Ranked Activity Streams

Abstract

A common component of social networks today is an activity stream, which is a list of recent activities performed by users. The purpose of an activity stream is to give users an easy way to keep up to date with other users. Activity streams can often grow very quickly and become so large and update at such a fast pace that they become unusable for the purpose they were intended. To ameliorate this problem an activity stream can be ranked, giving each activity a score based on its relevancy, bubbling relevant activities to the front or hiding activities that are below a certain threshold. Doing this will in theory help users keep up to date with activities that are relevant. The aim of this project is to create a system that manages a ranked activity stream called "FanFlow" on the photo-sharing website YouPic.com. The "FanFlow" activity stream consists of photos uploaded by YouPic.com users which a user is a fan of. The rst part implements a system termed an activity engine which routes each photo to the correct "FanFlow". The second part is to implement and evaluate ranking metrics on this activity stream.