TULIPS - Teachable, Understanding Natural Language Input Problem-Solver
نویسنده
چکیده
Th is paper desc r ibes the p r i n c i p a l f e a t u r e s of TULIPS program designed as a w i d e l y o r i e n t e d A I system t h a t accepts n a t u r a l language (NL) i n p u t . NL processor o f TULIPS is guided by i n f o r m a t i o n r e p r e sented i n models o f c u r r e n t " e x t e r n a l w o r l d " domain and of c u r r e n t user and u t i l i z e s deduc t i ve and i n d u c t i v e mechan isms. These f ea tu res a l l o w the program: t o d i scove r the most r e l e v a n t i n t e r p r e t a t i o n o f an i n p u t , t o e x t r a c t the d e s c r i p t i o n s o f u s e r ' s goa ls from NL u t t e r a n c e s , and to generate the s o c a l l e d T-problems ( T f o r TULIPS). Having so lved them TULIPS not on ly answers u s e r ' s request but a l so l e a r n s new knowledge on i t s e n v i ronment and improves i t s own mechanisms. I n t r o d u c t i o n I t need h a r d l y be sa id about a t t r a c t i veness of NL communications w i t h a p r o gram. TULIPS's user must no t l e a r n s p e c i a l i n t e r a c t i v e languages and can pass to a computer not a f o r m a l i » e d d e s c r i p t i o n o f the a l g o r i t h m bu t a d e s c r i p t i o n o f the problem i t s e l f ( an i n p u t f o r prob lems o l v i n g b lock o f the program ) • I t must be e s p e c i a l l y emphasised t h a t NL d e s c r i p t i o n of a problem can be of the p r e l i m i nary i n f o r m a l k i n d . For TULIPS i s i m p l e mented as a d i a l o g s y s t e m , user is ab le to s p e c i f y and supplement t h i s d e s c r i p t i o n i n case o f need. Thus TULIPS c a r d i n a l l y d i f f e r s f rom " t r a d i t i o n a l " computer program. I t i s not a d e t a i l e d fo rma l d e s c r i p t i o n o f an a l g o r i t h m f o r the c u r r e n t u s e r ' s problem, but a genera to r o f such a l g o r i t h m s . I t r ece i ves as i npu t not a command to a c t i va te known to user B t r i c t segment o f a l g o r i t h m , but on ly a d e s c r i p t i o n o f u s e r ' s problem ( i . e . user s a i s what he wants w i t h o u t any s p e c i f i c a t i o n how i t Bhould be done ) . Jus t t h i s p rope r t y of TULIPS i t s a b i l i t y a u t o m a t i c a l l y so lve the prob lem, be ing t o l d t he word ing o f i t a l l ows to cons ide r the program as AI sys tem t h a t automat izes so n o n t r i v i a l f u n c t i o n s o f human i n t e l l i g e n c e as s p e c i f y i n g of a problem to the w e l l f o r m e d one and s o l v i n g o f such w e l l f o r m e d prob lem. H i g h l y i n t e l l i g e n t work does, however , not on ly the So lve r of TULIPS but a l s o i t s Ana lyse r . I t s goal i s t o e x t r a c t f rom NL (Russian) i npu t the d e s c r i p t i o n s o f problem s i t u a t i o n s , u s e r ' s goa l s , and own d u t i e s ( represented in u t te rances bo th e x p l i c i t l y and i m p l i c i t l y ) . To ext r a c t i m p l i c i t i n f o r m a t i o n the Analyser should take i n t o c o n s i d e r a t i o n a contex t ( l o c a l and g l o b a l ) o f the cu r ren t commun i c a t i o n p rocess . A l o c a l con tex t of NL u t t e r a n c e i s an i n f o r m a t i o n ex t rac ted f rom the p rev ious i n p u t exp ress ions . A g l o b a l con tex t i s de f i ned by models o f " a b s t r a c t " u s e r , " e x t e r n a l w o r l d " as a who le , p a r t i c u l a r users known f o r TULIPS, s p e c i f i c problem-domains ( o f u s e r ' s a c t i v i t y ) , "gene ra l d u t i e s " o f TULIPS, and concre t h e u r i s t i c s and o the r t o o l s r e l a t ed to a concre t problem-domain or u s e r . These k inds of models must be i n t r o duced because of TULIPS is a w i d e l y o r i ented system. The program is to so l ve d i f f e r e n t k indes o f problems i n d i f f e r e n t concre t domains and to adapt to subset of Russian be ing used, s p e c i f i c f ea tu res o f problem-domains and u s e r s . Th is main f e a t u r e of TULIPS w i d e o r i e n t a t i o n de termines a l l the o the r i t s f ea tu res and, in p a r t i c u l a r , the approach to NL p roces -
منابع مشابه
TULIPS-2 - Natural Language Learning System
The learning of a natural language is considered to be an important aspect of man-machine communication in human language. The methods of the Russian language knowledge representation and acquisition implemented in the experimental understanding system TULIPS-2 are described. These methods provides for understanding utterances that contain words and structures unknown to the system wherther the...
متن کاملThe Complexity of Math Problems - Linguistic, or Computational?
We present a simple, logic-based architecture for solving math problems written in natural language. A problem is firstly translated to a logical form. It is then rewritten into the input language of a solver algorithm and finally the solver finds an answer. Such a clean decomposition of the task however does not come for free. First, despite its formality, math text still exploits the flexibil...
متن کاملUnit Dependency Graph and Its Application to Arithmetic Word Problem Solving
Math word problems provide a natural abstraction to a range of natural language understanding problems that involve reasoning about quantities, such as interpreting election results, news about casualties, and the financial section of a newspaper. Units associated with the quantities often provide information that is essential to support this reasoning. This paper proposes a principled way to c...
متن کاملExternal Propagators in WASP: Preliminary Report
State-of-the-art ASP solvers are based on a variant of the CDCL algorithm. One of the key features of CDCL is the propagation step, whose role is to implement deterministic consequences of the input theory. It is well-known that the performance of solvers can be considerably improved on specific benchmarks by adding custom propagation functions. However, embedding a new propagator into an exist...
متن کاملCan Computer Understand and Solve Turkish Arithmetic Problems?
Understanding is one of the most important research areas of Natural Language Processing. Problem solving is a way of realizing machine understanding. This paper presents a problem solver system which can understand and solve arithmetic problems in Turkish. There are three main phases of the system: morphology, syntax and semantic. The difference from other systems that is; data modeled as sema...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 1975