A Combined Method for Chinese Micro-Blogging Topic Tracking
Xiang Zhang, Bo Z. Shang, Dong Li, Yu Jie Zhu · Applied Mechanics and Materials · 2014
To the problem of Chinese micro-blogging topic tracking, a method combined LDA model and Bagging of ensemble learning was proposed. The method firstly used the LDA hidden topic modeling, effectively solved the issue that the dataset’s sparsity of the short text, then made the C4.5 decision tree as a weak classifier, through examples resampling to obtain multiple training set, compounding the training sets according to the voting rule, and ultimately getting the similarity of the micro-blogging topic. Experiments show that, compared with the model based on single vector model, classical TF-IDF and the tracking method of C.45Bagging similarity computing, this method have a better performance on precision, recall ratio and F1 value.