| C. Troussas, C. Papakostas, A. Krouska, Ph. Mylonas, C. Sgouropoulou |
| FASTER-AI: A Comprehensive Framework for Enhancing the Trustworthiness of Artificial Intelligence in Web Information Systems |
| 20th International Conference on Web Information Systems and Technologies (WEBIST 2024), 17-19 November 2024, Porto, Portugal |
|
ABSTRACT
|
| With increasing embedding of artificial intelligence (AI) in web information systems (WIS), the maximum assurance on the reliability of such AI systems is solicited. Although this aspect is gaining importance, no comprehensive framework has yet been developed to ensure AI reliability. This paper aims to bridge that gap by proposing the AI FASTER framework to enhance the reliability of AI in WIS. The key dimensions of concern within the framework are FASTER-AI: Fairness/bias mitigation, explainability/transparency, security/privacy, robustness, and ethical considerations/accountability. Each one guides in precisely the area where trust shall be accomplished: a decrease in bias, model interpretability, protection of data, resilience of models, and ethics in governance. The implementation methodology for these dimensions involves preliminary assessment, planning, integration, testing, and continuous improvement. Validation of proof for FASTER-AI was created based on in-depth case studies across different verticals: e-commerce, finance, health, and fraud detection. This work has demonstrated how FASTER-AI is applied through illustrative case studies showing promising performance. From the initial results of high improvement in terms of fairness, transparency, security, and robustness, it may be effectively inferred that FASTER-AI can be successfully applied.
|
| 17 November , 2024 |
| C. Troussas, C. Papakostas, A. Krouska, Ph. Mylonas, C. Sgouropoulou, "FASTER-AI: A Comprehensive Framework for Enhancing the Trustworthiness of Artificial Intelligence in Web Information Systems", 20th International Conference on Web Information Systems and Technologies (WEBIST 2024), 17-19 November 2024, Porto, Portugal |
[
BibTex] [
Print] [
 Back] |
