Study On Representation And Application Of Unstructured Information In Internet Consumer Comments

With the rapid development of the e-commerce, in order to enhance satisfaction of customers and share the customer’s shopping experience, online merchants allow customers to express opinions or recommendations has become a very common thing. Therefore, the number of reviews in the shopping site is rapidly increasing, and some of the best-selling products in large shopping site can reach tens of thousands as many user reviews. The emergence of a large number of network reviews, making product manufacturers or consumers to keep track of the purchased product user comments and suggestions difficult, which caused additional difficulties to their decision-making. Therefore, a new field of research--network reviews expressed and applied in this context is generating.This paper studies the unstructured information of the network comments, in order to make the information better to express their value. This article explores the theory and related applications in detail with reviews for network structure representation. This paper is completed to structure the network reviews via word processing, feature extraction and feature expression, and introduce the method of comprehensive evaluation about commodities with structured network reviews.This paper first researches reviews segmentation, POS tagging and the method of processing ambiguous words, unknown words, disabled words, and then analysis the advantages and disadvantages of the traditional feature extraction models, and on this basis this paper proposes the network reviews feature extraction model based on iterative ideology, and then it proposes the expressed model of the reviews based on fuzzy cognitive map through the knowledge representation and reasoning mechanism of fuzzy cognitive maps, give full consideration to the causal relationship and interactions between feature items and feature items, feature items and categories, categories and categories. On this basis this paper first introduces KNN classification methods to classify network reviews, and introduces the network reviews comprehensive evaluation model. We collected large customer reviews from the Internet during the experiment, and then do the corresponding experiment using the proposed models and methods。We get the effective experimental results and then verify the validity of the model.