A GENERAL SEMANTIC ANALYSER FOR DATA BASE ACCESS

The paper discusses the design p r i n c i p l e s and current s tatus of a na tu ra l language f r on t end for access to data bases. This Is based on the use, f i r s t, o f a semantic a l l y- o r i e n t e d question analyser e x p l o i t i n g genera l, language-wide semantic categor ies and pa t t e rns, ra ther than data base-spec i f i c ones; and, second, of a data base-or iented t r a n s l a t i o n component for ob ta in ing search s p e c i f i c a t i o n s from the meaning representat ions for quest ions der ived by the analyser. This approach is motivated by the des i re to reduce the e f f o r t of p rov id ing data base-spec i f ic mater ia l for the f r on t end, by the b e l i e f tha t a general analyser is we l l su i ted to the "casua l " data base user, and by the assumption tha t the r i c h semantic apparatus used w i l l be both adequate as a means of analys is and appropr iate as a too l for l i n k i n g the charac te r i sa t i ons of input and data language i tems. The paper descr ibes t h i s approach in more d e t a i l, w i th emphasis on the e x i s t i n g, t e s t e d, analyser. We are t r y i n g to bu i l d a f r on t end processor for natural- language access to data bases based on the p r i n c i p l e tha t the semantics as wel l as syntax of the language analyser should be general and not d a t a- s p e c i f i c. The func t ion of the analyser is to bu i l d a meaning representat ion of an input question which is then converted to a formal data base search query, s p e c i f i c a t i o n by a separate t r a n s l a t i o n component. The reason for adopting t h i s approach is to reduce the ana lyser ' s dependence on the spec i f i c p roper t ies of the data base, and hence the cost of dea l ing w i th new data bases. The proper t ies of the data base necessar i l y determine the form of search s p e c i f i c a t i o n s; but the assumption is tha t the t r a n s i t i o n from t e x t question to search query i