Role of PROLOG (PROgramming and LOGic) in Natural Language Processing

摘要

The field of Artificial Intelligence strives to produce computer programs that exhibit intelligent behavior. One of the areas of interest is the processing of natural language. This report will discuss the role of the computer language PROLOG in Natural Language Processing (NLP) both from theoretic and pragmatic viewpoints. The reasons for using PROLOG for NLP are numerous. First, linguists can write natural language grammars almost directly as PROLOG programs; this allows 'fast-prototyping' of NLP systems and facilitates analysis of NLP theories. Second, semantic representations of natural language texts that use logic formalisms are readily produced in PROLOG because of PROLOG's logical foundations. Third, PROLOG's built-in inferencing mechanisms are often sufficient for inferences on the logical forms produced by NLPs. Fourth, the logical, declarative nature of PROLOG may make it the language of choice for parallel computing systems. Finally, the fact that PROLOG has a de facto standard (Edinburgh) makes the porting of code from one computer system to another virtually trouble free. Perhaps the strongest tie one could make between NLP and PROLOG was stated by John Stuart Mill in his inaugural Address at St. Andrews: The structure of every sentence is a lesson in logic.