Intelligent interface to Unix

The new trend in computing is to provide intelligent assistance to computer users. Most commercial contemporary user-interface technology offers no concept of "intelligence" so that the user is relieved of learning command syntax. Existing methods of extracting user information are studied. Graph theory has been used to logically organize knowledge about Unix data manipulation commands and search algorithms were used to traverse the digraph many times, and in each case sequence of commands were generated. Few of these sequences define real-world user tasks. Research results showed that the adjusted breadth-first algorithm helped to generate the sequence of Unix commands from inputs extracted from user query. The approach followed could be implemented as a conventional software system or as an expert system that stores the digraph of Unix commands. For automatic task generation, we were facing an NP-Complete problem of subgraph isomorphism. With our small set of commands we were not able to define "tasks". A grammar approach generates different valid combinations, which can be studied by an expert system. (Abstract shortened with permission of author.).