Undergrad Research Project - Social media video data mining

Spring 2016

Nikhil Choudhary
Srinivasa Narasimhan, Yaser Sheikh
Project description

Current technology allows stitching together of multiple photos to recreate a scene in 3D (https://www.youtube.com/watch?v=IgBQCoEfiMs). Working together with Minh Vo (PhD student at RI), we hope to reconstruct the 3D world from internet videos by multiple individuals of the same experience (e.g., Paris attack, Boston bombing, etc.). This is the first part of the pipeline that will culminate in a product that allows testing of the algorithm against clusters of raw videos available online.

To support the goal of recreating these experiences, we aim to create a data mining engine that gathers data constantly from social media platforms such as Facebook, Instagram and YouTube. After sourcing a critical mass of videos, we will start with coarse-grain clustering based on time and space. The project will also have a user interface akin to onreel.news that allows for Minh and other researchers who need this data to view multiple videos of the same event and decide on what to feed into the algorithm. Further development of this project will lead to something like https://www.youtube.com/watch?v=ST6KKoHrmso.

Return to project list