An application-independent subsystem for free-text analysis

Abstract

The analysis of free English text is a major subproblem in medical computing. A freetext-analysis subsystem, written entirely in MUMPS language, is in operation at the Laboratory of Computer Science, Massachusetts General Hospital. This subsystem, built around a numerical taxonomy of anticipated input, services multiple application programs. Facilities are provided for matching, spelling correction, disinflection, and phrase analysis. A variety of administrative programs are provided for logging uninterpreted input and for dynamically incrementing the data-base so as to expand the range of correctly interpreted input. © 1976.