The original challenge was to understand what people thought (based on Twitter discussions) of the US Republican Debate that took place in Cleveland. The dataset consist of an annotated 20,000 randomly selected tweets from that night using #GOPDebate and related hashtags. The annotation was carried out by contributors based on the following guidelines:
- Is this tweet relevant and from a person? not from news outlet or a brand
- What candidate was mentioned in the tweet? this include “no candidate option”
- What subject was mentioned in the tweet? several options provided
- What was the sentiment of the tweet? +ive, -ive and neutral
Questions of interest:
- What issues resonated with voters?
- Which candidates were viewed most negatively?
- Are people really considering voting for a well-monied, sentient toupee?
- Using machine learning to analyze the GOP debate
- What the GOP Debate taught us about machine learning