What should adaptivity mean to interactive software programmers?

Abstract

International audienceWorks about adaptability and adaptivity in interactive systems cover very different issues (user adaptation, context-aware systems, ambient intelligence, ubiquitous computing), not always with the explicit goal of supporting programmers. Based on examples that highlight how weakly discriminative the present terminology is, we propose to separate two concerns: adaptivity as a purely analytical concept, relative to a given viewpoint on the software rather than to its very structure, and its programming as a non specific case of reactive behavior. We describe how simple adaptive behaviors can be programmed with simple interactive behavior patterns, and how more complex patterns can be introduced for intelligent adaptation. Finally we describe an application where, relying on the principles exposed in this paper, interaction and adaptation are combined in a simple and innovative manner

Djnn/Smala

“…First, the reduction of the number of basic concepts might diminish the complexity sometimes induced by the combination of several paradigms. Second, as it has been shown in [31], it allows to model complex interactive software, such as adaptive user interfaces. Finally, it provides a basis for the formal verification of new properties such as the visibility of a graphic object [11].…”

Dynamic user interface adaptation driven by physiological parameters to support learning

“…Such data is actually sent to the user model manager by dedicated pieces of software, which is connected to the peripherals, elaborates the data and sends them few times per second to the user model manager (4). When one or more of the subscribed events and conditions occur, the user model manager notifies the adaptation engine (5). The adaptation engine is aware whether one or more notification is associated to the triggering of adaptation rules concerning the current application (because it hosts an association table for this purpose).…”

“…Current screen coordinates observed by the user (detected by the eye tracker); Total time that the user has spent looking towards the display, but outside its borders in the last three minutes (detected by analyzing the history of eye tracker out-of-screen coordinates); Number of eye blinks in the last three minutes (blinks are detected by Neurosky biosensor); Currently observed paragraph element (each titled section of the document is considered to be a paragraph), which is highlighted in real time with a green border; Amount of time during which the user has observed a certain paragraph, filtered to avoid that short fixations 5 We wrote 3 pairs of complementary rules, in the following indicated as A/A', B/B' and C/C' (as shown in Table 1): the first one, named "Long Total Time Out Screen" (A) is triggered if the user has been looking outside the screen for more than 10 seconds. The complementary one, "Low Total Time Out Screen" (A') is triggered when the user has looked outside the screen for less than 4 seconds.…”

“…Regarding adaptive user interfaces however relevant information is limited to what can affect the current user activity [5]. In our investigation, to provide real time adaptation based on the mental state, we thus focus on contextual parameters that arise from user behaviour monitoring.…”

“…As stated in [5], the job of adaptive application programmers is basically to create mappings between events and transformations. This is done according to the "event-control-transformation" schema, consistent with the "event-condition-action" paradigm that we adopt.…”