us / dec-tweet-clustering

Unsupervised tweet clustering with DEC(Deep Embedding Clustering).

Unsupervised Tweet Clustering with DEC

The aim of this project is to clustering the unlabelled tweets with dec and I inspired by: "Unsupervised Deep Embedding for Clustering Analysis" (Xie et al, ICML 2016).

Before the dec I applied natural language processing techniques to preprocess to json type data. Then I use dec and I use the dec because that has more accuracy on clustering.

File Structure

.
+-- tweets.json	
+-- hey_thats_your.py
+-- keras_dec.py
+-- requirements.txt
+-- README.md
+-- .gitignore

Prerequisites

If you want a virtual enviroment, firstly create it:

python -m venv virenv

than activate:

source virenv/bin/activate

finally install packages:

pip install -r requirements.txt

Usage

Run:

python hey_thats_your.py

If we want to run on CPU we set the local variable set CUDA_VISIBLE_DEVICES=-1 then run it otherwise run on GPU.