Model Analysis Framework

Description

The Model Analysis Framework (MAF) project provides a powerful analysis tool for (meta) models built upon the Eclipse Modeling Framework (EMF). It allows to perform data-flow analysis (DFA) on meta models which can be defined using Java directives or an extended OCL syntax. The framework will then instantiate and evaluate the custom analysis directives for arbitrary instances of the meta model.
Use cases include but are not limited to the computation of the cycle time of business processes, the generation of test cases, the advanced validation of static semantics or the analysis of model metrics.

Currently supported features

• built upon the Eclipse platform (Helios)
• GUI independent framework core designed to run as background service
• can be easily extended using own input/output adapters
• write DFA equations using Java or OCL
• full support for EMF (meta) models
• internal model repositories for meta models / models / data flow definitions

Example Use Cases

• Several example use cases have been implemented on top of business processes which can be created using the Eclipse Java Workflow Tooling Editor
• Use cases include calculation of minimal/maximal predecessors and successor sets, identification of Strongly Connected Components (SCC), identification of Single-Entry-Single-Exit regions, validation of structural correctness and completion of quasi-structured components

Single-Entry-Single-Exit (SESE) component detection [1] and validation [2] for control-flow graphs has been implemented using flow-analysis (project MAF-DFA-JWT-TOKEN). The image below shows the SESE components of a JWT process model (one of which is highlighted green) and the results of the validation for structural correctness (click to enlarge).

[1] Token analysis of graph-oriented process models (Götz, Roser, Lautenbacher, EDOCW 2009)
[2] Faster and more focused control-flow analysis for business process models through sese decomposition (Vanhatalo, Völzer, Leymann, ICSOC 2007)

Roadmap

• The second stable version (1.1.0) has been released which includes editors for creating and configuring model analyses along with several example projects
• The source code can be obtained from the SVN (the plugins are described in the wiki)
• Current status is well-advanced, version 1.1 can be installed from the Eclipse Update Site (see below)

Installation

To install the Model Analysis Framework

• Download and unzip Eclipse Juno Modeling Package SR2
• In Eclipse, select "Help/Install New Software" and add the Xtext Update Site URL: http://download.eclipse.org/modeling/tmf/xtext/updates/composite/releases/ (MAF requires Xtext 2.4.1)
• In Eclipse, select "Help/Install New Software" and add the MAF Update Site URL: http://pvs.informatik.uni-augsburg.de/maf/1.1.0/
• Select the update site and install the available features from the "Model Analysis Framework" category
• Under "File/New/Model Analysis Framework" you can select and install several example projects

About the project

Model Analysis Framework is a research project located at the Programming Distributes Systems Lab at the University of Augsburg, Germany (project site).
The concept of using a data-flow based approach for model analysis is described in the paper Data-flow Based Model Analysis

from https://code.google.com/a/eclipselabs.org/p/model-analysis-framework/