The Dynamic Analysis Method Of Online Reviews Based On Opinion Mining

Nowadays, the rapid spread of information technology represented by Internet or wireless mobile Internet generates the network aggregation that comes from the traditional business data. Network business transaction data can be used by data mining methods and finally serve the business decision-making process and strategic development. At present, data mining has extensive application on customer relationship management (CRM) and product design, which are many of the successful applications in business field. As the new marketing mode that combines the data mining technology with the information management, data mining and business intelligence (BI) can get the mass transaction data from the Internet, use many algorithm analysis such as Apriori, machine learning or statistical analysis to recognize the change of the customers’ opinion of the certain products based on analytical results, and eventually enhance the capacity of the certain products to respond the market in time.The important needs of the developing E-business and research priorities on business information management come from the transformation from massive network trading data to knowledge that can be identified and used directly, the using and improving the text mining process of the business data, and the designing and analyzing the data mining system based on products reviews.At the same time, the Internet are becoming the critical medium that people can release and share their opinions and ideas, while the opinion mining technologies are also extensively researched and applied to those information resources. As the development of the Internet and mobile Internet, network-trading reviews are increasing fiercely. So facing the massive, complex, updating and dynamic network information resources, traditional text mining technology may not meet the demands that deal with the high dimensional, dynamic, large and updating Internet content.Therefore, this paper comes up with the dynamic opinion mining analytical method based on business demand. This method can not only process the large number of web reviews, but also classify the customer reviews and recognize the different degree of customer attention at various time period. Through the use of this method, this system can automatically summarize the sentiment orientation of the products features, and finally get the features that need to be promoted and improved. In the meantime, this method can dynamically update the mining result, and display the result to managers by using the images, so that those managers will have a better understanding of the market, know the variation trend conveniently and make better business decision.This paper adopts the damped landmark time window model and text features extraction technology to summarize the product features and classify the sentiment orientation of the certain features. Through the experiment, this method is proved very effective in terms of dynamic analysis. This algorithm can not only get the accurate classification results, but also dynamically respond the customer’s changing opinions and automatically recognize the concerning information of the certain features.