Yazid Benazzouz, Philippe Beaune, Fano Ramaparany, and Olivier


Abstract Servi es need to be adapted to users and auto-ongured for optimizing their fun tioning. Adaptation requires servi es to be aware of

their ontexts. However, making ontext-aware servi es osts many eorts

from designers and a deli ate setting. We propose to automate this pro ess

by identifying relevant ontexts for servi es and then adapting these servi es

a ordingly. To a omplish our purpose, we explore a ontext data-driven

approa h. This approa h refers to the pro ess of olle ting and storing data

from a wide range of sour es. Data are gathered in onformity with a ontext

data model, stored in databases and then ontext re ognition te hniques are

applied to identify relevant ontexts. Our proposal diers from the existing

approa hes for ontext-aware appli ations, where ontext models and appli-

ations are losely related and ignore how ontext is derived from sour es

and interpreted. Our approa h is supported by an overall ar hite ture for

servi e adaptation and intends to automati ally adapt servi es without any

prior knowledge about ontext models.