Immerse (#1): Playing with Twitter


By Vishal Kataria


Immerse, the institute research magazine, is the annual science publication of The Fifth Estate. It stands for IIT Madras Magazine on Research in Science and Engineering, where the endeavour is not only to showcase some of the recent developments in research and innovation at IIT Madras, but also to communicate the science behind them in the simplest way possible for better understanding and appreciation. In this series we publish each of the articles from Immerse 2017. You can browse through the previous editions here.


Twitter is the new global parliament. Everyone is a participant, and governments and world-influencers will be well served with a tool to study and shape democratic thought. Twitterplay is one such framework, which aims to be a control room for Twitter.


In this modern age, Twitter has become the place to be for world-influencers. From inciting revolts and planting the seeds of political revolution to managing election campaigns having far-reaching effects, Twitter is the new-age forum for real-time democratic debate. With over 600 million tweets sent out every day, Twitter is increasingly becoming representative of the views of the people of the digital, connected world.


Why Twitter, though? Why not Facebook, which has more users and greater variety of content? Or Reddit, home of the internet’s most intense and in-depth discussions? That’s because on these social networks, people present to the world the faces they want to. Posts and comments can be edited, altered and even deleted. However, Twitter is the closest one gets to a global real-time public conversation. People post unadulterated and personal views, knowing unapologetically that one wrong opinion will probably get buried in the huge swathe of tweets that overwhelms everyone’s feeds. Twitter is becoming the history textbook of the modern age, a log that is updated by millions of people writing about not only the big things of the world, but also the little things in their lives that only a few people are interested in. Want to know what Beyoncé wore to the Grammys? Or exactly what Barack Obama said at the State of the Union address? Look no further than Twitter, where someone or the other has documented each of these in painstaking detail.


Given how much of a tool and power social media services are becoming, it is imperative that they be put to the right use. Groups of people, organisations, and even governments seek control and domination. The notion of somebody having control over a publicly accessible feed may be autocratic, but this very autocracy could be a vital tool in preventing violence and sedition across the globe. This is where Prof Janakiram, of the Department of Computer Science and Engineering comes in.


What Prof Janakiram does is build a cutting-edge control room for Twitter. Equipped to handle all the tweets you can throw its way, it works and analyses them in real-time. This is no mean feat, keeping in mind that 600 million tweets are sent out every day. One can also track the tweets based on hashtags used, which enables identification of daily trending topics. Beyond hashtags, one can also track the personalities that are trending. Twitter does provide a list of trending hashtags and people, but use of this control room enables the tracking of trends as per one’s specific requirements. The Twitterplay framework, as it is called, also counts the number of tweets by country of origin and language tweeted in. As we speak, support for new languages and greater sensitivity to geographic location is being built into the framework.




This tool has many purposes. One could manage a marketing or advertising campaign. If you’re running public relations, Twitter is a place to address issues online before they balloon into disasters for the respective companies. Speaking of campaigns, an entire election campaign could be run via Twitter. If a certain humanitarian cause is to be highlighted in a certain region, or to a set of people, Twitter can serve as a method to identify the subset of population more likely to help with the cause.



Other functionality includes the ability to measure the sentiment of the incoming tweets. Tweets are classified as positive, negative or neutral. Although this technology is not as sophisticated as it will be in the future, it does a good job when averaged over the entire population. It’s surprising (or maybe not?) that we see general human behaviour from the sentiment data. On an average day, it’s observed that around 10% tweets carry positive sentiment and 5% of the tweets carry negative sentiment, with the rest being neutral. Inputs from other social media are also included, such as Google Trends and Instagram. One can imagine that in the not-so-distant future, such a control center would integrate and intelligently collate data from a plethora of media. That too is happening with the Twitterplay framework, with Google Trends and Instagram data being added in the latest version.


Social physics is the new buzzword, entailing the study of the way people interact socially. To extend the analogy to Twitter, each tweet has a velocity, which depends on the relevance and quality of the tweet, the hashtags used, and the number of people it reaches. There also exists an ‘escape velocity’, which if reached, ensures that the tweet reaches some sort of criticality, and then in all likelihood goes viral. Do you see a lot of physics words here? That’s because there is a lot of physics involved, and the study of social physics brings together the two exciting fields of social media study and statistical mechanics. However, as with all theories and models involving human decision-making, this too is a slave to the sometimes arbitrary sensibility of people’s everyday decisions. A human may choose to drink Pepsi today and not Coca-Cola; there is no way this seemingly random and whimsical choice can be modeled by logical rules.


One can know, from Twitter, the state of the world. The collective effort from deliberating parties will be amplified in the process of online expression of opinions. Systems involving people around the world can become more responsive and democratic. We already see that tweets to concerned authorities are in many cases the fastest way to get things done. Feedback on Twitter is near-instantaneous, and everyone will be interested. This holds true even for the important decisions that must be made, such as policymakers deciding on economic, foreign or environmental policy.


Twitterplay is still in a nascent condition, but is already quite powerful. About to be launched, it is merely an aggregation tool as of now, and future systems and frameworks may develop complicated machine-learning algorithms that will learn from the data that is currently being collected and stored. We see ahead of us an interesting phase in computer science, where deep learning techniques are slated to revolutionise multiple fields. We’ve seen this year a neural network based AlphaGo beat the best in the world at Go, a game requiring high skill and intelligence. Slowly, Twitter analysis will be coupled with deep learning systems, and we will see a spate of bots that post on Twitter.


Governments and organisations can use their analysis techniques to automatically engineer people’s sentiments on Twitter and other platforms. We may even soon end up in a future where one cannot distinguish between statements made by actual humans and artificial intelligence systems! “And there we will enter into a new world, a different paradigm”, says Prof Janakiram. There are multiple sides to this issue, but there is no doubt that we are headed there. While we’re at it, it becomes very important to start off this increasingly seminal field of social physics with a strong base, and Twitterplay is doing just that. It’s step 0 in heralding a new revolution.





Meet the Prof

Prof Dharanipragada Janakiram is a professor at the Department of Computer Science, IIT Madras. He heads the Distributed and Object Systems Lab, and builds large scale computing systems for cloud and grid computing technologies. He has recently acquired interest in the study of social media platforms from a computational perspective, believing that unlike some of his other abstract pursuits, this is a field that is very real, and relevant to most people in today’s world. He has received grants and awards from Yahoo! and IBM over the course of the last ten years to pursue his studies. Prof. Janakiram has also written a book on grid and cloud computing that has been published by McGraw Hill.


Meet the Author

Vishal Katariya is a fourth year B.Tech. student in Engineering Physics. He is interested in quantum physics among other things. He likes reading, writing, music and tennis, which includes a fanatical support of Roger Federer.

Write a Comment

Your email address will not be published. Required fields are marked *