In view of ignoring semantic relationship between words, high dimensionality of data and computational complexity when current text clustering algorithms deal with Chinese texts. This paper presents a new method to cluster Chinese texts based on semantics in a specific field-TCBS (Text Clustering Based on Semantics) algorithm. The algorithm is based on the agglomerative hierarchical clustering algorithm, it expresses Chinese texts with the characteristic words and sets relative threshold in order to improve the efficiency of clustering. Compared with the traditional algorithms, the experimental results show that TCBS has effectively enhanced the quality of the clustering.