jbhong / twitter-rumour-detection-

Twitter Rumour Detection

A Twitter rumour detection system using various machine learning models (tested using PHEME dataset)

Files contained in this repository:

1. feature_extraction.py
   Extract feature matrix from raw twitter dataset and store in a csv file "dataset.csv"

2. scikit-learn.py
   Read "dataset.csv", apply machine learning models and report the performance of testing set, plot confusion matrix

3. Gaussain Naive Bayes.py
   A Gaussian Naive Bayes model that read "dataset.csv" and implement training, testing and report performance.

4. dataset.csv
   Extracted features from dataset. The outcome of feature_extraction.py

5. Report
   Rumor_detection_report.pdf contains the report of this project including the experimental results.

Software requirement:

Python 3.0 or higher version

Library requirement:

pandas seaborn matplotlib sklearn nltk json

Library installation (based on Windows):

1. In windows command prompt
   

2. Navigate to the location of the pip folder
   

3. Enter command to install NLTK
   
   pip3 install nltk
   
   

4. Installation should be done successfully
   

5. Other libraries have similiar way for installation
   

PHEME Dataset Download is available at:

https://figshare.com/articles/PHEME_dataset_of_rumours_and_non-rumours/4010619

Usage:

1. Download PHEME dataset and unzip the file, put the file together with scripts and add them into current folder in Python.
   

2. Install relevant libraries through command prompt or Python platform.
   

3. Specify the dataset that desired to analyse in feature_extraction.py and run feature_extraction.py.
   

4. Run scikit-learn.py or Gaussain Naive Bayes.py to implement classification and get results.