中国科学院大学 人工智能学院,北京 100049

Abstract: To solve the technical challenge that how to effectively fuse the deep semantics of the text and the psychological cues, a personality identification model fusing BERT and syntactic dependency is proposed. Firstly, it uses BERT to extract the deep semantic information contained in the text, and obtains psychological cues with personality characteristics through syntax analysis, secondly, it designs a conditional fusion function to dynamically embed the mined knowledge as an external condition into the text representation vector, capturing the dynamic semantic interaction between these informations, finally, it uses the fully connected network to perform deep feature extraction and dimensionality reduction, so as to identify the personality. Experiments on the firstly constructed dataset for Chinese movie reviews have verified the efficacy of the proposed method. Compared with traditional neural networks and single BERT models, the model obtains a significant improvement in personality identification.

Key words: BERT model, dynamic semantic interaction, personality detection

摘要: 针对现有性格识别方法难以有效融合文本深层语义与心理学特征的技术挑战,提出了融合BERT与句法依存的性格识别模型。采用BERT提取文本蕴含的深层语义信息,通过词法与句法分析获得具有性格特征的心理学词汇,设计条件融合函数将该词汇作为外部条件动态嵌入到文本表示向量中,捕获文本深层语义与性格线索间动态的语义交互,基于融合后的特征向量使用全连接网络进行更深层的特征提取并降维,以此对性格进行识别。在首次构建的面向中文电影评论的性格数据集上的实验验证了该方法的有效性,该模型相较传统神经网络和单一BERT模型在性格识别准确性上有明显提升。

关键词: BERT模型, 动态语义交互, 性格识别