Our research group deals with major problems in diversified areas of computational sciences that can be inherently characterized by large scale systems of interacting entities, suitably represented as complex networks. Presently, we are involved in identifying and solving problems related to online social networks, peer-to-peer and lingustic systems.
Online Social Networks (OSNs) such as Facebook, Twitter and folksonomies such as Delicious and Flickr are now among the most frequently accessed sites on the Web. There has been an exponential rise in the number of users and amount of activity in OSNs in the past few years, and the popular OSNs and folksonomies have millions of users at present. Several practical issues related to OSNs are under active research.For instance, the popular OSNs are currently experiencing problems of scalability and increasing spam as a result of the enormous rise in the number and activity of users. We are studying the theoretical underpinnings of the methods to improve scalability used in OSNs, such as restricting the number of links that users can create, or utilize cloud-computing for better utility of resources. Another important goal of the OSNs is to improve the experience of the users of the networking service; this involves a variety of strategies such as restriction of spam and personalized recommendation of popular resources to users. We are also studying methods of spam-detection based on network properties, and algorithms for recommendation of resources to users in folksonomies.
Understanding the dynamics of large scale peer-to-peer (P2P) networks is an important research area for the network research community. The peer to peer networks are formed mainly as a result of the bootstrapping protocol followed by incoming peers. The bootstrapping protocol finds some online peers (nodes) that are already part of the network and sends the connection requests to them. It has been observed that this methodology of connecting the peers to some “good peer nodes” leads to the emergence of superpeer networks like Gnutella, KaZaA, Skype etc. Thus an important research problem is to understand the self-organizing behavior of these superpeer networks. Moreover, nodes in the P2P network frequently join and leave the network without any central coordination. This churn of peer nodes can partition the network into smaller components and breakdown the communication among peers. In addition to that, stability of the network can get severely affected through intended attacks targeted towards the important peers. Thus, an important research issue is to develop suitable theoretical frameworks to understand the emergence of the P2P networks and enhance their stability and scalability.
Several real world scenarios exist where (i) a network infrastructure cannot be deployed, e.g. under-water, in forests, in outer space, (ii) infrastructure setup is expensive in terms of money and time, e.g. sudden military exercise in border areas, disaster affected areas or (iii) the infrastructure is limited in reach, e.g. remote village. A Delay Tolerant Network (DTN) can provide communication support in such situations. Our focus is in improving the service qualities in DTN, such as broadcasting time, routing time, message delivery delay, message exchange policies, etc in networks with mobile agents. As understanding the inter contact time of the agents is important to the functioining of these networks, one of the major focus is studying the inter contact time distribution between the agents in the network which basically exploits the mobility pattern they follow. Further, we are exploiting several realistic mobility patterns that might arise while handling such critical situations.
Another important research area in the field of linguistics is the study of evolution of different langauges. It is observed that the sound inventories of the world’s languages show a considerable extent of symmetry. It has been postulated that this symmetry is a reflection of the human physiological, cognitive and societal factors. Although the organization of the vowel inventories has been satisfactorily explained for smaller inventories, the structure of the consonant inventories is an open problem since 1939. We reformulate the problem in the light of statistical mechanics and present complex network representations of these inventories. Two types of networks are considered – a language-consonant bipartite network and a consonant-consonant co-occurrence network. The networks are constructed from the UCLA Phonological Segment Inventory Database (UPSID). The complex-network theoretic study of these networks helps us in determining many of its interesting global structural properties.
Traditional graph theoretic approaches have their own limitations and are not applicable to these networks due to their large size and dynamic nature. We borrow concepts from statistical physics and apply them to solve the diverse scientific and engineering problems that arise in these domains.