Information and Decision Systems (IDS) research group conducts a broad range of research in analytical information and decision sciences
We focus on the effective and efficient collection, storage, retrieval, analysis and use of information and data, and subsequent computational and analytical techniques to automate and inform decision-making. IDS is based at the UCL Centre for Artificial Intelligence.
Our current research interests are:
- Web Intelligence, Information Retrieval, Conversational Systems and Natural Language Processing for the Web
- (Multi-) agent learning, game theory, machine decision making, and reinforcement learning
- Complex networks - communication, information and social networks
- AI for health
1) Web Intelligence, Information Retrieval, Conversational Systems and Natural Language Processing for the Web
Our research focuses on information retrieval and conversational search, dialogue systems, and applications of machine learning and natural language processing, primarily focusing on problems associated with the Web. Our research interests include user modelling and personalisation, inferring users’ interests, and recommendation based on the contents they have visited in the Web, search, conversational search, task completion assistants, fact checking and automated misinformation detection, fairness and bias in online systems.
Please contact Emine Yilmaz for more information.
2. (Multi-) agent learning, game theory, machine decision making, and reinforcement learning
Artificial Intelligence (AI) aims at creating a computerised system capable of acquiring and applying knowledge and skills as a result of experience. Perception and interaction are two key pillars of intelligence. It is widely accepted that the former corresponds to pattern recognition tasks, while the latter is related to machine decision making tasks.
With the decades of efforts, especially in supervised and unsupervised learning, many pattern recognition problems have been successfully explored, e.g. speech recognition and machine translation, and visual object detection and recognition. By contrast, machine decision making is in its infancy due to a lack of understanding of its intrinsic complexity.
Multi-agent learning arises in a variety of domains where intelligent agents interact not only with the (unknown) environment but also with each other. It has an increasing number of applications ranging from controlling a group of autonomous vehicles/robots/drones to coordinating collaborative robots in production lines, optimising distributed sensor networks/traffic, and machine bidding in competitive e-commerce, search and information retrieval and financial markets. Our research interests aim to address the fundamental challenges facing the field.
Please contact Jun Wang for more information
3. Complex networks - communication, information and social networks
Complex networks are large-scale systems with intricate and evolving patterns of connection between their elements. We study a wide range of complex networks in cyberspace, society and nature, including communication networks (e.g. the Internet), information networks (e.g. the Web), online social media networks (e.g. Twitter), as well as technological networks (e.g. global aviation network) and biological networks (e.g. gene co-expression network).
We utilise methods from network science and machine learning to obtain better understanding and more effective control of complex systems. For example, we modelled the Internet topology to assist Internet service providers to plan their business; we detected massive botnets in Twitter to prevent spamming attacks; and we proposed the critically hybrid spreading model to analyse the outbreak of Conficker worm on Internet and the HIV in vivo infection.
Please contact Shi Zhou for more information.
4. AI for health
Our primary research area is digital/computational epidemiology. Digital epidemiology uses data that was generated outside the public health system, i.e. data that was not generated primarily for the health purposes. Epidemiology research involves developing methods for using non-traditional (but evidently informative) data sources to provide early warnings of epidemics, understand disease properties better, assess the effectiveness of public health responses, and support the earlier diagnosis of health conditions. Our core methodological foci are in machine learning and natural language processing.
We are members (co-Is) of the i-sense (EPSRC) and VirusWatch (MRC) projects. Our current research is also supported by Google and the UCL Institute of Education. Our disease prevalence estimates for influenza and COVID-19 are fed directly to the UK Health Security Agency (formerly known as Public Heath England), and are included in their weekly disease surveillance reports.
We thank the following organisations for their financial support:
- UK Research and Innovation (EPSRC, MRC),
- National Institute for Health Research (NIHR)
as well as the following corporate organisations:
- Microsoft Research