extract domain thesaurus automatically

pip install DomainThesaurus==1.2.3




DomainThesaurus is a python package offering a technique of extracting domain-specific thesaurus which is commonly used in Natural Language Processing. Here is one item of generated thesaurus:

{ "internet explorer":
"synonym":["internet explorers", "internet explorere", "internetexplorer"],

Except for domain-specific thesaurus, the package also provides several useful modules. For example, DomainTerm for extracting domain-specific term and WordDiscrimination for discriminate words (e.g. abbreviation, synonyms). Details of the implemented approaches can be found in our publication: SEthesaurus: WordNet in Software Engineering. (IEEE Transactions on Software Engineering 2019)

Domain-Specific term

DomainTerm can automatically extract domain-specific terms from domain corpus. For example, Javascript in the domain of computer science and technology and karush kuhn tucker in domain of mathematics.

Abbreviations and Synonyms

The module WordDiscrimination can divide semantic-related words into different types. The default module can recognize semantic-related words as abbreviation and synonym. Note that, in our module, the synonym means that two words are semantic-related word and they are morphologically similar. For example, ie is the abbreviation of internet explorer and javascripts is the synonym of javascript.


DomainThesaurus is tested to work under Python 3.x. Please use it in Python 3.x. We will try to support Python 2.x.

Dependency requirements:

  • gensim(>=3.6.0)
  • networkx(>=2.1)

DomainThesaurus is currently available on the PyPi's repository and you can install it via pip:

pip install DomainThesaurus

If you prefer, you can clone it and run the file. Use the following command to get a copy from GitHub:

git clone


A simple example::
>>> dst = DomainThesaurus(domain_specific_corpus_path="your domain specific corpus path",
>>>                       general_vocab_path="your general vocab path",
>>>                       outputDir="path of output")
>>> # extract domain thesauruss
>>> your_thesaurus = dst.extract()

If you don't have any datasets, you can copy and run this code: . This code will automatically download datasets for you. The code design is flexible, you can replace the default function class with your own function class to get better performance. You can find more usage in


In this project, we use Levenshtein Distance and GoogleDriveDownloader from and Thanks for their code.