Research On Product Reviews Based On Domain Ontology

With the rise of online media such as blog, microblog, and forum, many users are inclined to publishtheir evaluations about products on the network, including the product performance. Corpus analysis doesnot only can make the manufacturers understand the evaluations of product users and product defects, butalso can keep users' needs and expectations in mind. Potential users are likely to check up the evaluationsof product users online before buying something, get to know the product performance and compare similarproducts in order to make a reasonable choice. At present, the network has accumulated a lot of productevaluation resources, and it is a current research focus to obtain the implication of the subjective opinions.Natural language description for product evaluation is unstructured data, from which the user wants toaccess to information only through the manual reading. However, the query in this vast information istime-consuming and error-proneness, in order to solve such problems, product evaluation technologycomes into being. The technology can automatically get product reviews information about the mainattributes of the product and evaluation words, finally, through analysis and statistics, providing users witha variety of evaluative information to help them to make a good choice. Evaluation object extraction andanalysis of subjective sentence tendency are the research emphasis and chanllenge. This article made someattempts for these two points with the introduction of domain ontology, arranged as follows:(1)Take portable computer for an example to build the product ontology and opinion vocabularyontology. The former is used for the evaluation of object extraction, and the latter aims to calculate thetendency of the subjective evaluation of sentence and its strength.(2)Using crawling tools to crawl pages from the network, extract content of each text, process textsand design an evaluation object extraction algorithm based on the product ontology.(3) According to set of rules for the sentence established in this paper, an N-POS model based on thecontinuous mode is proposed, which make subjective sentence extraction based on views lexical ontologybecome possible, and a subjective sentence tendency algorithm is designed.(4) Research on the product evaluation for applications: the portable computer evaluation and analysisapplication examples is introduced, a detailed description of the main function module in the system is given, and technology applied to the practical value of life is showed by realizing the main function forproduct evaluation analysis.