A sample feed that the program may be applied on
Ever been in a bad mood, and then see something that makes you feel even worse? Or been happy as can be, but suddenly read a post that brings you crashing back down? Given the hectic and random nature of social media, we're guessing you've felt this way before- and that's exactly what inspired us to create Feeltr!
What it does
Feeltr lets you filter out certain posts on your timeline based on what general mood you select. If things aren't going so great, you'll only see positive things while you have the filter going. If you are feeling particularly good you'll see everything, but be given a heads up when something potentially upsetting is flagged so you can chose to skip that post. And of course, it can just be a normal day same as ever and you can see your feed as is in the middle.
How we built it
We check for valid urls and text posts from social media, scrape the relevant data from the site, run it through the NLP API and filter posts depending on what it returns and the selected mood.
Challenges we ran into
A lot of the challenges came in the projects conception, namely how much negativity do we consider to be stressful to a person? Why not just remove every negative post on your feed so no one ever feels stressed? What do we do about neutral posts? At the end of the day, we ended up trying to be involved least as we could, and let the user make the decision of whether they want to see negative things or not, and let the API determine what 'negative' means.
Accomplishments that we're proud of
Using the NLP stuff was really cool, and implementing it in a way that could potentially help somebody feels really nice. We're also feeling good about how nice the front end looks, and in contrast how solid the base is, our back end!
What we learned
Overall, we learned a lot about the NLP API and how to use it, but we learned a lot more outside of the technical scope. We learned how to quickly adapt and come together as a team, even without knowing one another, and trusting everyone to do their part. Of course, we also learned about how to deal with deadlines and managing our time efficiently as well!
What's next for Feeltr
The next steps would be to integrate it with existing social medias, create a unique interface for the filter, dig deeper into its use of nlp and refine that, and bring it to mobile!