Behavior analysis based program understanding to support software maintenance

One of the most important tasks in software maintenance is to understand the behavior of the system’s parts one is deal-ing with. The common way for a maintainer then is to study the documentation of a system. However, more often than not, this documentation is far away from being up to date, which is due to the system’s continuous changes which are not reflected in its documentation. In such a situation a maintainer is lost and her/his only alternative is to dig in the system’s source code. This paper deals with the prob-lem of how to infer a (part of a) system’s behavior without having to look at its source code directly when the docu-mentation is not trustworthy. For this purpose we analyze the events which are emitted by a system. Such sequences of events contain much information which allow to reason about the behavior of the emitter. We present an approach to infer formal descriptions from event sequences and dis-cuss how these descriptions support the maintenance of a system