We are living in the age of cloud and big data where almost every physical and virtual system acts as a source of data, be it physical devices like smartphones and smart-sensors or virtual communication environments like social media and mass media technologies. Interestingly these data  produced are (socially) connected in nature and discovering/understanding these connections may unfold deeper and useful patterns. For example, barrages of tweets about a particular event may help us to understand the severity of the event. Collective assessment of smartphone sensor readings of commuters may help us in inferring the condition of a road. However, such collective sensing will have its own problems, many of the components (players) may not behave in a proper fashion — spam, virus, frauds and fake entities are common examples disturbing the smooth functioning of such a system. Also maintaining the huge amount of data brings in several system related challenges.

The main objectives that set the agenda for  research  are:

  1. Build innovative services leveraging the concept of collective sensing
  2. Tackle various problems arising due to anomaly in collective sensing
  3. Build up supporting services to seamlessly execute social sensing
  4. Building up several use cases to support social sensing
  1. We aim to develop different services over the data obtained from collective sensing, like, (i) detection of individual and group events, (ii) recommendation of activities, services and products, (iii) prediction of future events, facts and activities, (iv) opinion learning, (v) detection of anomalies in the information and (v) summarization and classification of information.
  2. As data are collected from human sources and devices, therefore the credibility of information and reliability of sources are fundamental issues in social sensing. Therefore our objective is to develop methods for truth discovery and assess the quality of information from the data. Typical examples include  identification of spams, fake reviews, fraudulent actions and various other similar noise items in data which may significantly affect the quality of top level services developed. Further, as data are collected from media or devices which are in possession of individuals, therefore privacy is another concern. In this direction, our target is to guarantee privacy preserving services.
  3. The sources of data as well as the target services will be mostly built over small resource constrained devices, like smartphones, IoT sensors, wearable devices, data centers, backbone networks and so on. Consequently, certain crucial system  related challenges are bound to stem up, like energy usage, storage optimization, computation efficiency, mobility, system management, network monitoring and deployment planning, network policy control, supporting ubiquitous connectivity among various devices and information sources etc.  We plan to  consider such system related issues during the development of the services.
  4. We plan to consider a few focused areas where such services can have immense applications. (i) Disaster Management: For disaster management, timely situational updates are vital for the concerned authorities (e.g., Governmental and non-governmental agencies) to gain a high-level understanding of the situation and accordingly plan relief operations. Information from microblogging/social media sites as well as smart devices can play a major role towards this. (ii) E-commerce: Applications like prediction of items’ popularity, recommendation of items based on user profile, summarization of user reviews and classification of user/product profiles etc. are some of the aspects that we plan to investigate in details.. (iii) E-Health: Wearable devices play a major role in today’s world and several e-health applications are continuously gaining popularity among people, like Google Fit or Apple Health. While extraction of necessary and meaningful information from wearable devices is one objective, preserving the privacy of data is another challenge, which we target in this domain.

Finally, knowledge of different fundamental computing techniques are required to fulfill this broad spectrum of objectives; some of these include (but are not limited to) machine learning, information retrieval, natural language processing, complex networks, network algorithms, data security and privacy etc. The Complex Network Research Group (CNeRG) has expertise in all these diverse domains, and target to focus on such aspects of social sensing with collective efforts and inter-domain collaborations among its faculty members and students.

 

 

Contact Us

Department of Computer Science & Engineering, Indian Institute of Technology (IIT) Kharagpur, Kharagpur, West Bengal, India PIN - 721302