Inferrer is an automata learning library written in Python. The library can be used to learn the grammar of a regular language from a given set of example strings or by systematically observing the behaviour of a target automaton and matching its behaviour. The library implements

• E. Mark GOLD's algorithm,
• The Regular Positive and Negative Inference (RPNI) algorithm,
• Dana Angluin's L* algorithm, and
• The NL* algorithm.

If you do not have pipenv installed on your system, then run the following:

sudo pip install pipenv 

The Pipfile specifies Python 3.7 as the development version, but the library should work for any Python 3 version, merely change the python_version value in the Pipfile to the Python version on your system.

If you want Inferrer to show you the DFA that was inferred by the different algorithms, then you need to install Graphviz on your system. The following will work:

• Ubuntu: sudo apt-get install graphviz
• Arch: sudo pacman -Syu graphviz
• OSX: brew cask install graphviz

Clone the repository, install the dependencies and then run the unit tests

$ git clone https://gitlab.com/steynvl/inferrer
$ cd inferrer
$ pipenv install
$ pipenv run python -m unittest discover

Since Inferrer is a library, you can import and use it in your own code, but the library is also exposed via a CLI:

$ pipenv run python cli.py --help

Documentation for the library is located in docs/public/index.html.

We want to try and learn the language A, where A is a binary string containing 101 as a substring. In the resources directory, there is a file positive_01.txt, this file contains positive example strings and a file negative_01.txt, which contains negative example strings.

We are going to try to learn this regular language using the RPNI algorithm:

pipenv run python cli.py resources/positive_01.txt resources/negative_01.txt rpni --show-dfa

RPNI then builds the following DFA

which is the correct grammar and thus we have successfully inferred the regular language A. This DFA is then converted to the following regular expression:

(11*00|0)*(11*(0(1(1|0)*)))

We want to try and learn the language L, where L is a language over the alphabet {a, b} where each string in L contains an even number of a's and an even number of b's.

We are going to try to learn this regular language using the RPNI algorithm:

pipenv run python cli.py resources/positive_02.txt resources/negative_02.txt rpni --show-dfa

RPNI then builds the following DFA

which is the correct grammar and thus we have successfully inferred the regular language L. This DFA is then converted to the following regular expression:

((ba(aa)*b|a)(b(aa)*b)*(b(aa)*ab|a)|ba(aa)*ab|bb)*

• Learning Regular Sets from Queries and Counterexamples
• Angluin-Style Learning of NFA
• Grammatical Inference: Learning Automata and Grammars