Natural-Language Interfaces

摘要

Publisher Summary Since the early 1960s when support decreased for machine translation, much of the research on natural-language processing in North America has been motivated by its potential use for communicating with software systems. Natural-language systems have been developed to extract information from databases, to control (simulated) robots, to interact with graphic systems, to specify simulation problems, and to communicate with systems embodying expertise in some task or problem area. This chapter discusses the interfaces to database management systems. Apart from being among the earliest interface systems developed, interfaces to databases account for most of the natural-language interfaces (NLIs) implemented and they are the subject of a substantial literature. The chapter discusses the main system architectures used in NLIs and the body of techniques developed for them. In doing so, it distinguishes between the task of an interface and its domain. Natural language is but one of the methods available for human–machine interaction. The reasons for its attractiveness are obvious. They are: it provides an immediate vocabulary for talking about the contents of the database and a means of accessing information in the database independently of its structure and encodings, shields the user from the formal access language of the underlying system, and is available with a minimum of training to both novice and occasional user.