ORGANIZING STRUCTURED AND UNSTRUCTURED DATABASE COLUMNS USING CORPUS ANALYSIS AND CONTEXT MODELING TO EXTRACT KNOWLEDGE FROM LINGUISTIC PHRASES IN THE DATABASE

Corpus analysis methods have previously been applied to text, typically to annotated text. The invention shows how to apply corpus analysis methods to information captured in databases, where the database columns include a mixture of both structured domains and unstructured domains containing text. It uses case-based methods to automatically organize cases for periodic review. The invention can help to identify opportunities for increasing knowledge about databases. By organizing a database around common lexical, semantic, pragmatic and syntactic relationships, the invention can be used to increase the effectiveness of previous corpus analysis methods, and to apply them to a diversity of commercial applications. The invention applies contextual constraints to focus the application of linguistic methods. This invention can provide a component for medical records, enterprise databases, information retrieval, question answering systems, interactive robots, interactive appliances, linguistically competent speech recognition, speech understanding and many other useful devices and applications that require a high level of linguistic competence within operational contexts.