Skip to content
Political Adverts Analyzer
Jupyter Notebook Python
Branch: master
Clone or download

Latest commit

Fetching latest commit…
Cannot retrieve the latest commit at this time.


Type Name Latest commit message Commit time
Failed to load latest commit information.

Russian Propagenda Facebook Political Adverts Analyze > Computational Social Science

Research Question:

“Are negative adverts more likely to be successful in Propaganda? “

How effective is an advert in relation to its positivity (negativity)?

How do we measure effectiveness?

We have access to AD_Click and AD_Impression in our data and we can calculate Click Through Rate measure to compare the success or effect of the advert. CTR = ClicksImpression

To measure campaigns’ success based on CTR plus Amount of Money spent (In other words, which campaigns were more affordably successful).

How to measure negativity?

We find a dictionary for labeling positive/ negative words or do supervised labeling to find negativity index for each row of our data.

Finding a correlation between negativity Index vs effectiveness?

Dictionary method for labeling using: SentiWordNet.

  • Esuli, Andrea and Fabrizio Sebastiani. “SENTIWORDNET: A Publicly Available Lexical Resource for Opinion Mining.” LREC (2006).

codes for using the dictionary:

Alternative Algorithms:

Too Naïve:

Java code:


Compare the results for different groups based on followings and age. Does any obvious difference exist?

Relate the Positivity/Negativity with the demographic measures.

Challenge: Missing data exists in our columns; which is observable.

We must include the Demographics for clustering. AD_TARGETING_INTERESTS, AD_TARGETING_EXCLUDED_CONNECTIONS


Age Platform (Instagram/ Facebook)

Which issues were more important for the Americans in comparison to topics for Russians Group?



All code is licensed under MIT License .

You can’t perform that action at this time.