Clustering Algorithm And Its Application In The Review Mining

With the increasing development of science and information technology, e-commerce is getting more and more popular. In this case, clients as well as consumers pay more attention to product evaluation when they go shopping. With the arrival of big data and massive data, people are faced with numerous and complicated products everywhere, so how to selectthe most satisfactory and cost-effective product is becoming a key point to clients and consumers. In the above background, digging into the product comments in order to acquire valuable information is getting widely recognized. Comments mining technology is an important aspect of data mining technology, and comments mining technology is becominga hot research area along with the progress of machine learning, information retrieval, natural language processing and data mining technology, more importantly, many research results are widely used in daily life.This paper mainly aims to put forward a product comments mining approach that can be broadly applied to multitudinous products when it comes to E-business in hotel. Based on the study of key technology of product comments, the method includes the following parts. The first step is to gain a large number of valuable comments and have them processed; the next step is to extract some keywords or phrases in the processed comments and at the same time, the part extracted is supposed to keep the original meaning; the third action is to divide the comments into different attributes by clustering algorithm or classification algorithm; and the last part is to calculate the affective satisfaction referring to related attributes or entirety, depending on affection lexicon, negation lexicon and affective computing method.By the extensive research on comments mining technology and related key technology (feature extraction of comments, clustering algorithm andclassification algorithm), this paper puts forward MMACA clustering algorithm as well as calculation method of affective satisfaction, and reveals a detailed discussion. At the same time, depending on the features of hotel reservation comments, this paper successfully makes the extraction of comment objects and evaluation terms come true by the means of a simple but efficient bilateral matching algorithm.Inspired by dividing clustering algorithm, hierarchical clustering algorithmand density clustering algorithm, aiming at the clustering technology used in comments mining area, this paper presents a new clustering algorithm-MMACA clustering algorithm. Due to the features of hotel reservation comments, MMACA clustering algorithm can realize the calculation of affective satisfaction on comments attributes, single comment and entire comments.Depending on the hotel reservation comments data and international conventional data, MMACA clustering algorithm is tested. What is more, the test reveals that MMACA clustering algorithm is not only reasonable and feasible, but also has a good applying prospect.