The Computational Models of Natural Language Processing

Transcription

1 J OU RNAL OF CH IN ESE IN FORMA TION PROCESSIN G Vol. 21, No. 3 May, 2007 : (2007) : Internet,,,, ( ),,,,, 2,, (, ) :,,,,, : ; ;;; ;; : TP391 : A The Computational Models of Natural Language Processing ZHAN G Bo (Department of Computer Science & Technology Tsinghua University, Beijing , China) Abstract : In this paper, we will discuss the computational models of natural language processing. There have been several kinds of computational models such as analytical model, statistical model, hybrid model, etc ; each has its own characteristics and limitations. As an ill2posed problem, we ll discuss what the essential hardness the natural language processing has, what challenge we will confront with, and what measures we ll adopted to solve the diffi2 culty. Key words : artificial intelligence ; natural language processing ; computational model ; analytical model ; statistical model ; hybrid model ; ill2posed problem 1 [1 ],,, : : : ( ) ; 973 (2003CB317007,2004CB318108) : (1935 ),,,

2 4 2007,,,,,,,,,,,,, (,, ), ( ) ( ),! 1956, ( Comp uter Chess) (Machine Translation),,,, H. Simon,, 1997, 40 ( 10 ), IBM (Deep Blue),?!,,,, [2 7 ],: ( ) (Local Ambiguity), (Contextual Dependency),,,,,,,,,,,,,, 2 N (Nat ural Language),, P ( Problem), P N ( I, O),P, M ( Mod2 el), A ( Algorit hm) M A P F ( Comp utational Model), M,, F M, M, F, ( I), O, O F, F,P ( 1 ) F 1 F( I, O), I,, O,,, 01, : F I, O, F I O

3 3 : 5 (Mapping), F : I O P, ( ) - F,, 2 F I O,, ( Ill2posed),,,,, (Strongly Ill2posed Problems),, :,,,, : ( ),, [8,9 ], [8 ],,(Constraint), ( Well2Posed), ( Regularization), ( ),,, 3 N. Chomsky, ( Generative Grammar) [10 ],,, ( Ac2 ceptable Grammatical Sentences),, ( Rule2Based), (Analytical Mod2 el),,,, ( I),,, ( Scalability) 4 Chomsky (Language Acquisi2 tion), Chomsky, B. F. Skinner [11 ], (Statistical Model) [12 ] (Learning From Data),,,, (Data Driven),, (Corp us2based),,, F: I O, [13 ],,,

4 6 2007,,, (Long2Distance Depend2 ency),,, (,, ),,,,, ( ), 5,,,,,,,,,,,,,, ( Hybrid Mo2 del), [14 17 ],,,, (60 ), : (1) The spirit is willing but t he flesh is weak. ( ) ( ),, : ten. (2) The Volt ka is strong but t he meat is rot2 (, ),, (1), (1) (2),, spirit (, ) Volt ka (),, flesh (, ) meat ( ),?,,,,?,,,,,,,,,,,?,,,,?,,,, IBM,,,,,, 10 15, 3 5,, IBM, 60 ( ), ( ),,, 6,,

5 3 : 7 01,,,, (, ),,,,, : [1 ],,. [ M ]. :,2005. [2 ] Gibson, E., Linguistic complexity : Locality of syntac2 tic dependencies [J ]. Cognition, 1998, 68 : [3 ] Daniel Grodner, Edward Gibson and Duane Watson. The influence of contextual contrast on syntactic pro2 cessing : evidence for strong2interaction in sentence comprehension [J ]. Cognition 2005, 95 : [4 ] Silvia Gennari and David Poeppel. Processing corre2 lates of lexical semantic complexity [ J ]. 2003, 89 : B272B41. Cognition [5 ] Tessa Warren and Edward Gibson. The influence of referential processing on sentence complexity [ J ]. Cognition 2002, 85 : [6 ] Gerry Altmann, Mark Steedman. Interaction with context during human sentence processing [J ]. Cogni2 tion 1988, 30 : [7 ] Douglas Roland, Jeffrey L. Elman and Victor S. Fer2 reira, Why is that? Structural prediction and ambiguity resolution in a very large corpus of English sentences [J ]. Cognition 2006, 98 : [8 ] Tikhonv, A. N., Arsenin, V. Y.. Solution of III posed problems [ M ]. New York : Winston/ Wiley [9 ] Bakushinsky, A., Goncharsky, A.. Ill2posed prob2 lems : Theory and Applications [ M ]. Dordrecht/ Bos2 ton/ London : Kluwer Academic Publishers, [10 ] Chomsky, N.. Syntactic structures [ M ]. The Hague : Mouton, [11 ] Skinner, B. F., Verbal Learning [ M ]. New York : Appleton2Century2Croft s, [12 ] Christopher D. Manning, Hinrich Sch tze. Founda2 tions of Statistical Natural Language Processing [ M ]. Cambridge, Massachusett s : The MIT Press [13 ] Vladamir N. Vapnik, Statistical Learning Theory [ M ]. New York : John Wiley & Sons, Inc., [14 ] Aue, Anthony, Arul Menezes, Robert Moore, et al. Statistical Machine Translation Using Labeled Se2 mantic Dependency Graphs [ A ]. In : Proceedings of the 10th International Conference on Theoretical and Methodological Issues in Machine Translation [ C ]. Baltimore, [ 15 ] Pinkham, J, and M. Corston2Oliver, Adding Domain Specificity to an M T System [ A ]. In : Proceedings of the Workshop on Data2driven Machine Translation at 39th Annual Meeting of the Association for Computa2 tional Linguistics [ C]. Toulouse, France, 2001, [16 ] Richardson, S., W. Dolan, A. Menezes et al. Achieving Commercial2quality translation with exam2 ple2based methods [ A ]. In : Proceedings of M T Sum2 mit V III [ C]. Santiago De Compostela, Spain, 2001, [17 ] Wu Andi. Statistically2Enhanced New Word Identifi2 cation in a Rule2Based Chinese System. In : Proceed2 ings of the Second Chinese Language Processing Workshop [ C]. Hong Kong, China, 2000,