Inspiration
I read about several projects that used a twitter sentiment analysis algorithm to determine opinions and geographic trends. Why not do this for just feelings?
What it does
This website uses the Twitter API and a simple natural language processing query (with keywords) to analyze not general sentiment (positive or negative), but specific emotions. These emotions are mapped onto a map of the US, and emojis and color are used to delineate between different areas of emotions.
How I built it
This is a website hosted on a Tomcat server on my laptop. It's built with HTML and CSS, and the JavaScript does the sentiment analysis. The Twitter API is used to stream and search tweets within the geographic borders of the US, and then this data is mapped visually onto a JSON of the USA.
Challenges I ran into
- Remotely accessing and referencing pubnub libraries
- Choosing words for the NLP that would not have double-meanings (for example, the word "damn", which typically means frustration or anger, but when I had included this word, a bunch of tweets of the "Damn Daniel" video came up)
- Getting the emojis to fit the visual space I allotted for them
- Titling this hack
Accomplishments that I'm proud of
I'm happy that I have a functional project at the end. I'm proud of the fact that I was able to problem-solve with mentors and I actually understood what the heck they were saying.
What I learned
How to API, how to host on a local web server, how to use JSONS
Log in or sign up for Devpost to join the conversation.