Blog‎ > ‎

On Multimedia, Social Networks and Computer Science

posted Jun 15, 2013, 8:28 AM by SD Roy   [ updated Jun 15, 2013, 8:33 AM ]

    Hi, my name is Suman Deb Roy. I am glad to serve as the new bulletin editor for IEEE Special Technical Community on Social Networking (STCSN). Below is a not-so-short excerpt of my interests, research, what brings me to STCSN and my thoughts on what makes multimedia in social networks a fascinating research opportunity: 

    I distinctly remember the humid summer evening in 2005, when I found myself reading a paper published in the Mind. I had spent most of the summer convincing myself that there is much more to Computer Science (my sophomore major) than what meets the eye. The first line of the paper really resonated with me. It was an effortless statement - "I propose to consider the question, 'Can machines think?'". Turing's simple proposition was a pivotal stone that laid the foundations of our still young field. However, simple does not imply easy. Six decades have passed since Turing's now rejoiced paper on the Imitation Game was published. We are still pretty far from quantifying what is thinking or consciousness. Computer Science has made huge strides in understanding learning, intelligence and knowledge. However, how the human mind combines all these into the conscious experience is still an unsolved mystery.

    At this point, it is natural for you to wonder what is the purpose of this philosophical introduction. Frankly, what role does multimedia play in the quest to understand the fundamental question raised by one of the founding fathers of Computer Science? Moreover, how does social networks fit into the picture. Let us take a moment and think about 'thinking'. Primarily, 'thinking' is electro-magnetic realities which are often directly or indirectly shaped by external stimuli and which affect our body, mind and actions. Scientists claim that consciousness is critical to thinking. Being conscious is to have unity of experience - a whole, seemingly indivisible and integrated phenomenon. I do not mean 'experience' as a causal series of events, such as when your flight was delayed due to bad weather and you experienced the agony of waiting for hours at the airport. Instead, experience fundamentally refers to description of the world you are sensing.  

    Experience is a really fascinating thing. It is not the isolation of different senses from each other. On the contrary, information from different sources and stimuli are blending together to generate an experience. Remember that different sources generate information/data in different formats, often in combination with one another. So, when you gasp at some movie scene or laugh at your friends joke, you are registering multiple formats of data - audio, video, image, interactivity and probably text too. This combination of content in different formats is what is otherwise known as 'Multimedia'. It is the convergence of the different data formats and types of signal into one that gives us the idea of 'experience'. Signals from multimedia are key to our feeling of experience. Trace back to a few higher abstractions, and we can see its relation to consciousness and thinking. 

    Finally, let us capture the social aspect. How come some books and movies live forever and deeply impact us. They contain plots, visuals, lines that never will die. We are attuned to liking media with certain aspects more than others. In the past, it was impossible to quantitatively analyze this phenomenon at large scale. But now, through social networks and social media we can study both the network and the media in unison. Analyzing shared multimedia in social networks gives us an unique opportunity to address two big questions simultaneously: (1) the property of the network structure and related dynamics of information diffusion, which facilitates spread of some media/memes more than others, and (2) from the other end of the spectrum, what property of the media/content itself makes us more receptive to it. Of course, there is a feedback involved here. Popular media is shared more, and sharing increases its popularity. It is thus not difficult to extrapolate that perhaps one reason humans experience and machines often find it hard to capture social reality is somehow rooted in social multimedia analysis. Multimedia Analysis in Social Networks is important, perhaps because it is the closest we will come to giving a machine the concept of social experience. This does not discount the great strides machine learning has made, nor does it ignore automated chat bots like Eliza. The two previously mentioned questions however have immense philosophical, technological and business implications. The trend is encouraging, for we are seeing an exponential rise in research and development using multimedia in social networks.

    I hope this long winding post resonates with you.  STCSN is a perfect place to explore these ideas, engender new approaches to give machines the power of experience. STCSN combines some very bright minds from both multimedia and social networks communities. As for me, my research involves learning from different multimedia sources/formats (social, video, streams, natural language) for improved understanding of socio-semantic context across domains. More specifically, I am working on exploring the duality between signal processing and social networks, modeling user attention and detecting impending popularity of multimedia using social network signals. You can find more details here.

If you are the author of an interesting article that relates to social networking and deserves to be read, feel free to send it over.