Research On Opinion Target Extraction Based On Syntactic Structure

With the development of the contemporary Internet, processing and getting network information by hand have not meet people’s needs, so it have widely used Natural Language Processing technology to process complicated information. Sentiment analysis is a key research topic in Natural Language Processing field, the main contents focus on processing subjective information, such as reviews, blogs and so on. Opinion target extraction is an important part of sentiment analysis which extracts the entities and attributes of each sentence in subjective texts and servicing other tasks for sentiment analysis.The main work of this paper is focused on the research of opinion target extraction based on syntactic structure which combines traditional methods based on rules or templates with the methods based on statistics model in order to explore more effective features for opinion target extraction. First, the features are extracted from preprocessed results. Then, the results of feature extraction with the corresponding template will be input to the Conditional Random Field module for training and recognition. In the process of feature extraction, three features are applied on the basis of existing features by analyzing syntactic structure for mining syntactic structural information in subjective texts. In the definition of templates, to select template with better performance and improve the effect of opinion target extraction, many comparison experiments have carried on a variety of templates with different sizes.We have carried on comparative experiments in four datasets, observing the performance changes in the same template with different features and the same features with different templates. By comparison with existed features which are the baseline, the performance of opinion target extraction in this paper outperforms the contrast method. F-measure which is the comprehensive evaluation indicator in optimal results improves 2.3 percent.