The Glasgow Information Retrieval Group within the School of Computing Science at the University of Glasgow was founded 32 years ago in 1986 by Professor C. J. ‘Keith’ van Rijsbergen, often considered one of the founders of modern Information Retrieval (IR). From its outset, the Glasgow IR group has focused on improving the effectiveness of IR systems, inventing new logic & probabilistic retrieval models in the 90’s and early 2000’s, followed by the development of adaptive query expansion techniques, interactive multimedia models, the Divergence From Randomness framework, as well as leading research into quantum, expertise search and search result diversification models in the late 2000’s. The group has currently 5 academics and about 25 research assistants and PhD students, researching machine learning/deep learning techniques for effective search and recommender systems.
The group has also a long history enhancing search engine efficiency, starting back in 2004 with the first release of the open source search engine Terrier, which is still being developed and maintained by the group to this day. Indeed, Terrier has acted as a testing ground for many early IR innovations, such as relevance feedback models, query performance/efficiency predictors, dynamic pruning strategies and learning-to-rank techniques. The group has a great breadth of experience working with different data types and scales, from small collections of academic, medical or government documents, up-to big data collections for Web search, enterprise search, social media analytics, IoT and user log data such as queries, check-ins and product purchases.
The group believes strongly in standardisation of IR evaluation, and has led international challenges and evaluation campaigns (Blog track, Microblog track, Real-time Summarisation track) at venues like TREC for over 10 years, as well as being a developer of metrics and open test collections for a range of IR tasks. The group’s current focus is on examining next-generation IR and recommender system challenges, including online evaluation, closed-loop effects and bias, data credibility, conversational agents as a new interaction paradigm and using heterogeneous streaming data to extract actionable information about the world. The group is also working on adapting and extending IR technologies for new tasks, such as cloud orchestration, identifying sensitive information within documents and enhancing emergency response during crises.
The academics of the group are:
- Professor Iadh Ounis specialises in large-scale IR with his research being at the intersection of information retrieval, data science/big data analytics, and sensing systems (e.g. social and physical sensors within smart cities).
- Professor Joemon Jose focuses on multimedia mining and search, multimedia interaction for information retrieval and adaptive and personalised search systems.
- Dr Craig Macdonald is interested in efficient yet effective search, as well as recommendation scenarios, as well as enterprise search and social media analytics.
- Dr Jeff Dalton works on conversational search and agent systems, relation extraction for knowledge graph completion and health knowledge discovery.
- Dr Richard McCreadie specialises in real-time IR, machine learning, big data stream processing and evaluation methodologies over streaming data.
Alumni from the Group have gone on to work at Amazon, Microsoft, Yahoo!, Yandex, Facebook, Spotify, IBM, JP Morgan, Telefonica, ASOS, Signal, Scoop Analytics, Reuters and Mendeley, as well as positions in universities in Scotland, UK, Brazil, China and the USA.
The Group have a long history of engagement with different public and industrial organisations on short-, medium-, and long-term interventions. For example, we can trial techniques or ideas through co-supervision of undergraduate or postgraduate projects (3 months – 6 months), perform knowledge exchange by trialling and tailoring existing state-of-the-art techniques in new search scenarios (3 – 6 months), engage with companies and the public sector through consultancy, innovation & collaborative R&D projects (e.g. through The DataLab), or through knowledge transfer partnerships (1-3 years), collaborate with a consortium within EU framework programmes such as Horizon 2020 (3-4 years), or perform new research into an emerging problem through co-supervision of PhD students (4 years). Indeed our research is funded by UKRI EPSRC, UKRI ESRC, DSTL, the EU, as well as via a wide range of public sector and industrial partnerships. Meanwhile, University-wide 50% match funding partnerships allow PhD co-supervision arrangements at very reduced expense levels to third parties.
We have built a long history both adapting technologies developed in the group and prototyping new solutions in collaboration with public sector and industry players. For example, we have supported the development of novel search and mining technologies for medical health records, financial stocks and shares, social media streams and government records, among others. For instance, our work on more efficient and green-search solutions prototyped within Terrier, were adopted by a major commercial search engine, markedly increasing query processing capacity, while also reducing electricity consumption and hence running costs.
Furthermore, the Glasgow Information Retrieval group, has placed significant investment into building and maintaining state-of-the-art high-performance-compute infrastructure to enable fast experimentation in-house for tasks such as learning-to-rank and deep learning. This allows agile development and experimentation of new technologies without the need to cost-in additional hardware to a collaboration. The group also has access to School and University-level infrastructure, enabling large-scale experimentation for tasks requiring big data, without relying on cloud services. Indeed, the School of Computing Science received £300k of funding for GPU compute infrastructure this year alone.
The wide-range of industrial collaboration mechanisms enables the Glasgow Information Retrieval group to work with companies large and small, either company-funded or via externally-funded programmes. The group has also access to a dedicated Business Development Manager who can facilitate IP & administrative discussions in a frictionless manner.
For collaboration opportunities with the Glasgow Information Retrieval Group, interested parties can contact us using the following details:
Professor Iadh Ounis, iadh.ounis[AT]glasgow.ac.uk