College of Information Science and Engineering, Lanzhou University, Lanzhou, China.

---

Abstract—Based on the unique structure of Chinese text data in the field of building materials, this paper proposes a Chinese text similarity calculation method based on part-of-speech tagging and word vector model. By analyzing the special structure of Chines text data in the building materials field, the method realizes specific part-of-speech tagging by using the machine, which saves a lot of manpower and time consumption required for manual labeling. Then, combined with the word vector model, the text similarity calculation is realized by these steps: Chinese word segmentation, syntactic analysis, and similar matching of the annotated text. This paper comprehensively compares the word level similarity calculation method based on the vector space model. Through analysis and experimental comparison, the algorithm in this paper obtains an average F value of 72.0%, which is improved by 20.67% compared with the method based on vector space model, and has achieved better test results.

Index Terms—Chinese text similarity, word vector model, part-of-speech tagging, vector space model.

[PDF]

Cite: Zhixin Ma, Mengguang Li, "Chinese Text Similarity Algorithm Based on Part-of-Speech Tagging and Word Vector Model," Journal of Computers vol. 14, no. 4, pp. 311-317, 2019.