Malick Ebiele
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Malick Ebiele has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Malika Bendechache
University of Galway
2 shared publications - Rob Brennan
University College Dublin
2 shared publications Brian Doyle
Centre for Innovative Human Systems, Trinity College, University of Dublin
1 shared publication- Lucy McKenna
Dublin City University
1 shared publication - Marie Elizabeth Ward
Dublin Trinity College Dublin School of Medicine
1 shared publication Nick McDonald
Centre for Innovative Human Systems, Trinity College, University of Dublin
1 shared publication
Latest Publications
- Open Access
- 24 June 2025
- Published by Association for Computing Machinery (ACM) in Journal of Data and Information Quality
- Vol. 17 (2) , 1-39
Quantitative Data Valuation Methods: A Systematic Review and Taxonomy
Background: Data valuation is an area of study encompassing but not limited to data quality, machine learning, applied energy, and information economics. The primary focus of data valuation research is the development of methodologies for determining the value (relevance or worth) of data. Quantitative data valuation methods are techniques that assign a numerical value to data. Method: The methodology used for this systematic review complies with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. A search string was run on five databases: IEEE, ACM, Web of Science, Science Direct, and AISeL. Studies introducing at least one new quantitative data valuation method and published before the 29th of February 2024 were extracted and analysed. Results: In total 75 studies have been included in this review paper. Quantitative data valuation research is mainly driven by the Machine Learning (including federated learning and supervised learning) and Data Management communities, which account for 48% (36 studies) and 20% (15 studies), respectively, of the included studies. In the last 3 years (2021–2023), the number of publications has at least increased by a factor of 1.5 every year, mainly driven by the Machine Learning community. Discussion: Under the current data deluge, quantitative data valuation has become crucial for effective data management in organisations, for example to enable prioritisation of data management activities. In parallel, the rapid rise in interest in explainable and federated machine learning depends on quantitative data valuation methods to identify the most relevant data to explain model outcomes and the most suitable federated clients to peer with. These two trends ensure that quantitative data valuation publications will continue to increase for many years to come. The data valuation methods favoured by the data management and machine learning communities overlap but also diverge so it is important to include both perspectives to have a comprehensive view of data value, data valuation, and their relationship to data quality. Conclusion: According to the studies included in this review, quantitative data valuation methods have been around for at least 25 years (2000–2024). However, such methods have received an increasing interest in the last 6 years mainly driven by the Machine Learning community exploring automated methods to select or focus attention on highly valuable (i.e., relevant) data for a model or the explanation of a model’s outputs. Quantitative data valuation has many application areas including but not limited to data management, military operations, system security, and economics. Other: The multidisciplinary nature of data valuation as a research topic has resulted in a lack of consensus on a definition of the term “data value”, with many domain-specific definitions emerging. After analysing the data value and data valuation definitions found in our literature search, we propose here a cross-domain definition of the term data value. We have also proposed a taxonomy of quantitative data valuation methods based on input types (data, metadata, bimodal) and application context to show the variety of techniques used in this field.
- 24 June 2025
- Published by Association for Computing Machinery (ACM) in Journal of Data and Information Quality
- Vol. 17 (2) , 1-39
Open AccessQuantitative Data Valuation Methods: A Systematic Review and Taxonomy
Background: Data valuation is an area of study encompassing but not limited to data quality, machine learning, applied energy, and information economics. The primary focus of data valuation research is the development of methodologies for determining the value (relevance or worth) of data. Quantitative data valuation methods are techniques that assign a numerical value to data. Method: The methodology used for this systematic review complies with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. A search string was run on five databases: IEEE, ACM, Web of Science, Science Direct, and AISeL. Studies introducing at least one new quantitative data valuation method and published before the 29th of February 2024 were extracted and analysed. Results: In total 75 studies have been included in this review paper. Quantitative data valuation research is mainly driven by the Machine Learning (including federated learning and supervised learning) and Data Management communities, which account for 48% (36 studies) and 20% (15 studies), respectively, of the included studies. In the last 3 years (2021–2023), the number of publications has at least increased by a factor of 1.5 every year, mainly driven by the Machine Learning community. Discussion: Under the current data deluge, quantitative data valuation has become crucial for effective data management in organisations, for example to enable prioritisation of data management activities. In parallel, the rapid rise in interest in explainable and federated machine learning depends on quantitative data valuation methods to identify the most relevant data to explain model outcomes and the most suitable federated clients to peer with. These two trends ensure that quantitative data valuation publications will continue to increase for many years to come. The data valuation methods favoured by the data management and machine learning communities overlap but also diverge so it is important to include both perspectives to have a comprehensive view of data value, data valuation, and their relationship to data quality. Conclusion: According to the studies included in this review, quantitative data valuation methods have been around for at least 25 years (2000–2024). However, such methods have received an increasing interest in the last 6 years mainly driven by the Machine Learning community exploring automated methods to select or focus attention on highly valuable (i.e., relevant) data for a model or the explanation of a model’s outputs. Quantitative data valuation has many application areas including but not limited to data management, military operations, system security, and economics. Other: The multidisciplinary nature of data valuation as a research topic has resulted in a lack of consensus on a definition of the term “data value”, with many domain-specific definitions emerging. After analysing the data value and data valuation definitions found in our literature search, we propose here a cross-domain definition of the term data value. We have also proposed a taxonomy of quantitative data valuation methods based on input types (data, metadata, bimodal) and application context to show the variety of techniques used in this field.
- Open Access
- 12 June 2025
- Published by Taylor & Francis in Ergonomics
- Vol. 69 (7) , 1-23
Trustworthy artificial intelligence and organisational trust: a scoping review using socio-technical systems analysis
Organisational trust enhances effectiveness, but its evaluation is often overlooked in real-world AI deployments. The relationship between trustworthy AI (TAI) and organisational trust is largely unexplored. This scoping review applies socio-technical systems analysis (STSA) to examine TAI mechanisms and their ability to build organisational trust, marking the first use of STSA in this context. A search of academic databases yielded 803 articles, with 54 included in the review. The findings show a lack of focus on the context, longitudinal aspects and organisational maturity in TAI implementation. While TAI guidelines aim to foster trust in specific technologies, successful implementation requires broader organisational maturity and active efforts to build, maintain and repair trust both within organisations and with the public.
- 12 June 2025
- Published by Taylor & Francis in Ergonomics
- Vol. 69 (7) , 1-23
Open AccessTrustworthy artificial intelligence and organisational trust: a scoping review using socio-technical systems analysis
Organisational trust enhances effectiveness, but its evaluation is often overlooked in real-world AI deployments. The relationship between trustworthy AI (TAI) and organisational trust is largely unexplored. This scoping review applies socio-technical systems analysis (STSA) to examine TAI mechanisms and their ability to build organisational trust, marking the first use of STSA in this context. A search of academic databases yielded 803 articles, with 54 included in the review. The findings show a lack of focus on the context, longitudinal aspects and organisational maturity in TAI implementation. While TAI guidelines aim to foster trust in specific technologies, successful implementation requires broader organisational maturity and active efforts to build, maintain and repair trust both within organisations and with the public.