Research on author attribution based on core topic

Subjects: Library Science，Information Science >> Information Retrieval submitted time 2023-02-09

MENG Xu XIE Jing LI Chunwang

Abstract：

[Background and purpose] Author recognition is developing towards the use of multilevel features. Compared with stylistic features, thematic features are still a few in the research and application of author recognition, especially for Chinese social media texts. At the same time, the research on the use of topic features focuses more on the innovation of the extraction technology and methods of topic features, but not on the identified topics and the application methods of topic features. Therefore, the basic purpose of this study is to study the use of topic features in the author recognition of Chinese social media texts, and further develop strategies to identify and screen the core topics in the topic features, optimize the use of topic features, so as to improve the use effect of topic features in the author recognition. [Methods] The research first uses the LDA topic model to extract the academic topics and social topics of the candidate authors, and then uses Word2vec to develop a merge screening strategy to identify and represent the core topics, and finally uses N-gram features and similarity calculation to achieve author recognition. [Results] The experimental results showed that the thematic features had a certain positive effect on the author's recognition in the corpus of this study, and the strategies and applications related to the core thematic features proposed in this study could also optimize the use of thematic features.

Peer Review Status:Awaiting Review

 Hits 2322  Downloads 334  Comment 0