Chinese Short Text Classification Based on Dependency Syntax Information
Yinggang Zhang, Hongguang Xu, Ke Xu · 2021
Short text has the characteristics of sparse features and discrete semantics. In order to extract the features of short text better, we propose a short text classification algorithm based on dependency syntax information in this paper. In terms of text representation, we train word vector based on sentences dependency triples. By concatenating the dependency word vector and original word vector, text can be represented at both semantic and syntactic levels. In terms of classification model, we use the dependency syntax information of the short text to guide the state update process of the recurrent neural network. In addition, we run experiments based on Chinese news-title dataset. Experiment results show that the proposed algorithm improves the performance of short text classification remarkably.