The internship was under the guidance of following eminent people:
- Dr. Kiran Garimella, IDSS, MIT, USA
- Prof. Aaditya Dar, ISB Hyderabad
- Dr. Vasundhara Sirnate, The Polis Project
Work done by Siba Smarak Panigrahi in the month of Dec 2020 - Jan 2021
The following description is extremely brief. To use this framework, look through the codes or read this .
- Creates a comprehensive dictionary of posts from a set of .txt files
- Can perform basic analysis on variation of posts of time, reactions, captions, and trending posts
- Create classifiers (currently Logistic Regression - you can very easily switch it 😅) for
- top 8 (as per the posts released) contesting parties
- any two contesting parties
- MGB and NDA
- Topic modeling and visualization of topics of any subset posts. Helper functions to ease topic modeling are available for
- posts which are not related to any party
- posts related to any particular party
Note: The work was done with the aid of Google Colaboratory Notebooks and free GPU resources available.
Ans: Definitely. But it is kind of rough and a bit muddled up. Still, if you wish to refer the Colab notebook (at your own risk 😅 😛), here you go: Link. If you find discrepancy with the final functions available here, then, I am really sorry, I have warned. Colab is rough work mostly - but provides a fair idea how to use the functions available in the framework!
2. Hey, I have an important query. I want to use your functions in framework directory. But, where is the data? Is it publicly available?
Ans. Hey! It is really wonderful to hear that you want to use my framework, to analyze and design classifiers for FB posts released during elections! 😍
I am extremely sorry 😥. The data could not be made public since the work is in progress. Hope you understand!
But, we have the shared the framework, so yay 😉🤩!
Ans: Do contact the author of the code!