twitter_data_analysis

The network analysis notebook does the following tasks:

1. Use Twitter API to collect 1000 tweets in which keyword ‘narendra modi’ appears, save the collected tweets in nm.txt

2. Convert the collected tweets into BoW vectors and find cosine similarity of a pair of tweets and print the top-10 most similar tweet pairs, print these pairs

3. Do the same using TF-IDF vectors

4. Find out unique users (N) who have posted these 1000 tweets. (N <= 1000) U1, u2, ….., uN

5. Find the followers and followee of each user from the N users obtained before

   • u1 - [followers list] [followee list]
   • u2 - [followers list] [followee list]
   • .
   • .
   • uN - [followers list] [followee list]

6. Followers and followees are also users, so create a follower-followee directed graph among them, G. (ui → uj) iff ui is followed by uj

7. Find popular users in this G based on

   • Degree centrality
   • Betweeness centrality
   • Closeness centrality

Run notebooks on local machine:

Create a twitter developer account from here

cd to the repository and create a virtual environment and activate it:

virtualenv -p usr/bin/python3.8 env
source env/bin/activate

Install the libraries from requirements.txt:

pip install -r requirements.txt

Create a file by the name secrets.py and copy the content of secrets_template.py to secrets.py. Add your secret keys, which you got by creating an app on twitter developers account, in secrets.py file.

Run jupyter notebook to run all notebooks inside this repository locally.

---