Our work "Maximal Format-Free Data Repair" has been accepted for publication at ASE 2026. This is a joint work by Jack Luo, Xi Wu, Hong Jin Kang, Alan Fekete, and Rahul Gopinath (me). It shows how to repair syntactically rich text data with minimal effort even when the syntax specification is unavailable. It combines passive and active blackbox grammar inference to achieve this feat.

You can read it here.

#SoftwareEnginnering  #DataRepair #GrammarInference

A tutorial of the #TTT algorithm for inferring regular input grammars using active membership queries. I had posted the L* algorithm (active grammar inference using membership queries), and the RPNI algorithm (passive grammar inference using examples) earlier.

This is part of my ongoing effort to document various algorithms relating to grammars including various algorithms for parsing, random sampling of fixed size strings, and grammar fuzzing.

Note: Pyodide takes a little time to initialize, but it should be faster to initialize than spinning up the binder service from Jupyter (but slower to execute).

https://rahul.gopinath.org/post/2026/06/09/ttt-grammar-inference/

A Simple Runtime Invariant Miner: A tutorial on how to build a Daikon like program invariant miner for Python.

A tutorial of the #RPNI algorithm for inferring regular input grammars when given only accepted and rejected examples (positive and negative examples). I had posted the L* algorithm before.

This is part of my ongoing effort to document various algorithms relating to grammars including various algorithms for parsing, random sampling of fixed size strings, and grammar fuzzing.

Note: Pyodide takes a little time to initialize, but it should be faster to initialize than spinning up the binder service from Jupyter (but slower to execute).

https://rahul.gopinath.org/post/2025/10/24/rpni-learning-regular-languages/

I am at São Paulo attending ISSRE'25If you are around, please come, say "Hi" :)