Balboa is a server-based process event data analysis system. It provides a framework for easing the construction of event data analysis tools. Included in the Balboa distribution are two analysis tools: Discovery, a tool that discovers state-machine patterns from event data; and Validation, a tool that measures the correspondence between an event stream (process execution) and a state-machine model of the process.
The Balboa framework consists of a server and data management tools that let a user define event collections and event mappings. The server then serves mapped event data to tools through TCP socket connections. The library that tools use to talk with a Balboa server also supports accessing local file-based event data, so that tools can be used independently of a Balboa server, without recoding or recompilation.
The Discovery tool uses statistical and algorithmic methods to find state-machine patterns in event data (traces of executions of a process). The methods in this tool allow control over the detail that is discovered, and are robust in the presence of noise in the data.
The Validation tool uses model-searching techniques to find and measure how closely an event stream matches a model of the process. This measurement represents how closely the actual process (the event stream) matches what is expected or prescribed (the model). The measurements are based on string distance methods, and count the number of event insertions and deletions that must be performed to make the event stream match exactly what the model expects.
Balboa requires a C++ compiler, Gnu make, and Tcl/Tk (at least 7.5/4.0). Additionally, all graph drawing for the Discovery and Validation tools is done using the 'dot' tool from ATT Labs.
|
|
Fetch a Balboa distribution from SRM |
|
|
Fetch Tcl/Tk from Sunlabs |
|
|
Fetch dot from ATT |
|
|
More info on Balboa |
|
|
Process discovery and validation |
©1998 Software Engineering Research Laboratory (SERL)