Francois de Villiers (MSc-I Computer Science).

Filtering Information Streams

The focus of my research is to make Twitter more user-friendly.  Although it is relatively easy to use, user friendly and brilliant, there is much room for improvement and here is my take on the issue.

About a year ago I myself started using Twitter and the number of people I followed increased monthly as expected.   This was for the reason that I constantly found new and interesting people who I wanted to follow.  Although my time spent using Twitter has increased over the last while, I have not gained more value from the stream of tweets.  After deciding to review the people I was following more closely I saw that none of the people I follow could be removed.

The next move I had to make was to categorise people into groups as it is evident that there are clear boundaries between the reasons I follow people.  I follow my friends for social information, and separately I follow industry experts who give me their expertise as well as other forms of acquaintances.  I sorted out the people accordingly that led me to be satisfied for only about a week.

The realization came to me that my victory was prematurely celebrated.  As a number of the industry experts I follow on Twitter also post social tweets about their lives, this was the reason for my short lived happiness.    This was information that was not valued by me in my Twitter stream.  A question popped up in my mind asking how noise  can be removed  from a user’s stream according to his preferences.

This lead to my focus point of research that is based in the field of topic detection and tracking.  The focus of my research therewith is to categorise different tweets into topics, by making use of recommendation techniques to infer a user’s preference to view a specific tweet.  The removal of all tweets from the user’s stream that he may not find valuable became my goal and thereby resulting in noise being filtered from a twitter stream.

Read Francois’s blog posts
Join Francois in 2012 at the MIH Media Lab 

My contact details: fdv@ml.sun.ac.za