Supervised Multi-Label Classifier

Supervised Multi-Label Classifier is a Python library for multi-class multi-label classification. This algorithm can classify texts into pre trained categories.

Consider the following scenario. We have a set of descriptions with two classifications ny and london

Category	Text
ny	New York is a hell of a town
london	The capital of Great Britain is London
ny, london	New York is great and so is London.

Based on my training set above the classifier would classify below tests as follows

Text	Classification
New York was originally dutch	ny
It rains a lot in London	london
New York and London both have the stock market	ny, london

Prerequisites

 ./scripts/setup.sh

• Installs pre-trained Word2Vec model
• Installs reuters corpus from nltk to run the test

Installation

Use the requirements.txt file to install dependencies

python3 -m venv ./venv
source ./venv/bin/activate  
pip install -r requirements.txt

Run tests

Tests are run using pytest. A simple model test is provided as an example with the csv file in /data_set folder. A more robust test is provided to test the model with reuters_21578 corpus

From the root folder run

pytest -s -r a

Usage

Do it yourself

From the root folder run

• Default mode - Reuters corpus

python run.py

• Pass csv as arguments

python run.py --model "path/to/file.csv"

To run the movie-matching algorithm use data/imdb_dataset.csv

Docker

docker build . -t supervised-multi-classifier

docker run --rm -it \
-v `pwd`/supervised_multilabel_classifier/model:/home/app/supervised_multilabel_classifier/model \
-v `pwd`/supervised_multilabel_classifier/data_set:/home/app/supervised_multilabel_classifier/data_set \
supervised-multi-classifier \
 --model /home/app/supervised_multilabel_classifier/data_set/<your-model>.csv

What can it do?

Let's run the movie algorithm

docker run --rm -it \
-v `pwd`/supervised_multilabel_classifier/model:/home/app/supervised_multilabel_classifier/model \
-v `pwd`/supervised_multilabel_classifier/data_set:/home/app/supervised_multilabel_classifier/data_set \
supervised-multi-classifier \
 --model /home/app/supervised_multilabel_classifier/data_set/imdb_dataset.csv
 
 ...
 
Enter/Paste your content. Ctrl-D or Ctrl-Z ( windows ) to save it.
I kissed a girl, I liked it
prediction: ['Drama', 'Romance', 'Comedy', 'Fantasy']
Most similar: [(0.6800497770309448, 'tt0102250/L.A. Story'), (0.6765267848968506, 'tt0386005/New Police Story'), (0.6711339950561523, 'tt0338763/Battle Royale II'), (0.669927179813385, 'tt0101507/Boyz n the Hood'), (0.6669382452964783, 'tt0258068/The Quiet American')]

The description: I kissed a girl, I liked it

Prediction: Drama, Romance, Comedy, Fantasy

Try it with proper plot synopsys

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.

License

MIT