Emberosaurus

Emberosaurus is an LM-based thesaurus for the English language. It uses WordNet lexical database and the spaCy NLP library to generate synonyms for words and then ranks them within the given context using Ember, a Sentence Transformer model.

Credits

This project is mostly a rework of Bertosaurus. The main difference is that Emberosaurus uses a different LM and updated models in general, which should make it more accurate.

Requirements

• You will need a recent-enough version of Python 3 installed on your system. The code was tested with Python 3.11.5 specifically.

• About 2 GB of free disk space and at least 2 GB of RAM are required to run the program (can be reduced by using a smaller spaCy model).

Installation

You can either download the code manually or clone the repository using git, like so:

git clone https://github.com/marcodsn/emberosaurus.git

Next, you need to install the required dependencies. You can do this by running the following command in the root directory of the project:

pip install -r requirements.txt

You will also need to manually download the spaCy model that will be used by the program. You can do this by running the following command:

python -m spacy download en_core_web_trf

Usage

CLI

To use Emberosaurus, you can run and interact with it directly from the command line, like so:

python cli.py

When prompted, enter a sentence and a word from that sentence that you want to find synonyms for. Here is an example:

Input a sentence: The quick brown fox jumps over the lazy dog
Input a word: quick
 
fast (0.517)
speedy (0.514)
immediate (0.506)
straightaway (0.493)
ready (0.491)
agile (0.488)
prompt (0.487)
spry (0.477)
nimble (0.473)
warm (0.462)
flying (0.459)
Input a sentence:

The program will then print out the available synonyms for the given word, along with their Ember scores (the closer to 1 the better).

To exit the program, simply press Ctrl+C.

API

You can also run Emberosaurus as a web service. I included a simple Flask app that you can use to do this. To start the server, run the following command in the root directory of the project:

python server.py

A simple client is also included, and it can be run like so:

python client.py

The context sentence and the base word are hardcoded in the client, as it is only meant to be used as an example. Modify the code to suit your needs.

License

This project is licensed under the MIT License. See the LICENSE file for details.