Research Newsletter - Issue 79: Spotlight

Through a recent strategy document ‘AI, Here for Good’ the Irish Government has flagged the ambition to position Ireland as a key player in responsible Artificial Intelligence, seeking to exploit AI in a positive way to retain Ireland’s global competitiveness and future productivity.

Notably, many of the strategy strands align well with existing DCU cross-disciplinary research strengths and interests. These include areas such as AI and Society; Governance and Trustworthy AI; AI Education, Skills and Talent; and Driving adoption of AI in Irish Enterprise.

The current National Challenges call reflect these priorities, with 2 strands (the Our Tech Challenge; and the Digital Resilience Challenge) seeking to fund collaborations in the area of AI in society and broader digitalisation themes. The closing date for these Challenges is the 10th February 2023, and we encourage researchers preparing or considering applications to get in touch with their Faculty Research Development Officer (RDO) or

By way of exemplar for the broad application of AI in Society,  Prof. Boualem Benatallah outlines his fascinating work below:


Spotlight on AI-Enabled Services and Processes 


Prof. Boualem Benatallah

Prof. Boualem Benatallah

Professor Boualem Benatallah joined the School of Computing in DCU in Jan 2022, with over 21 years as a senior lecturer, associate professor, professor and then Scientia Professor at UNSW, Sydney, Australia.

He was among the first researchers to establish a body of research in model-driven Web services engineering in the early 2000s. He advocated that heterogeneity and dynamicity of component services would mean that static and low-level programming-based systems, traditionally used in service composition and integration, would not scale. His work made seminal contributions in model-driven service composition languages that offered an abstraction layer where low-level service details are abstracted, organised, customised, and composed, and an automation layer where orchestration logic is automatically generated at run-time, in accordance with high-level composition models.


He and his team performed foundational work in model-driven service composition, Quality of Service-aware Web service synthesis, and Web service middleware, all of which has had lasting impact. He has earned not only peer recognition but also substantial government and industry funding, major industry and international collaborations, patents, and considerable influence on the design of service-oriented architectures, composition techniques and languages.


AI -Enabled Augmentation

More recently, Prof. Benatallah started a research stream on endowing digital processes with AI augmentation methods to orchestrate human–machine conversations over software-enabled services.

AI-enabled augmentation promises to transform services through data-driven automation and insights.

Increasingly, organisations are choosing to use AI to automate business processes, deliver data-driven insights, and augment and improve the productivity and effectiveness of their customers, workers and stakeholders. The widespread adoption of digital services, coupled with AI-enabled augmentation, is poised to usher in enhanced business agility – through greater simplicity and increased automation of digital processes.

Aside from creating new opportunities per se, AI-augmented services and processes are poised to bring positive outcomes to a broader community of service providers and recipients in government, education, health, security, agriculture, research and other sectors. For instance, AI enablement is clearly transformational when applied to elderly and blind aids, patient health advice and access to health, agriculture, and research knowledge.


Prof. Benatallah’s research on AI-enabled services has broad application – from software engineering and business processes, cyber security, academic research investigations and digital agriculture, to data-driven investigations and supporting the visually impaired.

In collaboration with former PhD students and colleagues, Benatallah investigated how AI-enabled services augmentation can partly automate some time-consuming research tasks such as screening research papers and creating systematic literature reviews. This will obviate the manual effort of analysing a large number of research studies to determine which are relevant to the research questions of concern.

Similarly, one of Prof. Benatallah’s former projects investigated support of data-driven investigators in the task of collecting and analysing large quantities of information (A project funded by the Data 2 Decisions Cooperative Research Centre, at UNSW, Sydney, Australia (2017-2019)).  As information accumulates during an investigation, it becomes vital to keep track of relevant events and detect possible findings from raw evidence logs. Many hours of collection and analysis could be consumed in processing thousands of emails, messages and documents, but this task can now be undertaken far more efficiently thanks to AI-enabled augmentation of the data curation processes.


Other significant applications of Prof. Benatallah’s research include:

(i) human-AI collaboration in the screening of vulnerability discovery reports in security testing platforms based on crowdsourcing, by helping platform professionals to analyse and validate a large number of varied quality vulnerability reports submitted by testers and researchers,

(ii) AI-enabled conversational services to support individuals who are visually impaired or blind, so they can communicate with online services using their voice, giving them access to online social, work and learning environments.


More broadly, recent research by Prof. Benatallah and his collaborators also aims to study and define software engineering methods to develop, deploy and continuously assess human-AI collaboration services. They advocate a transparent and trustworthy design process by making the process explicit, along with the goals, assumptions and risks of the envisioned AI-enabled services. This is not just to ensure fairness of AI-enabled systems, but also to consider the implicit and explicit assumptions that drive decisions. It should be noted that despite its early adoption, engineering quality in AI-enabled services is still in the early stages of development, with many unsolved theoretical and technical challenges stemming from the lack of support for complex tasks and growing concerns about the unintended consequences of poor quality-control methods, including bias, intellectual property risks, copyrights, safety, privacy risks, and malicious attacks.

Another research stream of Prof. Benatallah’s recent research (in collaboration with colleagues from France and USA on engineering quality control of AI enabled services) developed systematic techniques to mitigate bias in emerging pre-trained language models – such as the presence of discriminatory gender and race stereotypes across many social constructs.[1] This work proposed novel debiasing methods for pre-trained text encoders that both reduce social stereotypes and inflict next to no semantic loss.


To learn more about Prof. Benatallah’s work on AI-enabled services and processes, or to discuss potential collaboration opportunities, please reach out via



[1]Yacine Gaci, Boualem Benatallah, Fabio Casati and Khalid Benabdeslem. Debiasing Pretrained Text Encoders by Paying Attention to Paying Attention. EMNLP 2022, Research track, Association for Computational Linguistics 2022 (to appear in Proc. of EMNLP main proceedings)