NameFinder: software that finds names in text

Abstract

A common problem in the information industry is providing index terms for on-line text on the basis of company, person, location, or other named entities. The NameFinder™ system can provide high accuracy indexing for names of such entities. In addition to literal names, it recognizes many variations of the names listed in its database. Extensive customization facilities allow this recognition to be fine tuned for particular names, and automatic knowledge engineering rules help reduce the amount of human effort involved in this fine-tuning. This paper discusses the basic capabilities of the commercially-available NameFinder, some issues in knowledge engineering, and possible future extensions.