Adaptive systems based on user model are systems that create a digital representation of the user and use this model for adapting the content or the interface. Recommender systems are the most-well known example of content adaptation.
In particular, I'm working on a meta-model of the modeling process with the goal to create a holistic representation ofuser that puts together:
i) data from user beavour on real life that can be gathered from sensors from wearable devices and ambient intelligences,
ii) data from web and social networks user behaviours, in order to infer complex user features (such as habits, cognitive abilities, emotional states), ..and provide complex recommendations regarding real life such as what to do, goals to reach..
which considers all these different user features, their relations and conflicts, constraints.
Such model is particular useful for providing adaptive support to fragile people (i.e. people with cognitive disabilities, such as autistic people, people with dementia, etc ): in fact, for these people it is necessary to provide the right recommendations and avoid providing wrong suggestions, which can cause stress and anxiety, or irritation and anger with unpredictable consequences.
Moreover, to provide a real support in real life, we need many different information, much more than that needed for "neurotypical" individuals.
Traditional information about interests and preferences should necessary be combined with aversions, habits, idiosyncrasies, cognitive skills and abilities, emotions.
Currently, I'm working on applying this paradigm to a real use case of a project, PIUMA (Personalized Interactive Urban Maps for Autism), a three-year project founded by Compagnia San Paolo, with the goal provide an adaptive support in the movement in the city to people with autistic spectrum disorder.
In particular, in case of breakdown of the routine, the system will provide i) personalized recommendations of safe place to reach, ii) intelligent safe routing to get the place through crowdsourced interactive maps.