News Clustering

A simple class project to cluster a sample data-set of 1000 tweets to one of the four classes which are: Sports, Economics, Politics and Culture using K-Means & Mixture Models in Python. The data-set is crawled from Tasnim News Agency and the crawler used for this task is written in Julia language.

Path to training files:

All data files, including trained models are located in the data directory.
Raw data are pre-processed once but can be processed again in case there is new data available.

Instructions:

First of all do a pip install -r requirements.txt to install the required modules. You may need to install modules manually if this does not work as expected.

1. To cluster (train) news using K-Means algorithm, uncomment kMeans.fit() line in main.py module and run it. If you want to predict new data given trained model, uncomment kMeans.predict() line in the same file and run the module.
2. To cluster (train) news using Mixture Models algorithm, uncomment mix.fit() line in main.py module and run it. For prediction of new data, uncomment mix.predict() line in the same file and run the module.