A natural language processing (NLP) system has been developed as a component of a road accident data analyzer. The NLP extracts information from the English text within road accident records for joint analysis with the coded information in other parts of the records. The English text is characterised by deviant linguistic phenomena whilst being restricted in linguistic usage. A corresponding domain specific declarative grammar and lexicon are derived from the dataset using semantic information represented in a form of lambda expressions. A semantic oriented constraint-based formalism is adopted for the grammar and is enhanced by a system network technique. An interesting corner passing strategy based upon a bidirectional chart parsing algorithm is adopted: this provides convenience in the incremental development of the system. The information extracted from the text is represented in a form of predicates and can be readily incorporated into a functional database. Some useful techniques have been identified and developed, including a core phrase principle for system development, lambda variable localisation in meaning interpretation and a subsumption principle for constraint imposition.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £60 +P&P)