WP4: Realisation of Context Awareness enablers

Context Management in general

Here, context is any data linked to an entity, and an entity is anything that has context. Examples may include the localisation of a person, the number of persons in a room, the noise level in an area, or the favorite web site of an internaut.

Context management is the set of techniques and technologies used to collect, store and provide access to this context.

Collecting context starts with low-level data, using sensors, such as microphones in an area or GPS units on a person, or even posts on blogs that provide semantic data about users
Higher-level data is obtained either directly by existing sources, for example datamining in a train management database to get the most visited stations, or else deduced by the system based on low-level data. We call the latter technique context enhancement, it is generally done by combining several low-level sources with static repositories to establish semantic values; for example, one might combine person location, weather from an internet service, and the person’s agenda to deduce information about the person’s mood.
Providing the context means that APIs have to be created to enable applications to feed upon the context to adapt their behavior; for example, a warehouse microphone might switch off when a truck is approaching because of the noise it will generate.

Context Management with regard to Multi-Modality

Context management makes it possible to go further than proposing multiple modalities with a static choice, or dynamic choice of modalities. It enables the modalities to be activated or deactivated depending on context. For example, a camera might detect that there is not enough light for gesture recognition; the light will be managed as context, and when the multi-modal engine is notified that it is dark, it deactivates the camera, activates a microphone or a screen, and notifies the end-user of the change of modality and its cause.

Achievements

All partners were made aware of the principles of context management and the opportunities it offers. This is an internal dissemination result linked to WP7.

A Context Management Framework (CMF) was developed in WP4 based on prior art (in particular several European projects), partner specific requirements, and a need for genericity to enable re-use and extension.

Context sources from the consortium were integrated into the CMF. This includes voice analysis results, video analysis results, and geolocation data.

The CMF was integrated with the Multi-Modal Engine (MME) in order to propose a generic enabler to be specialized for each partner’s needs.

The CMF was integrated in several demonstrators, in particular the Public Safety Demonstrator and the Construction demonstrator.

Adapting to recent challenges, WP4 also started studying social context, in particular based on the analysis of social data available through social networks (Facebook, Twitter, etc.). A paper was published (see WP7 results), integration of such data in the CMF is ongoing.