THE GenParse PROJECT
Grammar inference for Domain-Specific languages
[Last Update: June 26, 2007]
[ Overview ]
One of the open problems in the area of domain-specific languages is how to make domain-specific language development easier for domain experts not versed in a programming language design. Possible approaches are to build a domain-specific language from parameterized building blocks or by language (grammar) induction. Many well established techniques exist for inferring regular languages. However, inferring context-free grammars (CFGs), which are more expressive and powerful than regular languages, is still an open research problem. Our approach to inferring CFG's previously made use of the genetic programming (GP) paradigm. Preliminary work using grammar-specific heuristic operators in tandem with non-random construction of the initial grammar population resulted in successful induction of small grammars. Our current focus is on incremental grammar learning.
Our project aim is to research on methodologies of CFG induction under various constraints (use of positive or negative samples, or both) limited not only to the GP model of computation, but also open to investigating other models of grammar inference like exploring the use of data mining techniques in grammar inference, the brute-force approach (augmented with heuristics) and investigate better algorithms and heuristics to incrementally construct grammars, all with the hope of being able to infer real world grammars.
This project is a collaboration among:
Lab, Department of Computer & Information Sciences,
Laboratory for Computer
Architectures & Languages,