YaLG - extending DCG for NLP
José João Almeida · 1995
Logical grammars are known to be powerful methods to specify syntax and semantics of language. In a prolog environment, dcgs are very popular and known to be simple and easy to learn and teach. When natural language is involved, dcgs are suitable for toy problems but difficult to scale up over big dictionaries, complex morphology or big texts to analyze. In this paper YaLG 1 (yet another logic grammar) is presented that enables the possibility of using dcg like grammars with external lexical analysis for natural language processing (NLP). YaLG is based on a set of modules that: ffl perform morphological analysis with external real size dictionaries and rules ffl give the possibility of scanning external files ffl give control over non-word elements of text ffl enable backtracking over multiple analysis Some difficulties on the interface prolog-C, namely large input, backtracking over the input and multi-analysis are discussed. 1 Introduction Design a Natural Language Processor i...