关键词: 深度文本聚类, 分布增强, 变分自编码器, 语义表征, 分布一致性约束

Abstract: Abstract: To address the issues of missing distribution information and distribution collapse encountered by deep variational text clustering models in practical applications, this study proposes a Deep Variational Text Clustering Model Based on Distribution Augmentation (DAVAE). This model can integrate enhanced latent semantic distributions into the original latent semantic distribution by enhancing distribution information to improve the completeness and accuracy of the latent distribution. Additionally, the model employs a distribution consistency constraint strategy to promote the learning of consistent semantic representations, enhancing the model’s ability to express the true information of the data through semantic distributions and thus improving clustering performance. Compared with existing deep variational inference models and deep clustering models, DAVAE's Normalized Mutual Information (NMI) metric is improved by 0.16, 9.01, 2.30, and 2.72 percentage points on the four real-world datasets of Abstract, BBC, Reuters-10k, and BBCSports, respectively. This validates the effectiveness of the model.

Key words: deep text clustering, distribution augmentation, variational autoencoder, semantic representation, distribution consistency constraint