Information Retrieval Based on Touch-Tone® Encoded Alphanumeric Keys

Abstract

Usage of the Touch-Tone telephone as a computer input device has increased dramatically during the past five years. The recent decrease in cost and increase in intelligibility of synthetic voice systems has made it possible to couple voice response systems with Touch-Tone telephones to create a low-cost, highly flexible computer terminal. However, usage is primarily restricted to applications that can be accomplished with a limited character set, do not need an extensive user protocol, consist of relatively short interactive sessions, and do not require immediate hard copy feedback. This paper discusses the first of these limitations: the unambiguous representation of 26 letters, 10 digits, and several special characters by the 12 keys of a Touch-Tone keyboard. Existing input schemes often require multiple keying to represent a single alphanumeric character or table look-up of numeric codes to represent a message. An alternative scheme would permit as-is encoding of alphabetic information. Using a specific application, as-is encoding is examined in detail, and the theory for recognition of encoded English phrases is generalized by two algorithms.