Our goal of this project was to examine public opinion and sentiments about the Ukraine War collected between Feburary 9th to March 31st. We used a following set of keywords to filter the Ukraine war-related tweets: 'ukraine','ukrainian', 'russia', 'russian','putin' 'zelensky'. Also, we looked at how Twitter influencers (e.g., mainstream news channels, verified Twitter accounts) are related to their followers via sentiment and NLP predictions.

We incorporated both supervised machine learning models (BERT) and dictionary-based sentiment analyses (Lexicone) to predict sentiments expressed in each tweet and compare similarity/difference across tweets of different influence networks. We end with graph/network visualizations to understand the structural differences among influential social media influencers.

For details, please see the notebook