Research On Several Key Problems Of Web Text Minin And Its Application In Online Evaluation Of Electromechanical Products

Recently, with the rapid development of Internet, the shortages of several key problems research of web text mining in text extraction of complex web pages, new words recognition and features clustering of product, are emerging gradually. Existing methods in accuracy and efficiency of complex web pages text extraction have yet to reach satisfactory result. The accuracy of new word recognition need to be increased. And most of the existing methods concentrate less on product features clustering of online reviews. In this paper, we study the above three problems based on relevant research work. First, a method for complex web text extraction based on statistic was proposed in this paper, this method can extract web text more accurately and efficiently from enormous source code of web pages and it can provide more pure and complete raw data for web text mining. Then, we proposed a method for Chinese new words recognition based on high frequency monosyllabic to improve the accuracy and efficiency of new words recognition, more importantly, it also can provide effective support for web text processing, such as information extraction, theme mining and semantic analysis. Finally, we proposed a method for product features clustering of online reviews based on semantic information, this method has an important in obtaining the attention from consumers about the performance of product, and it also can help businesses promote the product quality.In addition, we apply the above three methods in online evaluation of electromechanical products for providing effective support to extract valuable information. First, the proposed method for web text extraction is used to extract the text of online evaluation of electromechanical products, which can improve the accuracy and the degree of automation. Then, the proposed method for new words recognition based on high frequency monosyllabic is used to identify the new product features in the online evaluation, which can provide more useful information for businesses or users. Finally, the proposed method for product features clustering is used to cluster the same or similarity features which are in different description and that can help businesses obtain users attention on the product performance.