PURPOSE: A method for modeling user patterns to enhance recognition of a named entity and a named entity list generator are provided to reduce perplexity and increase accuracy by remarkably reducing an out-of-vocabulary rate. CONSTITUTION: The named entity list generator includes a named entity identifier(204) and an analyzer(216). The named entity identifier receives information partially matched with a specified user group from multiple different sources and identifies the named entities in the information. The analyzer obtains a group-specific named entity list and associated probabilities by calculating the probabilities associated with the identified named entities.
展开▼
