Astute symposium 2013-10-10_context_modelling_reasoning_wimboffe_annahristoskova

CONTEXT MODELLING & REASONING IN

SUPPORT OF PRO-ACTIVE & CONTEXT-

AWARE APPLICATIONS

Wim Boffé – IOS International

Anna Hristoskova – IBCN, iMinds

ASTUTE APPROACH

Large industrially-driven EU project

Development of • pro-active decision support • for data intensive environments • supported by HMI interfaces &

reasoning engines.

Focus on seamlessly accessing • relevant multimodal information • in a context-sensitive way • improving the way users deal

with complex situations and huge information quantities

• during real-time operations.

ASTUTE Symposium - Human factors and technologies for pro-active, context-aware and data-intensive applications

2

HORIZONTAL STRATEGY

VER

TIC

AL

STR

ATE

GY

GENERAL

SPECIFIC

Users Environment

Adaptive HMI Engine

Proactive Decision Support Engine

Context Engine

Physical world

Cross Domain Applications

Smart emergency dispatching

system

Intelligent cockpit: support for anticipation

Virtual control room: process management

Embedded driver

infotainment system

COMMON POINTS

COMMON PLATFORM


ASTUTE RA

3


SITUATIONAL AWARENESS

DURING EMERGENCY RESPONSE

WHO?

HOW?

WHAT? WHERE?

WHAT?

Fire Commander

Coordination of

Evacuation

Toxic smoke

formation

Site on fire

4


CONTEXT MODELLING METHOD

Context definition by domain experts • Field studies execution • Mental models creation • Goal-directed task analysis

Formal translation into an ontology & rules by ontology engineers in collaboration with domain experts • long track record of developing ICT solutions for

emergency management • accurate and complete reflection of the daily work

practices

Deployment of the architecture by domain experts Processing of usability feedback 5


FIELD STUDIES

In what environment do the users work?

What are the users’ goals? What information is needed?

6

TASK ANALYSIS RESULTING IN

MENTAL MODELS


7


ONTOLOGY DEVELOPMENT

METHODOLOGY

5 life cycle stages • Specification – define scope through various

scenarios based on the input from the field studies • Conceptualization – construct conceptual model

through extraction of concepts, relations and properties from scenarios

• Formalization – translate into a formal model through addition of rules and axioms restricting the possible interpretations

• Implementation – translate conceptual graphs and rules into a formal ontology language

• Maintenance – continuous evaluation, update and correction

8


EMERGENCY RESPONSE

ONTOLOGY Identification of knowledge within multiple context-aware application domains capturing generic data on devices, sensors, companies, people and the roles they perform.

Modeling of dedicated ontologies consisting of two layers: • 8 High-Level application-independent ontologies

Generic definition of persons, roles, scenarios, locations, medical parameters.

• 6 Low-Level emergency response specific ontologies Application-specific extension of the generic ontologies consisting of a

definition of firefighters, equipment, tasks, dangers, risks.

9


EMERGENCY PLAN LOW-LEVEL

ONTOLOGY

Defines the specific emergency incident and corresponding scenario including assignment of the required team having specific roles and competences.

10


CASE: INITIALIZATION OF AN

EMERGENCY INCIDENT Emergency Incident at a Location with assigned Team and required Roles. Corresponding Emergency Scenario has required assets such as Equipment, Vehicles, Devices, Roles and Competences to accomplish tasks. Automatic match between the assigned Team with the required assets.

11



EMERGENCY INCIDENT

Link search & rescue incident to a predefined search & rescue scenario with specific competence requirements.

Select qualified team with roles supporting these competences.

12



EMERGENCY INCIDENT

Additional semantic rules state

Emergency incident and scenario • describe emergency situation and available

resources, • track the progress of the emergency situation, • are used to assign tasks requiring specific

competences to people disposing of these competences.

Incident Scenario

Role Competence

hasApplicableScenario

requiredCompetence

hasCompetence

then

Incident

Role

hasAssignedRole

13

CONTEXT MODEL DEPLOYMENT

Server inside fire truck

Fire fighters On-sleeve HMI

Commander Tablet-size HMI

Crisis team Web-based HMI


14


REQUIREMENTS

Multiple emergency response teams

A multitude of sensors & devices

Safety critical situation

High sample rate

Asynchronous processing of events

Distributed reasoning on the context model

Orchestration by the context engine

15

ARCHITECTURE


16

ASYNCHRONOUS EVENT PROCESSING


17


IN CONCLUSION

3

1

2 Definition of an Emergency Response ontology divided into several application-independent ontologies, which are extended with emergency response specific ontologies.

Given the high number of input events the context model & decision engines need to work asynchronously to provide real-time responses to the user.

Field studies are a valuable tool for the identification of the users’ needs that can be used to build a formal context model.

18

THANK YOU!

QUESTIONS?

Astute symposium 2013-10-10_context_modelling_reasoning_wimboffe_annahristoskova

Technology

Transcript of Astute symposium 2013-10-10_context_modelling_reasoning_wimboffe_annahristoskova