Opportunistic Composition of Human- Opportunistic composition - PowerPoint PPT Presentation

Opportunistic Composition of Human- Opportunistic composition Computer Interactions in Ambient Spaces Ambient env. Our approach Plastic HCI Augustin Degas Requirements Ergonomic criteria Sylvie Trouilhet - Jean-Paul Arcangeli, IRIT Control of the Gaëlle Calvary - Joëlle Coutaz, LIG emergence of Stéphane Lavirotte - Jean-Yves Tigli, I3S assembly Multi-agent System Service agent Context agent Conclusion 1

Ambient Environnement • Fast changing software environment Opportunistic • Needs composition • Evolution of the context of use (platform, user, Ambient env. environment) Our approach • Opening and dynamics of the context Plastic HCI • A growing number of available devices in a connected Requirements Ergonomic criteria world • Smartphone, camera … Control of the emergence of assembly Multi-agent System Service agent Context agent Conclusion project 2

Component composition • Software bricks: a software component Opportunistic composition Ambient env. Provided service Our approach Plastic HCI Required service Requirements Ergonomic criteria • Application: assembly of components Control of the emergence of assembly Multi-agent System Service agent Context agent Conclusion 3

Our opportunistic approach • Opportunistic component (re)composition Opportunistic • Use components available in the environment, that can composition be assembled, in order to suggest to users appropriate Ambient env. applications Our approach • Take advantage of the circumstances to produce Plastic HCI Requirements applications: Ergonomic criteria • Useful • According to the context Control of the • Continuously (recomposition with new opportunities) emergence of assembly • Bottom-up software development • No pre-established plan Multi-agent System Service agent Context agent Conclusion 4

When and how compose? • 5 functional requirements Opportunistic • User request of a component composition • Context evolution Ambient env. • Appearance of a component Our approach • Disappearance of a component Plastic HCI • Update Requirements Ergonomic criteria Control of the • 7 extra-functional requirements emergence of • Decentralization assembly • Dynamic adaptation • Combinatorial optimization Multi-agent System • Recomposition Service agent • Learning and context awareness Context agent • Utility of the result Conclusion • Non dependent of user needs 5

Plastic HCI • Design time – help the developer Opportunistic • Maximize the reuse for each new target composition • Reduce the development cost Ambient env. • Take into account the use Our approach • Execution – help the end-user Plastic HCI • Distribute an application or a task on multiple devices or Requirements Ergonomic criteria another device • Keep ease of use and habit, ensuring also to adapt to the Control of the specificities of the devices emergence of assembly → Common preoccupations Software Development - HCI: Multi-agent System Service agent Mobility Context agent Adaptation to users Conclusion • Specificity: User role 6

An exemple of adaptation of HCI Opportunistic composition Ambient env. Our approach Plastic HCI Requirements Ergonomic criteria Control of the emergence of assembly Multi-agent System Service agent Context agent Conclusion 7

5 requirements of the field • Ergonomics of the resulting application Opportunistic • Application that respects ergonomic criteria composition Ambient env. • Continuous adaptation Our approach Plastic HCI • Controllability of the composition Requirements • Control by the user of his interactive ambient Ergonomic criteria environment Control of the • Non-intrusivity of the composition emergence of assembly • User assistance • Non-expert users Multi-agent System • Time saving Service agent Context agent • Response time Conclusion 8

Ergonomic criteria • Taking into account the “Significance of the Opportunistic denominations and behaviours” composition Ambient env. Our approach Plastic HCI Requirements Ergonomic criteria Control of the emergence of assembly Multi-agent System Service agent Context agent Conclusion 9

Ergonomic criteria • Second possible composition Opportunistic composition Ambient env. Our approach Plastic HCI Requirements Ergonomic criteria Control of the emergence of assembly Multi-agent System Service agent Context agent Conclusion 10

Ergonomic criteria • Third possible composition Opportunistic composition Ambient env. Our approach Plastic HCI Requirements Ergonomic criteria Control of the emergence of assembly Multi-agent System Service agent Context agent Conclusion 11

Ergonomic criteria • Fourth possible composition Opportunistic composition Ambient env. Our approach Plastic HCI Requirements Ergonomic criteria Control of the emergence of assembly Multi-agent System Service agent Context agent Conclusion • Four functionally equivalent solutions • Two solutions that satisfy the ergonomic requirement 12

Control of the emergence of assembly • Two types of components available: Opportunistic • Functional components composition • User Interface components Ambient env. Our approach Plastic HCI • A same composition process: Requirements • Try to compose what is Ergonomic criteria – Available Control of the – Connectable emergence of – Might be useful assembly Multi-agent System Service agent Context agent Conclusion 13

A three component model • Selection of services: opportunism for the control of the Opportunistic consistency composition • Connector factory Ambient env. • Semantic binding Our approach Plastic HCI • Assembly emergence: opportunism for the exploration of Requirements Ergonomic criteria the utility • Adaptive Multi-Agent System Control of the • Learning of previous connections emergence of assembly • Controllability Multi-agent System • Meta-HCI for the control of the composition by the user Service agent Context agent Conclusion 14

Meta-HCI • Set of functions that are necessary and sufficient to Opportunistic controle and evaluate the state of the interactive ambient composition space [Coutaz, 06] [Coutaz et al ., 16] Ambient env. Our approach • Used differently: • Catalogue of estimated relevant composition Plastic HCI • Validation of the composition of HCI Requirements Ergonomic criteria Control of the emergence of assembly Multi-agent System Service agent Context agent Conclusion 15

The multi-agent system • Agent Opportunistic • Autonomous software entity composition • Evolves in a software environment Ambient env. • Has its own goal Our approach • Life Cycle: Perception, Decision, Action Plastic HCI • SMA Requirements Ergonomic criteria Control of the emergence of assembly • AMAS Multi-agent System Service agent Context agent [Boes, 14] Conclusion 16

Service agent • An AMAS as composition engine Opportunistic • Composition based on the cooperation of agents composition • Agentification of component services Ambient env. Our approach Plastic HCI Requirements Ergonomic criteria Control of the emergence of assembly Multi-agent System Service agent Context agent composant 1 composant 2 composant 3 Conclusion composant 2 composant 2 17

Service agent • Local goal: connect at best Opportunistic • Life cycle composition Ambient env. • Perception Our approach - Messages of other agent services that want to be connected Plastic HCI Requirements - State of the neighbours Ergonomic criteria - Its state • Decision Control of the - Announce, Response, Connection, Disconnection, Do emergence of nothing assembly • Action Multi-agent System Service agent Context agent Conclusion 18

Context agent • Agentification of the knowledge Opportunistic • Learn the good composition, in terms of context and composition ergonomics Ambient env. • Thanks to the validation of the user in the meta-HCI Our approach Plastic HCI Requirements Ergonomic criteria Control of the emergence of [Boes, 14] assembly Multi-agent System Service agent Context agent Conclusion 19

Prototype • An AMAS in Java Opportunistic • A class by agent composition Ambient env. • That handle the connections of software components Our approach • WComp • UPnP Plastic HCI Requirements • Integrated in the execution environment Wcomp Ergonomic criteria Control of the emergence of assembly • To be experimented and evaluated Multi-agent System Service agent Context agent Conclusion 20

Conclusion • A novel approach for adaptive software engineering Opportunistic • Component-based composition • Bottom-up Ambient env. • Context-sensitive Our approach • Emergence and plasticity of applications and user Plastic HCI interfaces Requirements Ergonomic criteria Control of the emergence of assembly Multi-agent System Service agent Context agent Conclusion 21