tailieunhanh - Báo cáo khoa học: "NATURAL-LANGUAGE ACCESS TO DATABASES--THEORETICAL/TECHNICAL ISSUES"

Although there have been many experimental systems for natural-language access to databases, with some now going into actual use, many problems in this area remain to be solved. The purpose of this panel is to put some of those problems before the conference. The panel's motivation stems partly from the fact that, too often in the past, discussion of natural-language access to databases has focused, at the expense of the underlying issues, on what particular systems can or cannot do. To avoid this, the discussions of the present panel will be organized around issues rather than systems. . | NATURAL-LANGUAGE ACCESS TO DATABASES THEORETICAL TECHNICAL ISSUES Robert c. Moore Artificial Intelligence Center SRI International Menlo Park CA 94025 I INTRODUCTION Although there have been many experimental systems for natural-language access to databases with some now going into actual use many problems In this area remain to be solved. The purpose of this panel is to put some of those problems before the conference. The panel s motivation stems partly from the fact that too often in the past discussion of natural-language access to databases has focused at the expense of the underlying Issues on what particular systems can or cannot do. To avoid this the discussions of the present panel will be organized around issues rather than systems. Below are descriptions of five problem areas that seem to me not to be adequately handled by any existing system I know of. The panelists have been asked to discuss In their position papers as many of these problems as space allows and have been Invited to propose and discuss one Issue of their own choosing. II QUANTITY QUESTIONS Database query languages typically provide some means for counting and totaling that must be Invoked for answering how much or how many questions. The mapping between a natural-language question and the corresponding database query however can differ dramatically according to the way the database Is organized. For Instance if DEPARTMENT is a field in the EMPLOYEE file the database query for How many employees are In the sales department will presumably count the number of records In the EMPLOYEE file that have the appropriate value for the DEPARTMENT field. On the other hand If the required Information Is stored in a NUMBER-OF-EMPLOYEES field in a DEPARTMENT file the database query will merely return the value of this field from the sales department record. Yet a third case will arise If departments are broken down Into say offices and the number of exployees In each office is recorded. Then the .