A Commodity Information Search Model of E-Commerce Search Engine Based on Semantic Similarity and Multi-Attribute Decision Method

Beginning of article

1. Introduction

Nowadays, the advance of Internet and Web technologies has continuously boosted the prosperity of e-commerce. Through the Internet, it has become daily life for people to online shopping, and the number of people buying, selling and performing transactions on the Web is increasing at a phenomenal pace. With the further development of e-commerce, it will not be easy for customers to single out the best commodity when faced with the massive commodity information in the Internet. Usually, customers utilize various E-commerce search engines to search and compare commodities when they do online shopping in the Internet. Therefore, E-commerce search engines have largely become the main methods for customer to acquire commodity information and relevant services in the course of e-commerce activities. However, common search engines (such as Google, baidu, etc) and keyword-based search are not only low-efficient, but also sometimes the retrieved document contents of web pages are non-relevant with customer's query. The main reasons that result in these problems are: 1) the traditional information search techniques cannot express the semantic information correctly, and the information search based on keyword-matching still causes the semantic inaccuracy of retrieved results. 2) The heterogeneous characteristic of information organization is very obvious because of the diversity of e-commerce platform and the standard deficiency of relevant domain information description. 3) There are still not effective commodity evaluation and comparison mechanism so as to cause the information overload of the retrieval results.

In order to solve the existent problems that traditional information search methods have, many scholars both at home and abroad propose many new web search approaches, which are based on ontology and semantic web. Popov defines a general framework for document search that is supported by ontology, and integrate full-text search with ontology-based methods (Popov, 2004). Pablo presents a semantic search model based on an adaptation of the classic vector-space model, including an annotation weighting algorithm, and a ranking algorithm (Pablo, 2007). TAP utilizes semi-automatic techniques to extract relevant knowledge from free texts and semi-structured data (Guha, 2003). In TAP, the system transfers the document into the semantic web document format after analyzing the semantics of the document, and consequently utilizes the structured knowledge to improve the precision ratio of information. Wang also presents a semantic-similarity based information search method (Wang 2006), which utilizes ontology to describe semantics of the customer queries and Web documents, and computes the semantic similarity between the concept and property of domain knowledge to realize the semantic information search. Zhang presents a semantic annotation method based on ontology, which can be used in intelligent search systems for semantic reasoning (Zhang 2008).

All of these methods can solve the semantic inaccuracy of information to some extent, and can realize the semantic accordance between the customer's query and document information. However, in order to solve the information overload of commodities and provide accurate information search and shopping service for customers, one kind of effective commodity evaluation and comparison mechanism should be constructed, which is based on the realization of semantic retrieval of general information. In order to realize it, the paper presents an intelligent commodity information search model of E-commerce search engine, which integrates semantic retrieval and multi-attribute decision method. First, semantic similarity can be computed by constructing query semantic vector and document semantic vector respectively based on ontology, in order to realize the semantic consistency between retrieved results and customer's query. Besides, TOPSIS method is also utilized to construct the comparison mechanism of commodity by calculating the utility value of each retrieved commodity, and choose the most suitable one for customers. Therefore, the intelligent model not only …