THE GenParse PROJECT        

                                                            

                                  Grammar inference for Domain-Specific languages

                                 

 

                    home • research • people • contact • related links

 

                                                                     

 

[ Research ]

 

[ Papers ]

Journal

* Matej Crepinsek, Marjan Mernik, Barrett Bryant, Faizan Javed, Alan Sprague, "Inferring Context-Free Grammars for Domain-Specific Languages", Electronic Notes in Theoretical Computer Science (ENTCS), Vol. 144, Issue 4, pp. 99 - 116, 2005.

 

* [Abstract] Faizan Javed, "GenParse: An Evolutionary Approach to Context-Free Grammar Induction", The Journal of the Alabama Academy of Science, Volume 76, No. 2, pg 119.

 

Conference

 *NEW* Faizan Javed, Marjan Mernik, Barrett Bryant and Alan Sprague, "GenInc: An Incremental Context-Free Grammar Learning Algorithm for Domain-Specific Language Development”, accepted for publication as a Regular Research Paper (RRP) at the 2007 International Conference on Machine Learning: Models, Technologies & Applications (MLMTA ’07),  Las Vegas, NV, 2007

 

* Faizan Javed, Marjan Mernik, Alan Sprague, and Barrett Bryant, "Incrementally Inferring Context-Free Grammars for Domain-Specific Languages”, Proceedings of The Eighteenth International Conference on Software Engineering and Knowledge Engineering (SEKE'06), July 5th - July 7th,  pgs 363 - 368, San Francisco, CA, 2006.

  [pdf]

 

* Marjan Mernik, Goran Gerlic, Viljem Zumer, Barrett Bryant, "Can a parser be generated from examples?" , Proceedings of 18th ACM symposium on applied computing, SAC 2003, pp. 1063-1067. [pdf]

 

 

Workshop

 

*  Matej Crepinsek, Marjan Mernik, Barrett Bryant, Faizan Javed,   Alan Sprague, "Inferring Context-Free Grammars for Domain-Specific Languages", To In Proceedings of Fifth Workshop on Language Description, Tools and Applications, J. Boyland, G. Hedin (Eds.), pp. 64 - 81, 2005, [paper [ps], presentation [ppt] ]

 

 

Other Publications

 

* Matej Crepinsek, Marjan Mernik, Barrett Bryant, Faizan Javed, and Alan Sprague,, "Context-Free Grammar Inference for Domain-Specific Languages", Technical Report UABCIS-TR-2006-0928-1, UAB, 2006 [ps].

 

 * Faizan Javed, "Learning Context-Free Grammars for Domain-Specific Languages”, Doctoral Symposium, Object-Oriented Programming, Systems, Languages, and Applications (OOPSLA 2005), San Diego, CA, October 2005.

 

*  Matej Crepinsek, Marjan Mernik, Viljem Zumer, "Extracting Grammar from Programs: Brute Force Approach", ACM SIGPLAN Notices, Vol. 40, Issue 4, pp. 29-38, 2005.

 

*  Matej Crepinsek, Marjan Mernik, Viljem Zumer, Faizan Javed, Barrett R. Bryant and Alan Sprague, "Extracting Grammar from Programs: Evolutionary  Approach", ACM SIGPLAN Notices, Vol. 40, Issue 4, pp. 39-46, 2005.

 

* Marjan Mernik, Matej Crepinsekj, Goran Gerlic, Viljem Zumer, Barrett Bryant, Alan Sprague, "Learning Context-Free Grammars Using an Evolutionary Approach", Technical Report, University of Maribor and University of Alabama at Birmingham, 2003. [ps]

 

 

 

[Related Work]

 

The MARS Project:

 

 Domain-specific modeling (DSM) is a resource-efficient and expeditious alternative to the traditional software development process, which traditionally involves a series of mappings from the domain idea, to the domain models and the source code. DSM enables subject matter experts to describe the essential characteristics of a problem at the same level of abstraction with the domain itself.  The conventional approach to DSM involves creating a metamodel for a specific domain, from which instances pertaining to specific configurations of that domain can be constructed. From the defined models, software artifacts like design documentation and source code can be generated. However, as the metamodel undergoes evolutionary changes, repositories of instance models can become orphaned from their defining metamodel. This can result in instance models, which can contain important domain knowledge, failing to load into the modeling tool due to version changes that have occurred to the metamodel. In the model-driven software engineering realm, this problem highlights the need to have the capacity to recover the design knowledge in a repository of legacy models.

 

We propose MARS, a semi-automatic inference-based system for recovering a metamodel that correctly defines the mined instance models through application of grammar inference techniques. We leverage the fact that a correspondence exists between the domain models that can be instantiated from a metamodel, and the set of programs that can be described by a grammar. MARS has been successfully applied to various diverse domains with satisfactory results.

 

Project website: http://www.cis.uab.edu/softcom/GenParse/mars.htm  

 

http://www.cis.uab.edu/softcom/GenParse/mars.mht (single file webpage version)