Research On Information Refinement Technology Based On User Comments Data

With the influence of "Internet Plus" and "Made in China 2025",traditional manufacturing industry's sales and producting model are being in a profound change and adjustment.Explore the law of this change is given the great history mission to manufacturing industry practitioners and scientific research personnel.In order to study how to improve the competitiveness of manufacturing enterprises,this paper analyzes the user feedback in the production chain,found that manufacturing enterprises generally do not have real-time understanding of user feedback channels.In view of this phenomenon,this article on the Internet product user reviews as the starting point for research,mainly include the following three aspects:1.We have designed a data acquisition method with using element location.The localization method of XPath + ClassName is put forward based on the two tools of Selenium and Hibernate.And the experimental results show that the proposed algorithm achieves 99%accuracy and recall of data.2.We improved the Rwordseg word segmentation algorithm with the maximum probability chain principle,and strengthened the usability of the word segmentation lexicon.In this way,we reduced the word segmentation error rate by 2/3.And then the principle of DF,IG,?2 and MI feature extraction methods are analyzed and their computational quantities are compared on the basis of word segmentation.At last we chose DF word frequency method to realize the feature extraction of text.3.We propose a method of emotional analysis based on stepwise part-of-speech matching to analyze the influence of different parts of speech on the sentiment of the sentence and do the corresponding emotional score calculation.On the basis of this,the text representation model is established,and the result is more precise and detailed than the platform"?" mark.The data mining system developed in this paper can help the manufacturing enterprise to extract the user feedback in real time and reflect the user experience of the product in time.It is expected to effectively increase the sales volume of the product and provide consumers with a basis for purchasing.