Aspect term extraction of E-commerce comments based on model ensemble

Review aspect terms extraction is an important task in the field of emotional analysis in natural language processing. This paper uses the method of ensemble learning to extract aspect terms of the E-commerce reviews. And this method has a very significant meaning. Because of the particularity of Ecommerce reviews, we choose the traditional method combined with machine learning method to extract the aspect terms of Ecommerce reviews emotional analysis, the experiment proved effective. First, we did word segmentation, POS tagging and other data preprocessing for the original Chinese E-commerce data. Then based on the training set construct the dictionary of aspect terms with the use of word-based reversed search method, for the test set tagging aspect terms. In addition, we train efficient CRF model and carry out aspect terms annotation on the comment data. Finally, based on the first two methods, the effect of model ensemble is improved.