A Way to Detect the Opinion Sentences from Short Texts by the Vote-AdaBoost Combining Classify Method

Abstract

Because of the sparse sentiment features of network short texts, the identify result of common classification methods which are efficacious in texts with tradition structures are unsatisfactory. This article tries to combine multiple classifiers combination and integrated learning methods to solve this problem. Vote-AdaBoost combining classify method is constructed to optimize the appropriate classifier as the voting combination in iterative learning process. Finally, an effective classification method to detect the opinion sentences of short texts is obtained, and the effectiveness of this method is verified by experiments.