A Method of Calculating Comment Text Similarity Based on Tree Structure

摘要

Text similarity measure has a significance role in promoting the development of information processing. To address the comment text, this paper proposes a method, which is based on tree structure, of measuring text contents similarity. Taking advantage of comment text's content organization features to transform full text into a tree structure, this method divides the similarity measure of comment texts into that of the corresponding parts between the layers of trees. Accordingly the objects of similarity measure in each layer are the same type of words. Then suitable methods of similarity measure are adopted respectively, and different weights are given to the similarities in different layers. Finally, the overall similarity is achieved by combining the similarities in all the different tree layers. The experimental results on Amazon datasets show that the proposed method is more effective and has a higher accuracy than other common measuring methods.