The Lexicon in FCIDB : A Friendly Chinese Interface for DBMS

THE LEXICON IN FCIDB: A FRIENDLY CHINESE INTERFACE FOR DBMS Da Jinn Wang, Tsong-Y i Chen and Martha W Evens Department of Computer Science and Applied Mathematics 10 West 31' Street, Room 236 Illinois Institute of Technology, Chicago, IL 60616, U.S.A. EMAIL: {wangdaj, chentso}@harpo.cns.iit.edu, mwe@math.nwu.edu ABSTRACT FCIDB (Friendly Chinese Interface for DataBase management systems) can understand users' queries in the Chinese language. It works like a translator that translates Chinese queries into SQL commands. In the translation process, the lexicon of FCIDB plays a key role in both parsing and word segmentation. We designed some questionnaires to collect the frequently occurring words and add them to the public 'lexicon in FCIDB. FCIDB will produce a private lexicon for every new connected database. This paper will focus on the words included in the public lexicon and in the private lexicon. We also discuss the function, the structure, and the contents of the lexicon in FCIDB. 1. INTRODUCTION A natural language interface helps users communicate with computers in their own natural language. Natural language interfaces are an application of natural language processing. The goal of a natural language interface to database management systems (DBMSs) is to help users get information from a DBMS without knowing about databases or their contents. Some natural language interface systems have actually appeared, e.g., LADDER [3], DATALOG [2], and TELI [1]. Today, some domain-independent English interface systems for database are available as commercial products, e.g., English Wizard. English Wizard was developed by the Linguistic Technology Corp. [4]. It allows you to talk to several database management systems, which support Open DataBase Connectivity (ODBC), in the English language. FMDSIC (Friendly Medical Database System Interface in Chinese) is a domain-dependent interface [6]. A domain-independent Chinese interface system, as far as we know, has not appeared before. FCIDB (Friendly Chinese Interface for DataBase management systems) is a domain-independent interface system. It allows users to communicate with database management systems in Chinese. FCIDB translates Chinese questions into SQL, so that your DBMS receives the SQL it expects. The architecture of FCIDB includes the processing components: Look-Up & Word Segmentation, Parsing, Translating to SQL Commands, and Dialogue; and the information components: Lexicon, -215- Chinese grammar rules, SQL grammar rules, and Mapping Dictionary. Figure 1 shows the architecture of FCIDB. The function of the Look-Up and Word Segmentation component is to look up Chinese character strings in a lexicon and determine the word boundary. At the same time, this component obtains information about words from the lexicon, and sends them to the Parser. The Parser applies a set of Chinese grammar rules and a bottom-up parsing algorithm to parse user queries. The Parser must also choose the proper meaning for a word with multiple meanings. During translation the Parser output is reorganized into a legal SQL command. The legal SQL command will be sent to the Dialogue component. Dialogue sends the SQL command to the DBMS and gets the results back. Dialogue must then decide how to phrase the response to the users. …

Content truncated to 500 words