Rivrb's validation system is two-fold. The first step involves the intake of social media data that is generated on Rivrb and attempting to automatically determine its validity based on several key characteristics that define content validation on Rivrb, such as: The quality of a source, the bias, impartiality, credibility, and reputation of actor(s) involved/mentioned in the post.
Our algorithm then outputs an overall confidence level detailing how confident it is that the information contained in the post is valid. If this confidence level falls below a certain threshold, that particular post is escalated to Rivrb Content Validators. Only validators that have expertise in the domain of the post are contacted. These indipendent validators then conduct research to validate the information and submit their findings to the algorithm.
Content Validators' inputs are assessed against one another and then incorporated into our algorithm's decision making functionality. A new confidence score is generated in the process. All these various activities occur asynchronously, so validation scores on Rivrb are constantly being updated as new information comes in.
Our validation system is distributed in order to ensure content is provided to users in real time. An initial validation score is shown to users when content is posted on the app. As the various elements of Rivrb's validation system perform work and provide results to the central decision making algorithm, validation scores on posts are updated.
This is made possible by utilizing AWS Cloud infrastructure to host many small and compact algorithms that perform work to determine a single element of Rivrb's validation criteria. All these elements work in concert to provide a final output that determines Rivrb Validation Scores (RVS). If all of these elements end up being inconclusive, the post is escalated to Rivrb content validators.
Rivrb Content Validators are everyday social media users on Rivrb that have submitted an application to become a researcher within one of Rivrb's subject domains, such as politics, sports, trending topics, etc. Upon selection as a Rivrb Content Validator, said validator will be given an array of various content within their selected domain to research & submit findings on. Validators get compensated based on the accuracy of their research.
As Rivrb Content Validators research more and more content within their selected domain, they can earn Rivrb certified badges that indicate that they are an expert within that domain. These badges will only be conferred to researchers who have submitted quality research on a significant number of posts in their domain of expertise. These badges are a great source of social proof and a measure of domain expertise and can be used/added on resumes and other documents.
During the summer of 2020, amidst Covid-19 lockdown and the George-Floyd incident, a virtual hackthon called "NYC Coders Hack For Black Lives Matter" was hosted by NYC Coders, a Software Engineering meetup group based in NY that brings aspiring software engineers together to help one another prepare for software engineering related interview questions. The goal of the hackathon was to create software engineering solutions that would lead to better racial equity.
Rivrb's fit in this goal of software related solutions to racial equity was to create a software solution that would shatter online echo-chambers by flipping people's social media feeds. The inspiration came from the fact that our CEO, Calvin Goah, noticed that during the 2016 election when he was a Sophomore at Columbia University that his entire Facebook feed was full of posts that greatly suggested that Hillary Clinton was going to win the election. When that didn't happen, he kept wondering how many people had experienced a similar phenomenon where their social feeds were only dominated by content they typically engaged in. Was everyone in an echo-chamber? And if so, why?
This thought persisted until the 2020 Hackathon brought it all full circle. The impending 2020 Presidential Election was rife with outcries of "fake news" on both sides of the political spectrum. The initail hackathon MVP created a simple web app that would return tweets from a user's opposing political party.
Since then, Rivrb has evolved into a standalone social media application that automatically combats misinformation at the source: On Social Media. Rivrb is a social media application that uses artificial intelligence techniques to automatically fact-check content generated on the app. Moreover, content that Rivrb's automated model is unable to accurately diagnose are sent to Rivrb users that have been approved to be Rivrb Content Validators. This two-fold process ensures that content validation is done swiftly and at high fidelity.
Rivrb's vision is to create a more informed online community, and we are going to achieve that vision by creating a social media platform that automatically validates content!