Sara Saravi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Sara Saravi has a total of 33 co-authors. Here are the top co-authors by number of shared publications:
- Firat Batmaz
Loughborough University
5 shared publications Eran A. Edirisinghe
Loughborough Univ. (United Kingdom)
4 shared publications- Gary Storey
Loughborough University
4 shared publications - Hossein Nevisi
Loughborough University
4 shared publications - Parisa Derakhshan
Loughborough University
4 shared publications - Roy S. Kalawsky
Loughborough University
4 shared publications
Latest Publications
- Open Access
- 17 September 2026
- Published by Springer Nature in Virtual Reality
Holographic conversational agents for education
Conversational agents in universities are mostly screen-based and lack the presence needed for confident, back-and-forth teaching. We introduce life-sized, AI-powered Holographic Conversational Agents (HCAs) that run on Proto displays and portray Ada Lovelace and Averroes for use in education and public engagement. The paper makes three contributions. First, it sets out a documented, reproducible pipeline that takes still or video assets to an interactive, source-grounded, safety-checked avatar suitable for live education. Second, it details an EDI-informed persona design, pairing a woman in computing with a non-Western scholar, with deliberate choices about voice, accent and knowledge sources to support inclusive engagement. Third, it reports a mixed-methods evaluation with staff and students in real settings, showing strong perceived usefulness, engagement and social presence, with clear intent to reuse in classes and exhibitions. Results confirmed high perceived value, across 51 participants, 85.7% of all ratings were positive; Engagement averaged 4.31/5 (85.3% positive) and recommendation 4.37/5 with 88.2% rating 4–5. We close with practical guidance for building new subject-specific HCAs from the pipeline and deployment notes for classrooms, labs and galleries, positioning life-sized holography as a viable format for dialogic teaching with historical figures, with perceived credibility supported through source-grounding and scope boundaries.
- 17 September 2026
- Published by Springer Nature in Virtual Reality
Open AccessHolographic conversational agents for education
Conversational agents in universities are mostly screen-based and lack the presence needed for confident, back-and-forth teaching. We introduce life-sized, AI-powered Holographic Conversational Agents (HCAs) that run on Proto displays and portray Ada Lovelace and Averroes for use in education and public engagement. The paper makes three contributions. First, it sets out a documented, reproducible pipeline that takes still or video assets to an interactive, source-grounded, safety-checked avatar suitable for live education. Second, it details an EDI-informed persona design, pairing a woman in computing with a non-Western scholar, with deliberate choices about voice, accent and knowledge sources to support inclusive engagement. Third, it reports a mixed-methods evaluation with staff and students in real settings, showing strong perceived usefulness, engagement and social presence, with clear intent to reuse in classes and exhibitions. Results confirmed high perceived value, across 51 participants, 85.7% of all ratings were positive; Engagement averaged 4.31/5 (85.3% positive) and recommendation 4.37/5 with 88.2% rating 4–5. We close with practical guidance for building new subject-specific HCAs from the pipeline and deployment notes for classrooms, labs and galleries, positioning life-sized holography as a viable format for dialogic teaching with historical figures, with perceived credibility supported through source-grounding and scope boundaries.
- Top Cited PapersOpen Access
- 7 March 2026
- Published by Springer Nature in Education and Information Technologies
- Vol. 31  (10) , 3035-3075
The AI scaffold and engagement spectrum as a novel UDL-aligned system for supporting students with dyslexia
- 7 March 2026
- Published by Springer Nature in Education and Information Technologies
- Vol. 31  (10) , 3035-3075
Top Cited PapersOpen AccessThe AI scaffold and engagement spectrum as a novel UDL-aligned system for supporting students with dyslexia
- Open Access
- 4 November 2025
- Published by Springer Nature in Research Square
Holographic Conversational Agents for Education
Conversational agents in universities are mostly screen-based and lack the presence needed for confident, back-and-forth teaching. We introduce life-sized, AI-powered Holographic Conversational Agents (HCAs) that run on Proto displays and portray Ada Lovelace and Averroes for use in education and public engagement. The paper makes three contributions. First, it sets out a documented, reproducible pipeline that takes still or video assets to an interactive, source-grounded, safety-checked avatar suitable for live education. Second, it details an EDI-informed persona design, pairing a woman in computing with a non-Western scholar, with deliberate choices about voice, accent and knowledge sources to support inclusive engagement. Third, it reports a mixed-methods evaluation with staff and students in real settings, showing strong perceived usefulness, engagement and social presence, with clear intent to reuse in classes and exhibitions. Results confirmed high perceived value, across 51 participants, 85.7\% of all ratings were positive; Engagement averaged 4.31/5 (85.3\% positive) and recommendation 4.37/5 with 88.2\% rating 4–5. We close with practical guidance for building new subject-specific HCAs from the pipeline and deployment notes for classrooms, labs and galleries, positioning life-sized holography as a viable format for authoritative, dialogic teaching with historical figures.
- 4 November 2025
- Published by Springer Nature in Research Square
Open AccessHolographic Conversational Agents for Education
Conversational agents in universities are mostly screen-based and lack the presence needed for confident, back-and-forth teaching. We introduce life-sized, AI-powered Holographic Conversational Agents (HCAs) that run on Proto displays and portray Ada Lovelace and Averroes for use in education and public engagement. The paper makes three contributions. First, it sets out a documented, reproducible pipeline that takes still or video assets to an interactive, source-grounded, safety-checked avatar suitable for live education. Second, it details an EDI-informed persona design, pairing a woman in computing with a non-Western scholar, with deliberate choices about voice, accent and knowledge sources to support inclusive engagement. Third, it reports a mixed-methods evaluation with staff and students in real settings, showing strong perceived usefulness, engagement and social presence, with clear intent to reuse in classes and exhibitions. Results confirmed high perceived value, across 51 participants, 85.7\% of all ratings were positive; Engagement averaged 4.31/5 (85.3\% positive) and recommendation 4.37/5 with 88.2\% rating 4–5. We close with practical guidance for building new subject-specific HCAs from the pipeline and deployment notes for classrooms, labs and galleries, positioning life-sized holography as a viable format for authoritative, dialogic teaching with historical figures.
- Open Access
- 30 October 2025
- Published by Springer Nature in Research Square
The AI Scaffold and Engagement Spectrum as a Novel UDL-Aligned System for Supporting Students with Dyslexia
Artificial Intelligence (AI) is rapidly changing higher education, offering new opportunities to support diverse learners, including students with dyslexia, while raising important questions around ethics and pedagogy. This study introduces a novel, research-informed framework that supports students with dyslexia in reading, writing, and academic workflow planning, grounded in the principles of Universal Design for Learning (UDL) and constructive alignment. Using mixed methods, including student surveys and interviews with the Student Support and Wellbeing Services (SSWS) team, we identify key barriers such as information overload, linear writing demands, and uncertainty around AI use. The framework offers a step-by-step guide for responsible AI use aligned with UDL's principles of multiple means of engagement, representation, and expression. This paper’s primary contribution is a novel, dual-component pedagogical system designed to resolve this uncertainty. It comprises an academic-facing model, \textit{the AI Engagement Spectrum}, which shifts responsibility for ethical AI use from student policing to proactive instructional design, and a student-facing workflow, the \textit{AI Scaffold} that applies UDL principles for AI-supported academic tasks. The framework provides a process that embeds ethical checkpoints designed to support AI literacy, ensure human-in-the-loop agency, and mitigate the risks of uncritical cognitive offloading. Our findings suggest that with the right design, AI can become a practical and equitable support strategy, enhancing autonomy while addressing concerns like skill atrophy and academic integrity.
- 30 October 2025
- Published by Springer Nature in Research Square
Open AccessThe AI Scaffold and Engagement Spectrum as a Novel UDL-Aligned System for Supporting Students with Dyslexia
Artificial Intelligence (AI) is rapidly changing higher education, offering new opportunities to support diverse learners, including students with dyslexia, while raising important questions around ethics and pedagogy. This study introduces a novel, research-informed framework that supports students with dyslexia in reading, writing, and academic workflow planning, grounded in the principles of Universal Design for Learning (UDL) and constructive alignment. Using mixed methods, including student surveys and interviews with the Student Support and Wellbeing Services (SSWS) team, we identify key barriers such as information overload, linear writing demands, and uncertainty around AI use. The framework offers a step-by-step guide for responsible AI use aligned with UDL's principles of multiple means of engagement, representation, and expression. This paper’s primary contribution is a novel, dual-component pedagogical system designed to resolve this uncertainty. It comprises an academic-facing model, \textit{the AI Engagement Spectrum}, which shifts responsibility for ethical AI use from student policing to proactive instructional design, and a student-facing workflow, the \textit{AI Scaffold} that applies UDL principles for AI-supported academic tasks. The framework provides a process that embeds ethical checkpoints designed to support AI literacy, ensure human-in-the-loop agency, and mitigate the risks of uncritical cognitive offloading. Our findings suggest that with the right design, AI can become a practical and equitable support strategy, enhancing autonomy while addressing concerns like skill atrophy and academic integrity.
- Open Access
- 7 October 2025
- Published by Springer Nature in Research Square
A Novel Framework for AI Storytelling to CreateEngaging Educational Narratives
This paper addresses the challenges of creating educational content by intro-ducing a novel AI Storytelling Framework that transforms curricular contentinto engaging narratives. The framework is structured as an eight-step process and is grounded in the Right Human-in-the-Loop (R-HiTL) paradigm, a structured theoretical model for responsibly integrating generative AI with right expert guidance in education. By embedding right experts at critical points,the framework mitigates AI’s known limitations and ensures narrative, ethical, and pedagogical integrity. We tested the framework in student workshops, which served as a testbed for a high-stakes case study: the development of an educational podcast series Named after Nelson (NaN): Stories about places namedafter Nelson Mandela. The NaN project is an impact project in collaboration with the Nelson Mandela Foundation (NMF). Findings indicate that while generative AI can accelerate content creation, its effective use in education depends on a human-centred framework that improves efficiency and inclusivity without compromising quality or accuracy. Our R-HiTL-guided approach exemplifies this, offering a structured and replicable methodology for responsible use of AI in educational content creation.
- 7 October 2025
- Published by Springer Nature in Research Square
Open AccessA Novel Framework for AI Storytelling to CreateEngaging Educational Narratives
This paper addresses the challenges of creating educational content by intro-ducing a novel AI Storytelling Framework that transforms curricular contentinto engaging narratives. The framework is structured as an eight-step process and is grounded in the Right Human-in-the-Loop (R-HiTL) paradigm, a structured theoretical model for responsibly integrating generative AI with right expert guidance in education. By embedding right experts at critical points,the framework mitigates AI’s known limitations and ensures narrative, ethical, and pedagogical integrity. We tested the framework in student workshops, which served as a testbed for a high-stakes case study: the development of an educational podcast series Named after Nelson (NaN): Stories about places namedafter Nelson Mandela. The NaN project is an impact project in collaboration with the Nelson Mandela Foundation (NMF). Findings indicate that while generative AI can accelerate content creation, its effective use in education depends on a human-centred framework that improves efficiency and inclusivity without compromising quality or accuracy. Our R-HiTL-guided approach exemplifies this, offering a structured and replicable methodology for responsible use of AI in educational content creation.
- Open Access
- 24 June 2023
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 83  (4) , 10717-10734
Deep labeller: automatic bounding box generation for synthetic violence detection datasets
Manually labelling datasets for training violence detection systems is time-consuming, expensive, and labor-intensive. Mind wandering, boredom, and short attention span can also cause labelling errors. Moreover, collecting and distributing sensitive images containing violence has ethical implications. Automation is the future for labelling sensitive image datasets. Deep labeller is a two-stage Deep Learning (DL) method that uses pre-trained DL object detection methods on MS-COCO for automatic labelling. The Deep Labeller method labels violent and nonviolent images in WVD and USI. In stage 1, WVD generates weak labels using synthetic images. In stage 2, the Deep labeller method is retrained on weak labels. USI dataset is used to test our method on real-world violence. Deep labeller generated weak and strong labels with an IoU of 0.80036 in stage 1 and 0.95 in stage 2 on the WVD. Automatically generated labels. To test our method’s generalisation power, violent and nonviolent image labels on USI dataset had a mean IoU of 0.7450.
- 24 June 2023
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 83  (4) , 10717-10734
Open AccessDeep labeller: automatic bounding box generation for synthetic violence detection datasets
Manually labelling datasets for training violence detection systems is time-consuming, expensive, and labor-intensive. Mind wandering, boredom, and short attention span can also cause labelling errors. Moreover, collecting and distributing sensitive images containing violence has ethical implications. Automation is the future for labelling sensitive image datasets. Deep labeller is a two-stage Deep Learning (DL) method that uses pre-trained DL object detection methods on MS-COCO for automatic labelling. The Deep Labeller method labels violent and nonviolent images in WVD and USI. In stage 1, WVD generates weak labels using synthetic images. In stage 2, the Deep labeller method is retrained on weak labels. USI dataset is used to test our method on real-world violence. Deep labeller generated weak and strong labels with an IoU of 0.80036 in stage 1 and 0.95 in stage 2 on the WVD. Automatically generated labels. To test our method’s generalisation power, violent and nonviolent image labels on USI dataset had a mean IoU of 0.7450.
- Open Access
Use of Artificial Intelligence to Improve Resilience and Preparedness Against Adverse Flood Events
The main focus of this paper is the novel use of Artificial Intelligence (AI) in natural disaster, more specifically flooding, to improve flood resilience and preparedness. Different types of flood have varying consequences and are followed by a specific pattern. For example, a flash flood can be a result of snow or ice melt and can occur in specific geographic places and certain season. The motivation behind this research has been raised from the Building Resilience into Risk Management (BRIM) project, looking at resilience in water systems. This research uses the application of the state-of-the-art techniques i.e., AI, more specifically Machin Learning (ML) approaches on big data, collected from previous flood events to learn from the past to extract patterns and information and understand flood behaviours in order to improve resilience, prevent damage, and save lives. In this paper, various ML models have been developed and evaluated for classifying floods, i.e., flash flood, lakeshore flood, etc. using current information i.e., weather forecast in different locations. The analytical results show that the Random Forest technique provides the highest accuracy of classification, followed by J48 decision tree and Lazy methods. The classification results can lead to better decision-making on what measures can be taken for prevention and preparedness and thus improve flood resilience.
Open AccessUse of Artificial Intelligence to Improve Resilience and Preparedness Against Adverse Flood Events
The main focus of this paper is the novel use of Artificial Intelligence (AI) in natural disaster, more specifically flooding, to improve flood resilience and preparedness. Different types of flood have varying consequences and are followed by a specific pattern. For example, a flash flood can be a result of snow or ice melt and can occur in specific geographic places and certain season. The motivation behind this research has been raised from the Building Resilience into Risk Management (BRIM) project, looking at resilience in water systems. This research uses the application of the state-of-the-art techniques i.e., AI, more specifically Machin Learning (ML) approaches on big data, collected from previous flood events to learn from the past to extract patterns and information and understand flood behaviours in order to improve resilience, prevent damage, and save lives. In this paper, various ML models have been developed and evaluated for classifying floods, i.e., flash flood, lakeshore flood, etc. using current information i.e., weather forecast in different locations. The analytical results show that the Random Forest technique provides the highest accuracy of classification, followed by J48 decision tree and Lazy methods. The classification results can lead to better decision-making on what measures can be taken for prevention and preparedness and thus improve flood resilience.
- Open Access
A Model-Based Engineering Methodology and Architecture for Resilience in Systems-of-Systems: A Case of Water Supply Resilience to Flooding
There is a clear and evident requirement for a conscious effort to be made towards a resilient water system-of-systems (SoS) within the UK, in terms of both supply and flooding. The impact of flooding goes beyond the immediately obvious socio-aspects of disruption, cascading and affecting a wide range of connected systems. The issues caused by flooding need to be treated in a fashion which adopts an SoS approach to evaluate the risks associated with interconnected systems and to assess resilience against flooding from various perspectives. Changes in climate result in deviations in frequency and intensity of precipitation; variations in annual patterns make planning and management for resilience more challenging. This article presents a verified model-based system engineering methodology for decision-makers in the water sector to holistically, and systematically implement resilience within the water context, specifically focusing on effects of flooding on water supply. A novel resilience viewpoint has been created which is solely focused on the resilience aspects of architecture that is presented within this paper. Systems architecture modelling forms the basis of the methodology and includes an innovative resilience viewpoint to help evaluate current SoS resilience, and to design for future resilient states. Architecting for resilience, and subsequently simulating designs, is seen as the solution to successfully ensuring system performance does not suffer, and systems continue to function at the desired levels of operability. The case study presented within this paper demonstrates the application of the SoS resilience methodology on water supply networks in times of flooding, highlighting how such a methodology can be used for approaching resilience in the water sector from an SoS perspective. The methodology highlights where resilience improvements are necessary and also provides a process where architecture solutions can be proposed and tested.
Open AccessA Model-Based Engineering Methodology and Architecture for Resilience in Systems-of-Systems: A Case of Water Supply Resilience to Flooding
There is a clear and evident requirement for a conscious effort to be made towards a resilient water system-of-systems (SoS) within the UK, in terms of both supply and flooding. The impact of flooding goes beyond the immediately obvious socio-aspects of disruption, cascading and affecting a wide range of connected systems. The issues caused by flooding need to be treated in a fashion which adopts an SoS approach to evaluate the risks associated with interconnected systems and to assess resilience against flooding from various perspectives. Changes in climate result in deviations in frequency and intensity of precipitation; variations in annual patterns make planning and management for resilience more challenging. This article presents a verified model-based system engineering methodology for decision-makers in the water sector to holistically, and systematically implement resilience within the water context, specifically focusing on effects of flooding on water supply. A novel resilience viewpoint has been created which is solely focused on the resilience aspects of architecture that is presented within this paper. Systems architecture modelling forms the basis of the methodology and includes an innovative resilience viewpoint to help evaluate current SoS resilience, and to design for future resilient states. Architecting for resilience, and subsequently simulating designs, is seen as the solution to successfully ensuring system performance does not suffer, and systems continue to function at the desired levels of operability. The case study presented within this paper demonstrates the application of the SoS resilience methodology on water supply networks in times of flooding, highlighting how such a methodology can be used for approaching resilience in the water sector from an SoS perspective. The methodology highlights where resilience improvements are necessary and also provides a process where architecture solutions can be proposed and tested.
- Open Access
- 28 June 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 6, 38399-38410
A Systems Engineering Hackathon – A Methodology Involving Multiple Stakeholders to Progress Conceptual Design of a Complex Engineered Product
This paper describes a novel hackathon-style system engineering process and its value as an agile approach to the rapid generation and development of early design concepts of complex engineered products-in this case a future aircraft. Complex product design typically requires a diverse range of stakeholders to arrive at a consensus of key decision criteria and design factors, which requires effective articulation and communication of information across traditional engineering and operational disciplines. The application of the methodology is highlighted by means of a case study inspired by Airbus where stakeholder involvement and internal collaboration among team members were essential to achieve a set of agreed goals. This paper shows that a hackathon grounded on systems engineering approaches and structured around the technical functions within an engineering company has the capability and capacity to communicate a coherent vision and rationale for the conceptual design of a complex engineered product. The hackathon method offers significant benefits to these stakeholders to better manage, prioritize, and decrease excessive complexities in the overall design process. A significant benefit of this agile process is that it can achieve useful results in a very short timeframe (i.e., 80% reduction), where it could take up to a year to accomplish compared with using current/regular internal methods.
- 28 June 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 6, 38399-38410
Open AccessA Systems Engineering Hackathon – A Methodology Involving Multiple Stakeholders to Progress Conceptual Design of a Complex Engineered Product
This paper describes a novel hackathon-style system engineering process and its value as an agile approach to the rapid generation and development of early design concepts of complex engineered products-in this case a future aircraft. Complex product design typically requires a diverse range of stakeholders to arrive at a consensus of key decision criteria and design factors, which requires effective articulation and communication of information across traditional engineering and operational disciplines. The application of the methodology is highlighted by means of a case study inspired by Airbus where stakeholder involvement and internal collaboration among team members were essential to achieve a set of agreed goals. This paper shows that a hackathon grounded on systems engineering approaches and structured around the technical functions within an engineering company has the capability and capacity to communicate a coherent vision and rationale for the conceptual design of a complex engineered product. The hackathon method offers significant benefits to these stakeholders to better manage, prioritize, and decrease excessive complexities in the overall design process. A significant benefit of this agile process is that it can achieve useful results in a very short timeframe (i.e., 80% reduction), where it could take up to a year to accomplish compared with using current/regular internal methods.
- Open Access
- 13 May 2015
- Published by SPIE-Intl Soc Optical Eng
- Vol. 9481, 94810R-94810R-11
Illumination modelling of a mobile device environment for effective use in driving mobile apps
Copyright 2015 Society of Photo Optical Instrumentation Engineers. One print or electronic copy may be made for personal use only. Systematic electronic or print reproduction and distribution, duplication of any material in this paper for a fee or for commercial purposes, or modification of the content of the paper are prohibited. This item is also available here: http://dx.doi.org/10.1117/12.2177087The present generation of Ambient Light Sensors (ALS) of a mobile handheld device suffer from two practical shortcomings. The ALSs are narrow angle, i.e. they respond effectively only within a narrow angle of operation and there is a latency of operation. As a result mobile applications that operate based on the ALS readings could perform sub-optimally especially when operated in environments with non-uniform illumination. The applications will either adopt with unacceptable levels of latency or/and may demonstrate a discrete nature of operation. In this paper we propose a framework to predict the ambient illumination of an environment in which a mobile device is present. The predictions are based on an illumination model that is developed based on a small number of readings taken during an application calibration stage. We use a machine learning based approach in developing the models. Five different regression models were developed, implemented and compared based on Polynomial, Gaussian, Sum of Sine, Fourier and Smoothing Spline functions. Approaches to remove noisy data, missing values and outliers were used prior to the modelling stage to remove their negative effects on modelling. The prediction accuracy for all models were found to be above 0.99 when measured using R-Squared test with the best performance being from Smoothing Spline. In this paper we will discuss mathematical complexity of each model and investigate how to make compromises in finding the best mode
- 13 May 2015
- Published by SPIE-Intl Soc Optical Eng
- Vol. 9481, 94810R-94810R-11
Open AccessIllumination modelling of a mobile device environment for effective use in driving mobile apps
Copyright 2015 Society of Photo Optical Instrumentation Engineers. One print or electronic copy may be made for personal use only. Systematic electronic or print reproduction and distribution, duplication of any material in this paper for a fee or for commercial purposes, or modification of the content of the paper are prohibited. This item is also available here: http://dx.doi.org/10.1117/12.2177087The present generation of Ambient Light Sensors (ALS) of a mobile handheld device suffer from two practical shortcomings. The ALSs are narrow angle, i.e. they respond effectively only within a narrow angle of operation and there is a latency of operation. As a result mobile applications that operate based on the ALS readings could perform sub-optimally especially when operated in environments with non-uniform illumination. The applications will either adopt with unacceptable levels of latency or/and may demonstrate a discrete nature of operation. In this paper we propose a framework to predict the ambient illumination of an environment in which a mobile device is present. The predictions are based on an illumination model that is developed based on a small number of readings taken during an application calibration stage. We use a machine learning based approach in developing the models. Five different regression models were developed, implemented and compared based on Polynomial, Gaussian, Sum of Sine, Fourier and Smoothing Spline functions. Approaches to remove noisy data, missing values and outliers were used prior to the modelling stage to remove their negative effects on modelling. The prediction accuracy for all models were found to be above 0.99 when measured using R-Squared test with the best performance being from Smoothing Spline. In this paper we will discuss mathematical complexity of each model and investigate how to make compromises in finding the best mode