Research On Methods Of Customer Opinion Mining Based On Online Reviews

Intenet has become one important part of our daily life,it is inseparable with our work and learning. With the rapid popularization development of Internet,information on internet become more and more to enormous. Since Internet entered Web2.0 ages, all kinds of information resource is rich and colorful. Since time came to the new century, all kinds of new interactive websites gradually arised. Some familiar things such as blog, Wiki and instant messaging (IM) and so on have entered in our life. More internet users are involved, and they can published views and opinions through the network medias.This paper discusses related theories and methods of opinion mining based on online reviews detailedly, designs the process of the realization of opinion mining. This paper analyzes customer behaviors and opinion tendency through the realization of methods and the process of opinion tendency classification. This paper detailedly studies methods and the process of feature and polarity extraction, and analyzes the rules of customer opinions expressed.This paper summarizes and analyzes customers'semantic tendency of opinions of online users combined with the characteristics of online review contents. In the study we collected lots of customer comments from the internet, and these datas are structured after treatment. we also segmented and marked the contents for further analysis. After we extracted features from preprocessed datas, the datas are represented the form of characteristic vector. We use the method of computer automatic classification for the opinion tendency classification of preprocessed datas based on different attributes. In the process of classification this paper used support vector machine (SVM) method and using Weka software for classification based on every attribute which contains stars. In my study the process and the way of classification are improved for better result. Through the analysis of classification results we found the hidden rules of customer opinion expressions in comments, and These knowledge can help sellers learn customers pertinently.Online reviews often contain customers'evaluations of all kinds of goods and attributes, which are also valuable for us. This paper studies mining methods of the objects and their features evaluated and designs the mining process. In my study I design the process of objects and their features evaluated through the use of the method of frequency statistics and calculation of mutual information value. After that I design the process of polarity extraction combined with means of semantic rules and hownet semantic analysis. This paper extracts child features of features which contain stars and show what we find. This paper also use the statistical theorise and methods to analyze the impacts of customer word-of-mouth in business. customers'online reviews are spreaded in the form of word-of-mouth, this online word-of-mouth impact deeply trading behaviours of buyers and sellers for no doubt. In the study, the analysis of how customers'opinion tendency impact rate of second glance of shops can help us learn customers better. This paper attempts to study the opinion mining to provide references for the related field, help sellers understand customers and make customer relationship management better, and provide better supports for customers'decision-making.