Impact case study database
Search and filter
Filter by
- University of Greenwich
- 11 - Computer Science and Informatics
- Submitting institution
- University of Greenwich
- Unit of assessment
- 11 - Computer Science and Informatics
- Summary impact type
- Technological
- Is this case study continued from a case study submitted in 2014?
- No
1. Summary of the impact
The Fire Safety Research Group’s (FSEG) research in fire dynamics, human behaviour, and fire and evacuation modelling has driven the development of state-of-the-art software tools, SMARTFIRE and EXODUS. The societal impact of this research is evidenced by its contribution to the design of safer aircraft, ships and buildings globally. FSEG has generated economic impact from over 300 software license sales to organisations in over 30 countries, which have applied the software commercially to build safety into innovative designs. The research has led to, or enabled, the development of innovative safety products, such as dynamic and adaptive emergency signage, and low-cost domestic fire suppression systems. Public policy impacts entail the UK government’s use of its research to frame guidelines on the positioning of security bollards, and the Grenfell Inquiry’s adoption of FSEG recommendations to improve the safety of residential high-rise buildings. Impact on practitioners is demonstrated by their use of the government guidelines on security bollards, the application of EXODUS by UK government security services to assist in mitigating terrorist threats, and the wide-scale use of the FSEG software by engineers globally.
2. Underpinning research
From its inception, FSEG has engaged in multi-disciplinary research, bringing together computer scientists, mathematicians, engineers, psychologists and 3D modellers to investigate fire dynamics and human behaviour, develop effective software to model fire and evacuation, and support the design of safer environments. Initial research focused on understanding fire and human behaviour in aircraft accidents, through projects funded jointly by the UK CAA and EPSRC. This utilised commercial Computational Fluid Dynamics (CFD) software to develop one of the first CFD models to simulate fire within the complex environment of an aircraft cabin. Limitations of commercial CFD software to represent complex geometries and provide effective user support for fire engineers led FSEG, with EPSRC support [3a], to research and develop a more sophisticated fire specific CFD simulation software tool that could be used easily by fire engineers: SMARTFIRE [3.1, 2001 ]. SMARTFIRE is an unstructured mesh CFD fire simulation code written in C++. Research into its core physics sub-models, numerical solution algorithms and visualisation techniques, has driven its constant development.
Industrial funding [3b-3d] and research grants [3e-3g], supported FSEG to expand the fire simulation capabilities of SMARTFIRE to include water mist suppression [3d], advanced combustion, toxic gas, flame spread, advanced smoke models, as well as parallel computing techniques utilising networked PCs with dynamic load balancing to improve performance [3.1-3.6, 2001-2020 ]. A key feature of SMARTFIRE was the use of a knowledge-based systems approach to make the system highly usable by, and appropriate to, the needs of fire engineers, including a novel problem set-up phase supporting those not expert in CFD modelling. Close collaboration with practitioners informed development of the software’s capabilities and utility, including assisting users to handle complex setup tasks, such as boundary condition specification and meshing [3.3-3.5, 2006-2020 ]. This responsiveness assisted SMARTFIRE’s development, and its adoption for applications throughout the built [3.2, 2007 ], maritime [3.5, 2020 ] and aviation environments [3.3, 2006 , 3.4, 2017 ] worldwide. Applications include embedded advanced fire simulation within immersive training environments [3.5, 3c, 3g].
To understand how fire impacts safety, it is essential to consider both fire and human dynamics, and how they interact. Modelling both fire development, and human behaviour and evacuation, is therefore vital. FSEG software achieves this by coupling fire and evacuation simulations of populations exposed, and reacting, to the developing fire. The group’s evacuation research focused initially on developing a simulation tool that could predict the behaviour of passengers evacuating from a post-crash aircraft fire. This led to the development of the world’s first agent-based evacuation model coupling fine-grained spatial resolution, adaptive human behaviour, toxicological models and fire hazard data: airEXODUS [3.4]. A series of EU-funded aviation fire safety research projects, e.g. [3e], built on the unique interactive coupling between fire simulation (SMARTFIRE), human behaviour and evacuation modelling (airEXODUS). The sophistication of both the fire and evacuation modelling tools was enhanced over the years through development of advanced combustion models to accommodate modern materials, including composites, and advanced rule-based behaviour models to simulate the interaction of passengers with cabin crew. These advanced modelling capabilities enabled the coupled airEXODUS – SMARTFIRE software to address complex fire scenarios involving interaction of people with the evolving fire environment e.g., explaining why 55 people lost their lives in the Manchester Boeing 737 fire from 1985 [3.4].
FSEG evacuation research expanded to include the built environment. The agent-based modelling concept using a fine spatial mesh was adapted from aircraft evacuation. A variety of grants including [3h, 3i], supported the extension of the modelling to represent behaviour associated with stairs, route-finding, movement of the mobility impaired, interaction with signage, groups, movement in smoke filled environments, elevators etc [3.7, 2001 , 3.8, 2016 ]. These behavioural and modelling capabilities were further enhanced to include the ability to simulate the impact of marauding armed terrorists in crowded places [3.9, 2018 ]. To address real-time applications, FSEG developed the world’s first parallel implementation of a rule-based evacuation model. The approach, using domain decomposition and dynamic load balancing, also enables the simulation of extremely large problem domains [3.10, 2017 ].
The value of FSEG research on behaviour in the built environment extends beyond evacuation models. Research into how people react to signage revealed that only 38% ‘see’ conventional static emergency signage in emergency situations [3.8]. To address this and enable signs to adapt to the evolving hazard environment, FSEG developed the concept of the Active Dynamic Signage System (ADSS). This was developed and demonstrated further in the EU FP7 GETAWAY project, improving detection of signs by over 100% [3i, 3.8]. The concept’s value was recognised in 2014 through ‘The Guardian University Award for Research Impact’ ( http://bit.ly/FSEG_guardian_award). Another example is FSEG research characterising the interaction of pedestrians with security bollards, which demonstrated that strategically placed bollards had little negative impact on exit flow [3k, 3.11, 2014 ].
FSEG research again expanded to include maritime/ship environments, which pose additional challenges, such as a heeled deck, and the impact of lifejackets, on human performance. Here, the coupling of SMARTFIRE with maritimeEXODUS enabled fire to be reliably represented within a ship evacuation scenario for the first time [3j]. This work was supported by a major EU FP5 project [3j] enabling the collection of data on the time required by passengers on ships at sea to respond to the evacuation alarm and begin the evacuation process [3.12, 2007 ], an essential parameter in evacuation modelling. It also enabled data collection on passenger performance in dynamic situations involving roll, and the impact of smoke on passengers in heel and roll situations. Data generated from this work was used in the development of the maritimeEXODUS software [3.5, 3.12], particularly in modelling the interaction of passenger movement in smoke-filled corridors subjected to heel. The significance of this was recognised through the 2001 Royal Institution of Naval Architects/Lloyds Register ‘Safer Ships’ award.
The overall quality of FSEG research is demonstrated through the award of several national and international prizes: • 2001 British Computer Society award for IT, CEO Judith Scott commenting, “The winners not only demonstrate technical innovation, but also show how technology can be used to benefit society at large.”; • 2002 Queen’s Anniversary Prize: “The University is a recognised world leader in the area of evacuation model development. Use of its software technology by businesses and public authorities greatly enhances public safety and its specialised training offers vital expertise to the user community worldwide.”
3. References to the research
Wang, Z., Jia, F., Galea, E.R., Patel, M.K., Ewer, J. “Simulating one of the CIB W14 round robin test cases using the SMARTFIRE fire field model.” Fire Safety Journal, 36, pp661-677, 2001. https://doi.org/10.1016/S0379-7112(01)00018-2
Galea E.R., Jia F., and Wang Z. “Predicting toxic gas concentrations resulting from enclosure fires using local equivalence ratio concept linked to fire field models”. Fire and Materials, Vol 31, Issue 1 pp 27-51, Jan/Feb 2007. https://doi.org/10.1002/fam.924
Jia F, Patel M, Galea E, Grandison A, Ewer J. “CFD Fire Simulation of the Swissair Flight 111 In-Flight Fire – Part 2: Fire Spread analysis”, The Aeronautical Journal. Vol 110, Number 1107, pp 303-314, 2006. https://doi.org/10.1017/S0001924000013178. This paper won the 2017 gold medal from the Royal Aeronautical Journal.
Galea, E.R., Wang, Z., and Jia, F (2017). Numerical Investigation of the Fatal 1985 Manchester Airport B737 Fire, The Aeronautical Journal, Vol 121, Number 1237, pp. 287-319. https://doi.org/10.1017/aer.2016.122. Winner 2017 gold medal from Royal Aeronautical Journal
Woolley, A ., Ewer, J., Lawrence, P., Deere, S., Travers, A., Whitehouse, T., Galea, E.R. A naval damage incident recoverability toolset: Assessing naval platform recoverability after a fire event, Ocean Engineering, (2020), 207, 107351. https://doi.org/10.1016/j.oceaneng.2020.107351
Grandison, A. Galea, E.R., Patel, M.K., and Ewer.J. Parallel CFD fire modelling on office PCs with dynamic load balancing, Int J Numer Meth Fluids, 55, pp29-39, 2007. https://doi.org/10.1002/fld.1278
Gwynne S., Galea, E. R., Lawrence, P.J. and Filippidis, L. “Modelling Occupant Interaction with Fire Conditions Using the buildingEXODUS model”. Fire Safety Journal, 36, pp327-357, 2001. https://doi.org/10.1016/S0379-7112(00)00060-6
Galea, E. R., Xie, H., Deere, S., Cooney, D., and Filippidis, L. (2016) An international survey and full-scale evacuation trial demonstrating the effectiveness of the active dynamic signage system concept. Fire and Materials. https://doi.org/10.1002/fam.2414
Galea E.R., Blackshields, D., Lawrence. P., Deere, S. Simulating a Marauding Terrorist Firearms Attack (MTFA) ** * **, Final Report. 23/02/18. Secret Report for DSTL and DfT.
Grandison, A., Cavanagh, Y., Lawrence, P.J. and Galea, E.R. (2017). Increasing the simulation performance of large-scale evacuations using parallel computing techniques based on domain decomposition, Fire Technology, 53, pp1399-1438. https://bit.ly/38gQ8Md
Galea, E.R., Cooney, D., Xie, H., Sharp, G.G. Impact of Hostile Vehicle Mitigation Measures (Bollards) on Pedestrian Crowd Movement. Phase 2 Final Report, Dept for Transport, CPNI, Oct 2014. Published on UK government website on 15 Nov 2016. https://bit.ly/3bit3uq
Galea, E. R., Deere, S., Sharp, G., Filippidis, L., Lawrence, P. J., & Gwynne, S. (2007). Recommendations on the nature of the passenger response time distribution to be used in the MSC 1033 assembly time analysis based on data derived from sea trials. International Journal of Maritime Engineering, 149(A1), 15–29. http://gala.gre.ac.uk/id/eprint/1076
Example research grants:
E. R. Galea. EPSRC Grant (GR/L56749/01). The SMARTFIRE Fire Simulation Environment. 1997-2000. £171,000.
E. R. Galea. Borealis projects, Evacuation (Pt1) and Fire toxicity (Pt2), Jan 02 – Mar 05, £85,000.
E. R. Galea. Australian DSTO. Virtual ship simulator. Part I – III, Jan 14 – Feb 18, £202,000
E. R. Galea. TSB/EPSRC (SKTP). Plumis (project 1000921), Water Mist Fire Suppression Modelling, Nov 13 – Nov 14, £48,550.
E. R. Galea. EU FP6 (project 516068): NACRE , Apr 05 – Mar 09. €590,000.
E. R. Galea. EU FP7 (project 218761): FIREPROOF. Jun 09 – Jun 12, €345,000
E. R. Galea. EU Horizon2020 (project 653590): AUGGMED, Jun 15 – May 18, €640,000
E. R. Galea. Part of HEED Consortium (Project concerned the evacuation of the World Trade Centre). EPSRC (GR/S74201/01 and EP/D507790). Sep 04 – Oct 07. £1.5 million.
E. R. Galea. EU FP7 (project no. 265717): GETAWAY. Nov 11 – Oct 14, €572,438
E. R. Galea. EU FP5 (contract GRD2-2001-50055): Fire Exit. 2001-2005. £325,000.
E. R. Galea. CPNI (UK Home Office). Experimental analysis of the impact of Bollard Arrays on Pedestrian movement – Parts I and II, Nov 12 – Aug 15, £200,000.
4. Details of the impact
(1) Economic Impact: (i) During the REF period, UoG has generated over £1.15million from over 300 license sales of SMARTFIRE and EXODUS software to organisations in over 30 countries [5.1]. (ii) Licensees included engineering consultancies e.g. Bureau Veritas, regulatory authorities e.g. China Maritime Classification Society, national laboratories e.g. DSTO Australia. The software was used in cutting-edge design to explore and improve the evacuation safety of complex structures, generating considerable consultancy income. EXODUS software was used for the Airbus A330-X, A340 and A380 projects [5.2]. FSEG and the airEXODUS software were used in the preliminary design of the multi-billion-euro A380, and to de-risk the A380 full-scale evacuation certification trial. The software, and the research underpinning it, contributed to the A380’s safety [5.2] and potentially saved the manufacturer millions of euros by identifying possible problems that might occur during the trial, which could have caused cost overruns, resulting in a higher unit cost. Through the REF period, the A380 world fleet safely carried 150 million passengers over 3.3 million flight hours. Since 2013, Airbus has delivered 131 A380s, with 228 in service (May 2020). FSEG research assisted these sales by contributing to product safety, and keeping the price down, critical considerations for airline customers [5.2]. (iii) FSEG fire modelling expertise and SMARTFIRE fire simulation software assisted three-person start-up company PLUMIS to develop and refine their innovative prototype fire suppression system into an efficient cost-effective life-saving product. In just 12 years, the start-up has grown to have a £3.2M turnover employing 43 people [5.3]. (iv) EXODUS and SMARTFIRE tools give fire engineering firms a competitive edge when bidding for projects, enabling them to win important contracts, generating significant income. Examples during the REF period include global engineering firms Arcadis and Thornton Tomasetti which used FSEG software, under license, to undertake early design assessments of the life safety and emergency management systems for major projects such as Rolls Royce Goodwood (UK), Jaguar Land-Rover Castle Bromwich (UK) [5.4], a large Data Center in the USA and a car rental facility at a major US airport [5.5]. (v) FSEG research, which lead to the development of the ADSS concept [3i] has created a new market in emergency signage technology. The ADSS concept improves signage detectability 100% enabling the sign to adapt to a changing hazard environment [3.9]. Signage manufacturers globally have adopted and adapted the concept. EVACLITE and CLEVERTRONICS are 2 companies focused on the FSEG ADSS concept. EVACLITE was set up in the UK to manufacture and sell ADSS [5.6]. CLEVERTRONICS, Australia’s leading emergency lighting company, adopted the ADSS concept and started a new business to develop and manufacture the product [5.7]. They are both founded on the FSEG concept, with many other companies globally adapting similar concepts based on FSEG fundamental research.
(2) Impact on Public Policy. (i) Security bollards have become a common feature protecting public spaces, part of the UK’s Hostile Vehicle Mitigation strategy. The Centre for the Protection of National Infrastructure (CPNI) is the government organisation reporting to the Home Office that advises on security issues related to national infrastructure. CPNI had produced guidance on the positioning of bollard arrays, but the advice did not include their impact on evacuation flow. This is an important issue because the initial evacuation safety analysis used in the design and certification of these structures did not take into consideration that a ring of security bollards would be placed outside the exits. FSEG research, sponsored by CPNI and DfT [3k], investigated the impact that security bollards have on evacuation flows using a series of full-scale experiments, which identified and quantified, for the first time, not only how bollards impact evacuation flow, but also how they could be positioned to minimise their impact [3.11]. The research resulted in a new set of DfT guidelines (written by Prof Galea), which are used internationally to optimally position security bollards around critical infrastructure [5.8]. (ii) Prof Galea, one of six experts to the Grenfell Fire Inquiry, provided a report to the Inquiry containing 42 evidence-based [3.7] interim recommendations to improve evacuation of residential high-rise buildings [5.10]. Of these, 22 were partially/fully adopted in the Chairman’s Phase 1 report. For example, recommendation 2.9 on luminous floor numbering was adopted as recommendation 15 (para 33.27); recommendation 2.11 on PEEPs was adopted as recommendation 12e (para 33.22), and recommendations 3.1–3.3 and 3.6-3.8 on full building evacuation were adopted as recommendation 12a (para 33.22) [5.11]. The recommendations aim to improve public safety through UK regulatory/policy change.
(3) Impacts on Practitioners and Professional Services. (i) The new guidelines [5.8] for optimal positioning of bollard arrays impact professional practice through the modification of the previous design practice. (ii) A restricted version of the EXODUS software has been developed for use by UK government security services to assist in planning mitigation strategies for marauding armed terrorists in crowded places [3.9]. The use of this approach has had a significant impact on how security professionals plan for and develop counter measures [5.9]. (iii) Over the REF period, FSEG software has been used by over 300 licensees in 30 countries (see 1i), becoming standard engineering design tools for safety analysis, used by fire safety engineers around the world. The software therefore has impact on engineering professional practice globally [5.4, 5.5].
(4) Impacts on Social Welfare. (i) The main impact of FSEG research on society is public safety: safer aircraft, buildings and passenger ships through more effective safety standards and policy, and safer designs. Examples: ensuring that security bollards around infrastructure have minimum impact during emergency evacuation (see 2i); through the introduction of novel concepts in emergency signage, making them more effective, and buildings safer (see 1v); through recommendations to improve safety of residential high-rise buildings (see 2ii); through assisting in the development of cost-effective water mist fire suppression systems for domestic environments, now installed in over 5000 UK properties (incl. 350 sheltered scheme flats), to protect the most vulnerable from fire (see 1iii). (ii) Members of the public make an important contribution to their personal safety, safety of families and of society. Engagement to improve public understanding of safety and how to minimise risks associated with fire and evacuation is therefore critical, and a further societal impact of FSEG. FSEG promotes public understanding of science via the media, helping to build resilience to science scepticism, which the COVID-19 pandemic has highlighted as a growing issue. This is achieved through media coverage of FSEG research and interviews with Prof Galea on evacuation and pedestrian dynamics. It has also informed future industrial partners and policy makers. Examples: FSEG research on pedestrian interaction with autonomous vehicles: BBC Radio 4 ‘All in the Mind’, 08/05/18 ( https://bbc.in/2yBjxxZ). Importance of standing on London Underground escalators in rush hour: BBC Radio 4 Today, 10/03/16 (audience 1.2m); BBC Radio 5 Breakfast Programme, 07:25, 18/01/16 (audience 444k); BBC World Service Weekend Programme 07:50 16/04/16. FSEG EU Horizon2020 project AUGGMED, developing VR training environments for security services: BBC CLICK 28/04/17 ( https://bbc.in/3qdBuLW). Prof Galea interviews on Grenfell Tower Fire: BBC Radio 4 Inside Science 15/06/17 ( https://bbc.in/3uVOmdg);NYTimes, (4.7m digital subscription) 24/06/17 ( https://nyti.ms/3qiPeFh), The Economist, (24/07/17) (1.7m global print and digital circulation), ( https://econ.st/2yFZXR9).
5. Sources to corroborate the impact
University of Greenwich sales accounts, SMARTFIRE & EXODUS licences (Aug 13–July 20)
Testimonial: Airbus Chief Engineer, France.
Testimonial: Plumis Director, UK
Testimonial: Arcadis, Associate Technical Director, UK
Testimonial: Thornton Tomasetti, Principal, UK
Evaclite: identifies FSEG research incl GETAWAY [3i] https://www.evaclite.com/about-us/; https://www.evaclite.com/directional-safety-signage-systems/
Testimonial: Clevertronics Managing Director, Australia
Traffic Advisory Leaflet 01/16, 15 Nov 2016, Dept for Transport. http://bit.ly/UKgov-bollards
Testimonial: DSTL, lead Modelling and Simulation Strategy, Platform Systems Division, UK
Galea E.R., Interim Phase 1 Recommendations for the Grenfell Inquiry - Final, 02/04/19. https://bit.ly/3qkzbXm
Grenfell Tower Inquiry: Phase 1 Report. Chairman: The Rt Hon Sir Martin Moore-Bick October 2019. Vol 4 (Part 5). https://bit.ly/3kNvQPr
- Submitting institution
- University of Greenwich
- Unit of assessment
- 11 - Computer Science and Informatics
- Summary impact type
- Economic
- Is this case study continued from a case study submitted in 2014?
- No
1. Summary of the impact
Research led by Dr Georgios Samakovitis on fraud detection and anti-money laundering has resulted in multiple impacts related to financial services and their regulation. The nature of the impact is commercial-economic and public policy-related, with key beneficiaries being UK government, industry and financial services regulators, and society as end-users. The techniques and concepts introduced by Samakovitis had distinct material contribution to changing the perception of UK public bodies and the financial services industry, initially through attracting stakeholder interest in industry fora, and later through Samakovitis’ direct contribution in the UK Payments Strategy Forum and the HMG Cabinet Office Counter-fraud Data Advisory Group.
More specifically, successful application of Case-based Reasoning (CBR) for fraud detection in electronic money transfer has proven that intelligent techniques that rely on prior knowledge (confirmed illicit transactions) mitigate costly misclassification risks borne by other, often more advanced, practices. Combined through an operational framework for secure transaction data sharing (the Collective Intelligence Hub – (CIH)), the above results provided “ strong evidence for regulators, payment service providers, and users that financial transaction sharing is both feasible within the required data privacy frameworks, and necessary to significantly enhance mitigation capabilities in Counter-Fraud” [5.3]. The work was fractionally supported by Dr. Stelios Kapetanakis who departed from Greenwich in late 2013.
2. Underpinning research
The underpinning research stems from the development of intelligent automated methods and tools for identifying financial transactions fraud and, in so doing, facilitate users (financial institutions, policy makers and regulators) to inform their approach and policies in investigating illicit money exchange. Tested on different financial transaction datasets, the research highlighted the capability to leverage characteristics of previously identified fraud to automate detection by predicting patterns of such illicit exchange; this further demonstrated the significant practical value that AI techniques carry, in informing decisions on counter-fraud and anti-money laundering policy.
The work highlights early use of AI techniques to structure financial transaction data, which has uncovered patterns of mechanisms that informed regulations. Even though the specific AI techniques used (CBR) may not scale-up as well as more recent AI and Data Science techniques, it is the heuristic contributions, in terms of pattern detection that made it possible to inform regulations. These heuristics will outlast the use of a specific AI technique, notwithstanding the fact that CBR maintains an advantage in terms of representation semantics over techniques that have since gained greater popularity.
The earlier research [3.1] presented a prototype system introducing a workflow approach to identify abnormal financial transactions by applying CBR (Case Based Reasoning, an Artificial Intelligence approach) for transactions classification. The approach features characteristics which are directly usable in fraud and money laundering identification, such as the construction and representation of integrated cases from transaction data and its ability to derive and leverage case similarity as a critical identifier of suspected fraud. Grounded on results of the above research, [3.2] then proposed an intelligent Financial Fraud Detection framework and architecture that acts as both an analytical and policy model for anonymised financial transaction data sharing.
On a parallel research stream, impact was underpinned by work [3.1, 3.3] investigating the use of Case-based Reasoning (CBR) for identifying fraudulent transactions in mobile money transfer systems [3.1] and credit card transactions [3.3]. In particular, work in [3.3] focused on the development and testing of CBR techniques to detect fraudulent transactions with pre-classification of proven fraud cases, which succeeded in predicting a significant part of unknown fraud cases. The research offered a tool at a proof-of-concept level that is testable on real transaction datasets. The work described in [3.3] above expanded in scope to investigate transaction fraud in mobile money transfer (MMT) thus addressing an area of significant exchange value (presently at $130bn, expected to grow to $203bn by 2024), especially at a time where MMT forms the core mechanism for Peer-to-peer (P2P) and so-called Social Payments, which accounted for approximately 80% of transactions globally in 2019, according to Juniper Research.
Similarly, the research in [3.4] underpins impact through investigating the novel use of combined artificial intelligence approaches (natural language processing, CBR and deep learning) to detect social engineering in networks; the proposed model delivers high accuracy in detecting attacks, and is tested on real data. The approach’s underlying significance is in its potential use for critical support for payment networks; especially where money transfer occurs without financial intermediation (P2P), the proposed model can deliver important insights to users on potential malicious adversarial activity. In turn, its particular significance is largely reflected by the present and projected future growth figures in peer-to-peer (P2P) money transfer discussed above.
At the operational and policy levels, the research [3.2, 3.3] highlighted the critical role of technological architectures for secure cross-institution privacy-preserving data sharing for anti-money laundering, proposing “ the establishment of an independently commissioned Collective Intelligence Hub as the infrastructure and governance framework guaranteeing synergistic intelligence against money laundering networks” [5.2]; at the technological level (computational intelligence), the research developed, tested and reviewed knowledge engineering and machine learning techniques in anti-money laundering, fraud prediction in financial transactions, and social engineering. At that same technological level, later research by Samakovitis investigated applications of Distributed Ledger Technologies (DLTs – blockchain) for privacy preservation and transaction auditability. Notably, during the impact period (and particularly post-2017), emergent blockchain solutions challenged the incumbent technological paradigm in data sharing, across multiple areas of intelligent data analysis (not excluding fraud, identity verification or privacy preservation). However, it is important to highlight that, despite such disruptive barriers, the technological and operational framework underpinning the Collective Intelligence Hub (CIH) proposition is far from obsolesced through DLT, because of “the critical requirement for control and oversight by a centralised delegated authority” [5.2]. Conversely, several technological components of the CIH may arguably be improved with blockchain technologies, albeit not challenging its core operating principles. This is further evidenced by the present industry trajectory in the UK, where Vocalink, the operator of BACS, FPS and LINK in the country, has been developing network-wide counter fraud solutions along the same principles of centralised oversight as these recommended in the CIH (see indicatively https://www.vocalink.com/services/financial-crime-solutions/).
3. References to the research
Adedoyin A., Kapetanakis S., Samakovitis G., Petridis M. (2017) Predicting Fraud in Mobile Money Transfer Using Case-Based Reasoning. In: Bramer M., Petridis M. (eds) Artificial Intelligence XXXIV. SGAI 2017. Lecture Notes in Computer Science, vol 10630. Springer, Cham. https://doi.org/10.1007/978-3-319-71078-5_28Work in [3.1] was disseminated in the BCS Specialist Group in Artificial Intelligence (SGAI) conference and received the Best Paper award in the event’s Application Stream in 2017 .
Samakovitis, G. and Kapetanakis, S. (2013), Computer-aided Financial Fraud Detection: Promise and Applicability in Monitoring Financial Transaction Fraud, Proceedings of the International Conference in Business Management and Information Systems, (ICBMIS 2013) Nov. 19-21, Dubai, United Arab Emirates. https://gala.gre.ac.uk/id/eprint/17237/The work in [3.2], presenting a proposition for counter-fraud systems architectures, focuses primarily on the policy-related aspects of the recommended model (rather than on its technical feasibility) and, as such, was disseminated to a venue which, while not core to the relevant Unit of Assessment, allowed for the necessary visibility to how the proposed Collective Intelligence Hub solution can be modelled and commercially operationalised.
Kapetanakis, S., Samakovitis, G., Gunasekera, B. and Petridis, M. (2012), Monitoring Financial Transaction Fraud with the use of Case-based Reasoning, Seventeenth UK Workshop on Case-Based Reasoning (UKCBR 2012) 11th December 2012, Cambridge, UK. https://gala.gre.ac.uk/id/eprint/9780/ [Full text available from university on request]Work in [3.3] that formed the grounds for investigating the uses of Case Based Reasoning (CBR) in fraud detection, was disseminated in the specialist workshop (UKCBR) which is part of the SGAI conference and the key CBR-specialist venue in the UK.
Lansley, M., Polatidis, N., Kapetanakis, S., Amin, K., Samakovitis, G., and Petridis, M. (2019). Seen the villains: Detecting Social Engineering Attacks using Case-based Reasoning and Deep Learning. In Workshops Proceedings for the Twenty-seventh International Conference on Case-Based Reasoning co-located with the Twenty-seventh International Conference on Case-Based Reasoning (ICCBR 2019), http://ceur-ws.org/Vol-2567/paper4.pdfWork presented in [3.4] was disseminated in the International Conference in Case Based Reasoning (ICCBR), which has a Core B ranking ( http://portal.core.edu.au/conf-ranks/).
4. Details of the impact
Underpinning research work, “ presented to a wide range of payment professionals and officials informed and, to an extent, influenced that subsequent work – both in government and in the payments industry – on data sharing for the combatting of public sector fraud and money laundering” [5.2]. Specifically, the impact of the underpinning research is visible across commercial and public policy, through (1) informing the UK Financial Services industry strategy for electronic payments, (2) influencing policy discussions on countering fraud and money laundering practices, which shaped the Payment Regulators’ position, and; (3) informing the perception of government stakeholders and policymakers on how data sharing strategies, especially pertaining to financial fraud and money laundering practices, can improve Civil Service operation, via direct advisory contribution with the HMG Cabinet Office. Further societal impact materialised through (4) motivating the public debate on counter-fraud among beneficiaries, including the citizen as end-user of financial services.
The underpinning work, carried out from 2012 to 2020, focused on enhancing the performance of counter-fraud and Anti-Money Laundering regimes through collaborative models grounded on Artificial Intelligence (AI) and data analytics. The impact of this has primarily taken the form of influence on UK public policy and regulation, informing the public and industry debate on technologies and approaches for counter fraud and anti-money laundering in the UK, [5.5 – 5.9] and informing the introduction of policies which, in turn had an impact on economic growth.
The research came at a time (2013) well-predating UK regulators’ efforts to encourage or support such models, which later became mainstream, as also witnessed by the relevant New Payment Systems Operator (NPSO, later Pay.UK) strategies which were only released in 2016. More specifically, the analytical model delivered in [3.2], branded under the concept ‘Collective Intelligence Hub’ (CIH) “was seen at the time to offer a realistic and usable approach to how data sharing for counter-fraud purposes in the public sector (e.g. for Universal Credit) could be automated to provide additional protection for public money.” [5.2] and was presented in December 2014 to a panel of financial services executives and payments industry stakeholders, coordinated by CIFAS (Credit Industry Fraud Avoidance Scheme) and advising the Government Coordination Committee, as outlined in (2) below. While the proposal was not adopted at the time, the Collective Intelligence Hub was later presented in the form of a technological framework and solution in March 2016 to the Payments Strategy Forum (PSF; established Feb 2015), at the early stages of deliberations in the Financial Crime Data and Security Working Group, whose “remit was to identify initiatives, where the industry could collaborate and work with Government, to deliver step-change capabilities to tackle fraud and financial crime risk in payments” [5.3]. The CIH proposal was formally recognised to be commensurate with the strategic vision of the PSF.
The underpinning work has addressed solutions applicable in both public and private sectors; the academic research, outlined in Section 2, concentrated on the use of AI approaches for combatting fraud in electronic payments networks and on developing Anti-Money Laundering models and solutions. Throughout the impact period (2012-2020), the research contributed directly on the recommendation of appropriate tools, techniques and infrastructure for future deployment, in both the UK financial services and public sector. In the impact period, the contribution trajectory took the form outlined here:
The underpinning research has attracted attention in the commerce field, and led to invited disseminations of research propositions in industry conferences and sector discussion panels. These events [5.5 - 5.9] thematically focused on operational and economic significance of intelligent techniques against fraud and money laundering. The resulting visibility of underpinning research attracted further attention by stakeholders in public policy (DWP) and regulation (FCA, CIFAS), subsequently leading to his direct involvement in activities outlined in (2), (3) and (4) immediately below.
On 17 December 2014, Samakovitis’ proposal for a UK-wide data sharing collaborative model was formally presented to a panel of banking professionals, (including representatives of the Bank of England and Financial Fraud Action (FFA UK)) advising the UK Government Coordination Committee. Termed Collective Intelligence Hub (CIH), the proposed model was “ one of the very few approaches at the time that suggested leveraging anonymous transaction data sharing between financial institutions” [5.2], as discussed later in this section. Although the legal challenges associated with such data sharing, both in the public and private sectors were such as to prevent practical actions being taken at the time, the “ *CIH approach added value by making explicit the potential benefits that could be realised if financial institutions could find a way to share data in order to counter money laundering and fraud, through **[*the CIH ] specific model for doing so” [5.2]. This involvement further motivated his recommended introduction to the then newly-founded Payments Strategy Forum (PSF).
From April 2015 (and until the completion of its objectives in December 2017) Samakovitis was a member and direct participant of the UK Payments Strategy Forum (PSF), a body commissioned by the new Payment Systems Regulator ( https://www.psr.org.uk/about-psr) as the main industry and policy forum on the UK’s future payments strategy for the next 20 years ( https://www.psr.org.uk/developing-payments-strategy-forum). Throughout that period (2015-17) “ the contributions by Dr. Samakovitis had continued presence and hence materially influenced the direction of the Working Group, particularly with reference to infrastructure and governance for anti-money laundering data sharing and utilisation” [5.3]. His distinct material contribution concentrated on the outputs of the Financial Crime Data and Security Working Group ( https://consultation.paymentsforum.uk/workinggroups/financial-crime), primarily under two Strategic Solutions: (i) Trusted KYC Data Sharing; a solution framework for a central shared repository to support financial services providers in their Know-Your-Customer (KYC) procedures and; (ii) Payments Transaction Data Sharing & Data Analytics; a solution framework that involves a centralised transaction storage facility and analytical capability, residing with a public body. Both solutions formed part of PSF’s “ recommended development of a transaction data analytics capability to be built for Faster Payments and BACS, the UK’s retail payments market infrastructure..[..]..Subsequent to the Forum’s work, this solution was implemented for Faster Payments in 2018, called MITS (Mule Insights Tactical Solution), and will be incorporated and enhanced in the new NPA [New Payments Architecture] infrastructure” [5.3]. In both Strategic Solution spaces, the operational principles and technological architecture proposed by Samakovitis were “ recognised as commensurate with the strategic vision of the PSF, and considered as supporting elements of the solutions that the Working Group later submitted to the Payment Systems Regulator” [5.3]. Collaboration with UK Government stakeholders and PSF Working Group participants has catalysed Dr. Samakovitis’ direct membership in the Counter Fraud Data Analytics Advisory Group of the HMG Cabinet Office (Counter Fraud Centre of Expertise), “..an established group of industry leaders, academics and third sector representative, …[which]…provides a forum for ongoing consultation, challenge and support for the development of the Profession” ( https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/730050/Annex_B_-_GCFP_Brochure.pdf). Within the remit of the Advisory Group, “ his underpinning research contribution has directly and materially supported clearer understanding of the technological solution space for identifying and mitigating fraud in the public sector” [5.4]. Along the same lines, the insights grounded on Dr. Samakovitis’ underpinning research “have had distinct and material contribution to informing the Counter Fraud Data Analytics Standard and, in so doing, supporting the establishment of role and education specifications for the Counter Fraud Profession in Government” [5.4]. In his capacity as academic advisor, he has directly supported the Cabinet Office Ministerial Thought Paper on fraud in the government sector [5.1] through a named contribution (pp. 64-67) proposing four core themes along which Government should pursue data sharing strategies. According to the Programme Director, Counter Fraud Centre of Expertise, “ research input by Dr. Samakovitis, introduced via his participation in the Advisory Group has materially informed and, to an extent, influenced the Centre’s position on how data sharing strategies can improve Civil Service operation, and contributes to its work in shaping the direction of the Counter Fraud function in Government” [5.4]
5. Sources to corroborate the impact
Evidence of the impacts:
HMG Cabinet Office Thought Paper (2019): Tackling Fraud in Government with Data Analytics: Starting the Conversation, Department for Digital, Culture, Media & Sports, June 2019 (pp. 64-67), https://www.gov.uk/government/publications/tackling-fraud-in-government-with-data-analytics
Testimonial – Innovation Lead, CDIO, HMRC, UK.
Testimonial – Former Payments Strategy Forum Working Group Programme Lead, UK
Testimonial – Programme Director, Counter Fraud Centre of Expertise, Cabinet Office, UK.
Evidence of the impact translation activities (media interviews; trade conference
presentations; industry events):
‘AI and advanced analytics in AML: From rule-based controls to intelligence-led capabilities’ Interview by C. de Monts-Petit, Editor, The Economist Intelligence Unit, Feb. 2020, available at: https://eiuperspectives.economist.com/technology-innovation/ai-and-advanced-analytics-aml-rule-based-controls-intelligence-led-capabilities
Samakovitis, G. (2018), ‘RegTech and the Role of Artificial Intelligence’ Invited Panel Session MLROs London Conference Three, London, 26 Sep 2018
Samakovitis, G. (2017), ‘Collective Intelligence for Client Onboarding: What AI can do for you (and what it cannot)’, Client Onboarding in Financial Services (22-23 Feb 2017),
Samakovitis, G. (2016), ‘Towards Collective Intelligence for AML: an Operational & Technological Framework for Addressing the Risk Balance across the Value Transfer Ecosystem’, Client Onboarding in Financial Services (24-25 Feb 2016),
Samakovitis, G. (2014), ‘Financial Transactions Monitoring & Fraud Detection – Challenges in the Current Environment‘, Info Crime: Information Security and Cyber Crime Summit (18-19 Feb 2014).