Project Title: Using FB Product User Engagement to Model American Party Issue Importance
Executive Summary → Facebook’s importance to politicians, political interest groups, and party activists in the American political environment is at an all-time high. Both parties are regularly scaling up their digital advertising spends and infrastructure, as they are stuck in an equilibrium of escalating campaign digital advertising budgets. As content history from these groups grows, user engagement scales up as well (more posts, more likes, more comments, etc.). Therefore, an interesting research gap would be to understand how issue importance has changed over the last few years as engagements between politicians and constituents, activists and interest groups have gone digital.
Project Proposal →
Use python packages insta-scrape (built on Selenium and for pandas) and facebook-scraper (pandas-friendly) to web-scrape the comments from the five highest engagement posts of the five highest follower count politicians (for both Instagram and Facebook) from both American political parties.
Use a natural language processing model to find the most frequently used terms that get classified in a list as important political issues to find out what the most talked about issues are for each party.
Use a natural language language processing model to perform a sentiment analysis of a random sample of comments from accounts who are either business or verified to attempt to define how party activists and groups feel about a particular issue.
Here is the link to the master comments - https://drive.google.com/file/d/1qoFjdlIiwWPMd3Dsh8aCQnu1CwKE2I0f/view?usp=sharing
On Start_NLP if you can't scrape the IG comments because you don't have enough warm instagram accounts or fb accounts then you can use this file and turn the comments into a nested list segmented by the profileURL to replicate the analysis and tables I made.
Files are run in order of the number in the bracket until you get to Start_NLP.
HMU at marleyrosario@uchicago.edu for any explanation.