Embed Size (px)
Citation preview
Using Ontologies to Reduce User Intervention to Deploy Sensing Campaigns with the InCense Toolkit
Presented by:
Marcela D. Rodríguez
CICESE/UABC, Ensenada, México
1st International Workshop on Ubiquitous Mobile Instrumentation
Challenges to deploy a sensing campaign Deciding the granularity of the sensed
informationComponents that collect low-level data vs
high-level data Calibrating the sensing components to
the population to be monitored.To the particular participants characteristics Indicating a calibration criteria
The target users are researchers with little or no technical background
Developing a tool for behavioral data collection from mobile phones to enable researchers with low technical
skills to implement a sensing application: InCense
InCense implementation model
Session: group of components connected to achieve a sensing goal.
Sensors: act as interfaces with the mobile phone’s sensors
Filters: preprocess raw data from sensors
Survey: multiple choice or open- ended questions
Triggers: start sessions if certain conditions are met
Sink: data pool wherein the sensed information is assembled into files
OntoInCense<ontology>
User
Customize
Deploy
Analyse
Implement
Ontology-based GUIOntology to support customization
Code generation
Sensors Library Filters Library
Specification language and re-usable components
InCense API
Template Engine
Class Builder
<plug-in>
Contextual Database
Mobileapplication
Project Server
JSON< / >
Filter Explorer
Filter Generator
<template>
Configuration file Generator
<template>
InCense Architecture
InCense Manager
Use of the InCense API for implementing a sensing application
OntoInCense<ontology>
User
Customize
Deploy
Analyse
Implement
Ontology-based GUIOntology to support customization
Code generation
Sensors Library Filters Library
Specification language and re-usable components
InCense API
Template Engine
Class Builder
<plug-in>
Contextual Database
Mobileapplication
Project Server
JSON< / >
Filter Explorer
Filter Generator
<template>
Configuration file Generator
<template>
InCense Architecture
InCense Manager
OntoInCense
OntoInCense<ontology>
User
Customize
Deploy
Analyse
Implement
Ontology-based GUIOntology to support customization
Code generation
Sensors Library Filters Library
Specification language and re-usable components
InCense API
Template Engine
Class Builder
<plug-in>
Contextual Database
Mobileapplication
Project Server
JSON< / >
Filter Explorer
Filter Generator
<template>
Configuration file Generator
<template>
InCense Architecture
InCense Manager
OntoInCense
OntoInCense<ontology>
User
Customize
Deploy
Analyse
Implement
Graphical WidgetOntology to support customization
Code generation
Sensors Library Filters Library
Specification language and re-usable components
InCense API
Template Engine
Class Builder
<plug-in>
Contextual Database
Mobileapplication
Project Server
JSON< / >
Filter Explorer
Filter Generator
<template>
Configuration file Generator
<template>
InCense Architecture
InCense Manager
Scenario: “A public health organization (PHO) is interested in
comparing the walking habits of older adults in the winter and in the spring. They began using InCense for data gathering from 392 individuals during two weeks in the middle of January, and then again in May. The application captures the individual location, the activity level obtained from the accelerometers. A filter infers from the GPS and accelerometer, if the individual is walking or in a vehicle as he leaves his home. When InCense detects that the user is back at home, the mobile phones, will ask the individuals to complete a survey with question related to the activity being performed and their wellness. The data captured from the individuals is sent to the PHO to find interesting correlations with standard statistical packages.”
Extending the Filter Library
Implement a Filter Register a FilterRegistrar variables
to callibrate
Add the Filter to OntoIncense
Graphical Widget
Filter Explorer
Extending the Filter Library
Implement a Filter Register a FilterRegistrar variables
to callibrate
Add the Filter to OntoIncense
Graphical Widget
a
b
Extending the Filter Library
Implement a Filter Register a FilterRegistrar variables
to callibrate
Add the Filter to OntoIncense
Graphical Widget
a
b
Extending the Filter Library
Implement a Filter Register a FilterRegistrar variables
to callibrate
Add the Filter to OntoIncense
Graphical Widget
Develop a sensing campaign
Select/drag Components
Add Relationships
Calibrate components
Participant height
Conclusions and Future work The ontology acts:
As a representational model: Facilitates to understand the implementation model of InCense
As a graphical user: Adds flexibilty to InCense Toolkit for customizing a sensing application.
We plan to evaluate InCense
