A New Approach to Programming Man-Machine Interfaces
The application of recent developments in web languages and machine learning of heuristics to problems in programming a flexible interface mechanism between a user and an application program. This report considers an adaptive communicator interposed between an application program and a set of input/output devices. The communicator is coupled with these I/0 devices by programs that transform raw signals from them into low-level logical objects. The communicator is envisioned as a series of context-analysis systems, each having a separate context and set of operating rules, and communicating with each other by transmitting logical objects. The report concludes that: (1) A labeled directed graph, or web, is an appropriate database organization for man-machine interaction. (2) Web grammars, in the form of pattern-replacement rules, can be used to manipulate that database. (3) Pattern-replacement rules can be viewed as heuristics suitable for machine learning.