Sections
You are here: Home » Data » Documents » Development » RDF files » Validation

Validation

— filed under: , , ,

Validation. The RDF representation is created to be as close as possible to original XML proposal.

Overview Validation  Classification information

Validation URI is defined by dc:identifier

Validation timing is defined by ot:CPUTimeTesting and ot:elapsedTimeTesting. (TODO is it UNIX time?)

The Model being validated is defined by ot:validationModel

The Dataset with predicted values is defined by ot:validationPredictionDataset

The Test Dataset is defined by ot:validationTestDataset

    <ot:Validation rdf:ID="Validation_19">
        <dc:date rdf:datatype="http://www.w3.org/2001/XMLSchema#string">Creation date</dc:date>
        <dc:identifier rdf:datatype="http://www.w3.org/2001/XMLSchema#string"
            >http://myservice/validation/{validationid}</dc:identifier>
        <ot:CPUTimeTesting rdf:datatype="http://www.w3.org/2001/XMLSchema#long">1000</ot:CPUTimeTesting>
        <ot:elapsedTimeTesting rdf:datatype="http://www.w3.org/2001/XMLSchema#long">1000</ot:elapsedTimeTesting>
        <ot:hasValidationInfo>
            <rdf:Description>
                <rdf:type rdf:resource="http://www.opentox.org/api/1.1#CrossValidationInfo"/>
                <ot:crossvalidation rdf:resource="#Crossvalidation"/>
                <ot:fold rdf:datatype="http://www.w3.org/2001/XMLSchema#int">2</ot:fold>
            </rdf:Description>
        </ot:hasValidationInfo>
        <ot:hasValidationInfo rdf:resource="#v1"/>
        <ot:validationModel rdf:resource="#DemoModel"/>
        <ot:validationPredictionDataset rdf:resource="#DatasetPredicted"/>
        <ot:validationTestDataset rdf:resource="#DatasetTest"/>
    </ot:Validation>

Further validation information can be included with the use of ot:hasValidationInfo element, which can have multiple entries. The validation information entries are of four types: Crossvalidation Information, Classification Information, Regression Information and Y Scrambling.

Validation information entries can be defined as separate RDF objects, as well as an anonymous classes (embedded in the Validation representation).

Crossvalidation Information

        <ot:hasValidationInfo>
            <rdf:Description>
                <rdf:type rdf:resource="http://www.opentox.org/api/1.1#CrossValidationInfo"/>
                <ot:crossvalidation rdf:resource="#Crossvalidation"/>
                <ot:fold rdf:datatype="http://www.w3.org/2001/XMLSchema#int">2</ot:fold>
            </rdf:Description>
        </ot:hasValidationInfo>

Classification Information

    <ot:ClassificationInfo rdf:ID="v1">
        <ot:areaUnderROC rdf:datatype="http://www.w3.org/2001/XMLSchema#float">0.75</ot:areaUnderROC>
        <ot:falseNegativeRate rdf:datatype="http://www.w3.org/2001/XMLSchema#float">0.0</ot:falseNegativeRate>
        <ot:falsePositiveRate rdf:datatype="http://www.w3.org/2001/XMLSchema#float">0.0</ot:falsePositiveRate>
        <ot:fMeasure rdf:datatype="http://www.w3.org/2001/XMLSchema#float">1.0</ot:fMeasure>
        <ot:kappaStatistic rdf:datatype="http://www.w3.org/2001/XMLSchema#float">1.0</ot:kappaStatistic>
        <ot:numberCorrect rdf:datatype="http://www.w3.org/2001/XMLSchema#int">1</ot:numberCorrect>
        <ot:numberIncorrect rdf:datatype="http://www.w3.org/2001/XMLSchema#int">0</ot:numberIncorrect>
        <ot:numberUnclassified rdf:datatype="http://www.w3.org/2001/XMLSchema#int">0</ot:numberUnclassified>
        <ot:numFalseNegatives rdf:datatype="http://www.w3.org/2001/XMLSchema#int">0</ot:numFalseNegatives>
        <ot:numFalsePositives rdf:datatype="http://www.w3.org/2001/XMLSchema#int">0</ot:numFalsePositives>
        <ot:numTrue_negatives rdf:datatype="http://www.w3.org/2001/XMLSchema#float">0.0</ot:numTrue_negatives>
        <ot:numTruePositives rdf:datatype="http://www.w3.org/2001/XMLSchema#int">1</ot:numTruePositives>
        <ot:percentCorrect rdf:datatype="http://www.w3.org/2001/XMLSchema#float">100.0</ot:percentCorrect>
        <ot:percentIncorrect rdf:datatype="http://www.w3.org/2001/XMLSchema#float">0.0</ot:percentIncorrect>
        <ot:percentUnclassified rdf:datatype="http://www.w3.org/2001/XMLSchema#float">0.0</ot:percentUnclassified>
        <ot:precision rdf:datatype="http://www.w3.org/2001/XMLSchema#float">1.0</ot:precision>
        <ot:recall rdf:datatype="http://www.w3.org/2001/XMLSchema#float">1.0</ot:recall>
        <ot:trueNegativeRate rdf:datatype="http://www.w3.org/2001/XMLSchema#float">0.0</ot:trueNegativeRate>
        <ot:truePositiveRate rdf:datatype="http://www.w3.org/2001/XMLSchema#float">1.0</ot:truePositiveRate>
    </ot:ClassificationInfo>

Regression Information

    <ot:RegressionInfo rdf:ID="RegressionInfo_1">
        <ot:meanAbsolutError rdf:datatype="http://www.w3.org/2001/XMLSchema#float">0.1</ot:meanAbsolutError>
        <ot:relativeAbsoluteError rdf:datatype="http://www.w3.org/2001/XMLSchema#float">0.2</ot:relativeAbsoluteError>
        <ot:rootMeanSquaredError rdf:datatype="http://www.w3.org/2001/XMLSchema#float">0.2</ot:rootMeanSquaredError>
        <ot:rootRelativeSquaredError rdf:datatype="http://www.w3.org/2001/XMLSchema#float">0.4</ot:rootRelativeSquaredError>
    </ot:RegressionInfo>

Y Scrambling

The unscrambled Dataset is defined by ot:unscrambledDatasetURI

    <ot:YScrambling rdf:ID="YScrambling_2">
        <ot:unscrambledDatasetURI rdf:resource="#DSSToxCPDBAS"/>
        <ot:yScramblingEnabled rdf:datatype="http://www.w3.org/2001/XMLSchema#boolean">true</ot:yScramblingEnabled>
        <ot:yScramblingSeed rdf:datatype="http://www.w3.org/2001/XMLSchema#int">1</ot:yScramblingSeed>
    </ot:YScrambling>

 

 

Document Actions

crossvalidation resource

Posted by Martin Gütlein at Dec 02, 2009 10:09 AM
Thanks for the work. AFAIC the validation rdf representation looks good.
I just added the a crossvalidation resource to the crossvalidation-info object. The crossvalidation resource is still to be definded. I will do that, as well as moving the validation rdf object to the validation API, within the next days...

Martin