We see old tweets or old posts coming up all the time to haunt us. A few days ago, a reporter was criticized by Carolina Football Quarterback Cam Newton, a few days after Cam Newtons backlash, the reporter's old tweets were dug up of her using racist comments. These comments have put the reporters job at stake, and even the company that has hired her. Our team sought to eliminate this problem.
What it does
Unwantd parses through all the social media and public information the user has available and targets things that may cause a flag later in that person's life. Were utilizing Microsoft's Azure technology to really take parsing information to the next level. Not only do we do a full Bing search of the user's username, but once we get results, we use a basic natural language processor to parse the text and Microsoft Azure's image recognition API to scan the images for red flags. The web app can handle ton's of data and the results of it are things that we recommend the user remove from their profiles.
We have 3 use cases:
Susie is a hiring manager at Microsoft and she has screened Steve and deemed he is a great candidate for the job. Susie runs Unwantd on Steve's profiles to verify he doesn't have any baggage that could potentially harm the company later. Susie runs the software and it returns that Steve is posting images of supporting a hate group and the group goes against all of Microsoft's company values, so Susie decides not to hire someone that goes against her company goals.
Samantha is trying to find a job, but she is not sure if her social media is appropriate. Samantha runs Unwantd and the results returned that she has 4 inappropriate images that show her smoking marijuana on her instagram. Samantha removes these images prior to her job hunt.
Alex is star quarterback at UC Berkeley and he is going to get drafted by the Los Angeles Chargers. Alex wants to remove all of his dumb inappropriate college posts before making it big, so none of them come back to haunt him when he's famous. Alex also has 12,000 tweets, 15,000 Instagram posts so he doesn't have the time to go through them 1 by 1, and he can't delete his accounts, or else he will lose his followers. So Alex runs Unwantd to discover these posts for him and protect him from future problems.
How we built it
This project would not be possible without 3 major things.
Microsoft Azure's custom computer vision for image recognition API - This API allows us to filter out things like alcohol or drugs through a users Instagram or Facebook.
Microsoft's Bing - Bing lets us pull up other accounts from the users, just based off one username. Bing's API returns us more results so we can better dig into the user's background.
Our makeshift natural language processing algorithm - This algorithm basically extracts key words from a posts and understands if its a bad posts or not.
Challenges we ran into
We ran into some web socket problems and some API restriction problems. We solved the socket problem by using a well known library and just digging into the documentation and we solved the API restriction by consulting the Microsoft Team.
Accomplishments that we're proud of
We're extremely proud we were able to finish the project and we're also proud of our custom trained image recognition tool.
What we learned
We all learned tons about Vue.js as well as how to use sockets with Vue and Express.
What's next for Unwantd
We're going to deploy Unwantd to a Amazon Web Server to allow anyone to use the service for free. We may add some ads on the webpages to pay for AWS fees.