An Online Framework for Supporting the Evaluation of Personalised Information Retrieval Systems
Catherine Mulwa, Luca Longo, Séamus Lawless, Mary Sharp, Vincent Wade · Electronic workshops in computing · 2011
Scope - Personalised Information Retrieval (PIR) has been gaining attention because it investigates intelligent ways for enhancing content delivery. Web users can have personalised services and more accurate information. Problem - Several PIR systems have been proposed in the literature; however, they have not been properly tested or evaluated. Proposal – The authors propose a generally applicable web-based interface, which provides PIR developers and evaluators with: i) implicit recommendations on how to evaluate a specific PIR system; ii) a repository containing studies on user-centred and layered evaluation studies; iii) recommendations on how to best combine different evaluation methods, metrics and measurement criteria in order to most effectively evaluate their system; iv) a UCE methodology which details how to apply existing UCE techniques; v) a taxonomy of evaluations of adaptive systems; and vi) interface translation support (49 languages supported).