Stages in lexical knowledge acquisition

A prerequisite for lexical acquisition is to define the following items:

• An application dependent lexicon model   which defines relevant types of lexical information.
• A lexical database  model which defines the storage structure for lexical information.
• A characterisation of the sources of lexical data (e.g. orthographic transcriptions, other databases; manual information provided systematically or ad hoc by a lexicographer)
• Definition of procedures for constructing a lexicon from the data on the basis of these models and sources.
• Definition of procedures for validating the consistency of the database.
• Definition of procedures for extracting complex lexical information (e.g. coverage statistics for attributes and attribute combinations).
• Tools for implementing the procedures.

The logistics involved in these procedures are entirely non-trivial in a large project or consortium, and especially so with a complex spoken language understanding system or speech-to-speech translation system.