Laboratoire d'Informatique de Grenoble Équipe Ingénierie de l'Interaction Humain-Machine

Équipe Ingénierie de l'Interaction

Fuzzy4U : a Human-Computer Interfaces adaptation system in fuzzy logic for accessibility

162 pages. 2020.

Tanguy Giuffrida


With the massive spread of the use of the Internet and the computer tool in general, the accessibility of user interfaces is an increasingly pressing need. Ease of use of computing devices is indeed a critical point in fostering adoption. Thus, the difficulties experienced by certain users, such as the elderly or in a situation of neurocognitive difficulties, require more acutely the question of accessibility. Much work has focused on this subject in order to define generic or situational accessibility recommendations. One possible approach is to adapt the interfaces to the context of use. However, difficulties remain, particularly linked to the complexity of the possible contexts of use, leading to a multiplicity of characteristics to be taken into account, an imprecision of the values assigned to these characteristics and the combination of multiple adaptation rules. We address these three obstacles through the following research question: how to improve the accessibility of UIs through adaptive interfaces by taking into account the multiplicity of context characteristics, the imprecision of context values ; as well as the complexity and the combination of rules?By relying on a generic adaptation architecture, we propose a dynamic adaptation engine based on fuzzy logic, used to implement accessibility recommendations. We show how fuzzy logic manages the three obstacles identified through a prototype. We will highlight these advantages compared to a classical approach in Boolean logic. In order to validate these advantages, we carried out a comparative analysis between the 2 types of rules (fuzzy and binary logic) in a theoretical way and with an evaluation by experts. The results validate the interest of fuzzy logic for adapting interfaces for accessibility purposes.