Research And Design Of Merchandise Query Service System

Having many apparent advantages such as great convenience, low cost, high efficiency and strong selectivity, online shopping has been widely accepted and become a business form that can not be neglected. Nevertheless, in the initial period of shopping on the Internet, consumers have to follow a time-consuming procedure to first find the most favorable merchandise or product: they may try to visit a number of different websites, repeatedly fill up certain "search form" on each one of these sites before they can get any information. In order to make people's online shopping experience more natural, friendly and humane, this dissertation brings forward the research and design of an intelligent Merchandise Query Service System which allows people to use natural language to inquire about merchandise information from online shops. The system first translates natural query sentences to corresponding SQL queries of background databases of those online shops, then forwards these SQL queries to each website where they are executed, collects data replied and finally returns integrated merchandise information to its users. Our Merchandise Query Service System is practically an application of Natural Language Interface to a Database(NLIDB) which aims at converting natural language query in Chinese into SQL query to relational database where merchandise data is stored. In this dissertation, we first describe the design of system knowledge base in detail. Then in the course of lexical analysis, an improved Chinese word segmentation method called IMM(which is based on the Maximum Matching Method) is proposed and used to segment the query sentence into word-chain, methods of how to identify numeric words and disambiguate the database semantic meanings of words are also discussed; In the course of query semantic understanding, we propose a SQL Conversion Oriented Query Understanding Model(SCUM) which is used to translate word-chain to executable SQL command. To prove the feasibility and validity of our Merchandise Query Service System, we have developed a prototype system—iAnswer, which can provide natural language query service(including query about hidden information) to databases of multiple online shops on the basis of effective extraction of database semantic knowledge.