Information BaseQoIhasQuality

ContextAppDomainCntx

name: stringdescription: URLstandardName: stringstandardDescription: URL

QoISpatialCntxAttrname: stringlowerBound: QoIMetricupperBound: QoIMetric

QoITemporalCntxAttrname: stringlowerBound: intupperBound: int

QoIBaseAttrname: string

QoIQualifier

QoI attributesQoI attributes A

1 1

The base QoI metadata model

Annual Conference of ITAAnnual Conference of ITAACITA 2009ACITA 2009

Building Principles for a Quality of Information Specification for Sensor-originated Information

The end-to-end information flow

App_i needs information items: α, β, γ, and δ; it receives: α’, γ’, and δ’

App_i needs information items: α, β, γ, and δ; it receives: α’, γ’, and δ’

Networks of data sources

Networks of fusion, processing and intermediary decision centers

CommNets

?

Networks of end-users

α β γ δ

δ‘γ'α'

app_i

• On-demand, just-in-time deployment of sensor-enabled applications requires creation of elaborate sensor-based services with on-the-fly, dynamic binding capabilities to available data/sensor and communication resources

• QoI can play central role in indexing/searching these sensor-based services

QoI for sensor information

• The world comprises objects and linkages between these objects

• The “world state vector” (wsv) describes the world as a function of time

• Elements of the wsv are inferred using data sources

• Information is the estimated value wsv elements

x1={ location, velocity, type, allegiance}

x2={ location, velocity, type, allegiance}

information gathering

information processing

information distribution

information: estimate of object state(s) xQoI: collection of information attributes:

accuracy, latency, provenance

information: estimate of object state(s) xQoI: collection of information attributes:

accuracy, latency, provenance

information delivery

x3={ location, direction, strength, allegiance}

QoI?

sensor nodes

Information + Metadata

Information’ + Metadata’

Information’ + Metadata’

Information + QoI MetadataInformation

AnnotationOperator

Information

Modifier Operator

Information + Metadata

Fusion Operator

Information + Metadata

Information + Metadata

Information + QoI Metadata

Information + QoI Metadata

QoI Annotator

VoI Annotator

• Information flows and processed during its end-to-end journey

derivedmetadata(mtd(ym))

;;

derivedinfo(yn)

metadata(mtd(xm))

info (xm)

Object processorObject

processor

local or remote repositories where metadata maybe retrieved from or stored to for future use

local or remote repositories where metadata maybe retrieved from or stored to for future use

… …

… …

m incomingdata objects

n outgoingdata objects

metadata(mtd(x))

info (x)

An information objectAn information object

metadata(mtd(x1))

info (x1)derived

metadata(mtd(y1))

derivedinfo (y1)

Information operators

• Shooter localization through a collection of acoustic sensor arrays

• The information product produced comprises the “shooter state:– Shooter location– Direction of shooting

LocationAppQoIIntegrityowner: StringsensorModel: StringsensorPower: String

LocationAppQoIAccuracysnr: realrange: realtemporalResolution: real

LocationAppInformationshooterPosition: CoordinateshooterDirection: CoordinatereportTime: double

QoIAccuracyAttr QoIIntegrityAttrInformation

DOAalgorithm(s)

TDOAalgorithm(s)

localizationalgorithm(s)

spectra

location

time

shooter location(bearing)

shooter location(bearing & range)

shooter location(Cartesian co-ordinates)

A shooter localization example

QoI is…

• Mixture of inherent and context-dependent properties of quality complicates progress

• Quality vs. Value of Information

• Quality of Information (QoI)– The body of evidence (described by

information quality attributes) used to make judgments about the fitness (or, utility) of information

• Value of information (VoI) – The utility of the information when

used in a specific application context

VoIVoI

QoIQoI

“holistic” view of QoI“holistic” view of QoI

“separable” view of QoI“separable” view of QoI

assessment inapp-context

inherent qualityattributes

VoIVoIVoIVoIApp_1

App_2 App_N

• Has been studied within the context of data collected and maintained by large industrial and governmental information technology systems– Quality attributes: Consistency, syntactic/semantic accuracy, timeliness/currency/volatility

Chatschik Bisdikian Lance M. Kaplan Mani B. Srivastava David J. Thornley Dinesh Verma Robert I. YoungIBM Research, US ARL, US UCLA, US Imperial College, UK IBM Research, US Dstl, UK