This thesis puts forward a kind of smoothing mechanism on the basis of traditional statistical information, which guarantees recall rate of candidate names extraction reach 99%;

Algorithms that scan text for common patterns, such as email addresses, phone Numbers, and people and place names, are useful for named entity extraction.

The recognition of organization names is one of the fundamental tasks in Information Extraction and Machine Translation.