电子商务中针对产品的摘要挖掘技术研究计算机科学与技术专业论文.docxVIP

哈尔滨工业大学工学硕士学位论文 哈尔滨工业大学工学硕士学位论文 哈尔滨工业大学工学硕士学位论文 哈尔滨工业大学工学硕士学位论文 - - PAGE IV - - - III - Abstract With the rapid development of Internet, an increasing number of people go shopping online for convenience, the era of e-commerce began. For now, there are massive goods for people to buy on online shopping site. There is a great number of goods and the amount of information about goods is huge, so it is inconvenient for the buyers to find the commodities they needed. In fact many goods of the same kind which are sold by different sellers is a product made by the same manufacturer, so constructing virtual product nodes and mining the product information based on the description of goods is usefull. The information of products includes plane text descriptions, product summaries, product attributes, product reviews and price trends. In a simple and friendly way it uses comprehensive and detailed information to display products. This paper focuses on the product summary, and presents a method of mining the summary of products in e-commerce. Product summary uses a brief language which is formed by 3 to 5 phrases to descripe the product and aims at highlight the products selling points and features. Product summary hears from the users’ point of view, provide a unique description of the product and gives users an overall impression when the users purchase the products, then help them make better choices, and also reduce user the purchase time. Products summary mining is a multi-document summary problem, but unlike traditional summary which construct summary through selecting important sentences. The product summary is formed by the phrases. This method involves two key parts, the first is the sentences clustering to form the sub-themes; Second, extracting important phrases from the sub-theme as summary phrases. For clustering sentences to form sub-themes, this paper discusses the sentence representation, similarity calculation and clustering methods. We use hierarchical clust