Research On Sentiment Orientation Analysis Of Online-shopping Review Based-on HNC Theory

In recent years, the development of e-commerce is in its heyday. The online-shopping activity is increasingly increase. More and more scholars in the field of natural language processing turn their attention to the sentiment orientation analysis of online-shopping review. Online shopping reviews could not only help the seller to determine the key indicators improving goods and services, but also provide an important reference advice for customers about to buy the same products. The analysis of the massive comment is conducive to the management and development of e-commerce. Therefore, it has high research value.Chinese online-shopping is a special free short text. The past natural language analysis methods can’t effectively deal with such non-normative text. Combined with the true review, this paper proposes a new sentiment orientation analysis method based on HNC theory. First, this paper uses primitive symbol system of HNC theory to process the original review. In the stage of feature extraction, we use the sentence analysis of HNC theory and context unit extraction technology. The effective divisions of the typical sentence greatly improve the accuracy of the feature extraction. This paper build a characteristics granularity tree combined with many specification manuals and improve the tree with true review. And then complete the particle size correction, implicit reasoning, classifying of review object based on the characteristics granularity tree. This paper designs new appraise criteria take advantage of the HNC theory. Simultaneously, not only redesigning the processing strategy of extent word and negative word, but also considering the impact to emotional polarity come from special punctuation. Finally, the results of the general review analysis will be presented to the user in the form of a comprehensive review abstracts.In the experimental stage, this paper consider review from Taobao as the experimental data and select cellphone as the main research object. The comparative experiment results show that the proposed feature extraction method has some advantages in improving the accuracy. The new emotional tendencies criteria improve the processing speed on that basis of meeting the application requirements. The general review abstract is a microcosm of the review content. It not only helps PC users familiar with the product information as soon as possible, but also solve the browse problem of the mobile end-user.