A support system for software function discovery using histories of function executions

The majority of current application software packages have an extremely large number of functions because of the many different users envisioned for the software. This means that it is not easy for users to find useful functions. In this paper, the authors propose a system that supports the discovery of useful functions for users by allowing for shared referencing of function execution histories among users with similar uses for the software. The proposed system gathers a wide range of function execution histories over a long period of time, efficiently processes them, and then displays in a format that is easy to understand candidate functions that are unknown and useful to the user. The results of evaluation experiments using Microsoft Word 2000 and PowerPoint 2000 showed that an average of 63.7 functions were displayed as useful candidate functions to a total of nine subjects, with an average of 16.1 of them being unknown functions. In addition, the percentage of functions that users said were actually useful among the unknown functions discovered was 39. 2% on average for Word and 46.3% on average for PowerPoint.