Automated FAQ Answering: Continued Experience with Shallow Language Understanding

The subject of this researchis development of an evolving automated FAQ(Frequently Asked Question) answering systemthat providespre-stored answersto user questions askedin ordinaryEnglish. Thenatural languageprocessing technique developed for FAQretrieval does not analyze user queries; instead, analysis is applied to FAQs in the databaselongbefore anyuser queriesaxe submitted. Thus, the work of FAQ retrieval is reduced to keyword matching without in ferring; the systemstill creates an illusion of intelligence. Additional researchis donein orderto process phrases. The system is designed for ordinary websites such as those belonging to university, laboratories, software developers, etc.