private FastVector createWhereFastVector()
{
FastVector fvWhere = new FastVector(5 + whereWordsBefore * 2 + whereWordsAfter * 2);
fvWhere.addElement(new weka.core.Attribute("word", (FastVector)null));
fvWhere.addElement(new weka.core.Attribute("wordCount"));
fvWhere.addElement(new weka.core.Attribute("sentence"));
//fvWhere.addElement(new weka.core.Attribute("position"));
//fvWhere.addElement(new weka.core.Attribute("sentenceStartProximity"));
fvWhere.addElement(new weka.core.Attribute("wordScore"));
for (int i = whereWordsBefore; i > 0; i--)
{
fvWhere.addElement(new weka.core.Attribute("word-" + i, (FastVector)null));
}
for (int i = 1; i <= whereWordsAfter; i++)
{
fvWhere.addElement(new weka.core.Attribute("word+" + i, (FastVector)null));
}
for (int i = whereWordsBefore; i > 0; i--)
{
fvWhere.addElement(new weka.core.Attribute("postag-" + i, fvPOS));
}
for (int i = 1; i <= whereWordsAfter; i++)
{
fvWhere.addElement(new weka.core.Attribute("postag+" + i, fvPOS));
}
FastVector fvClass = new FastVector(2);
fvClass.addElement("yes");
fvClass.addElement("no");
fvWhere.addElement(new weka.core.Attribute("where", fvClass));
return(fvWhere);
}
private FastVector createWhyFastVector()
{
FastVector fvWhy = new FastVector(8 + whyWordsBefore * 2 + whyWordsAfter * 2);
fvWhy.addElement(new weka.core.Attribute("candidate", (FastVector)null));
fvWhy.addElement(new weka.core.Attribute("wordCount"));
fvWhy.addElement(new weka.core.Attribute("sentence"));
fvWhy.addElement(new weka.core.Attribute("candidateScore"));
fvWhy.addElement(new weka.core.Attribute("numWho"));
fvWhy.addElement(new weka.core.Attribute("numWhen"));
fvWhy.addElement(new weka.core.Attribute("numWhere"));
for (int i = whereWordsBefore; i > 0; i--)
{
fvWhy.addElement(new weka.core.Attribute("word-" + i, (FastVector)null));
}
for (int i = 1; i <= whereWordsAfter; i++)
{
fvWhy.addElement(new weka.core.Attribute("word+" + i, (FastVector)null));
}
for (int i = whyWordsBefore; i > 0; i--)
{
fvWhy.addElement(new weka.core.Attribute("postag-" + i, fvPOS));
}
for (int i = 1; i <= whyWordsAfter; i++)
{
fvWhy.addElement(new weka.core.Attribute("postag+" + i, fvPOS));
}
FastVector fvClass = new FastVector(2);
fvClass.addElement("yes");
fvClass.addElement("no");
fvWhy.addElement(new weka.core.Attribute("why", fvClass));
return(fvWhy);
}
private static Instances CreateInstanceOnFly(double[] a, double[] b)
{
FastVector atts;
Instances data;
double[] vals;
// 1. set up attributes
atts = new FastVector();
// - numeric
atts.addElement(new Attribute("att1"));
atts.addElement(new Attribute("att2"));
// 2. create Instances object
data = new Instances("MyRelation", atts, 0);
for (int i = 0; i < a.Length; ++i)
{
// 3. fill with data
// first instance
vals = new double[data.numAttributes()];
// - numeric
vals[0] = a[i];
// - nominal
vals[1] = b[i];
data.add(new weka.core.DenseInstance(1.0, vals));
}
return(data);
}
public FastVector GetDataSetAtts()
{
if (_fvWekaAttributes != null)
{
return(_fvWekaAttributes);
}
// Declare features
FastVector fvWekaAttributes = new FastVector(DomPool.SelectorFeatures.Count() + 1);
foreach (Feature currFeature in DomPool.SelectorFeatures)
{
weka.core.Attribute feature = new weka.core.Attribute(currFeature.ToString());
fvWekaAttributes.addElement(feature);
}
// Declare the class attribute along with its values
FastVector fvClassVal = new FastVector(2);
fvClassVal.addElement("yes");
fvClassVal.addElement("no");
weka.core.Attribute ClassAttribute = new weka.core.Attribute("theClass", fvClassVal);
// Declare the feature vector
fvWekaAttributes.addElement(ClassAttribute);
_fvWekaAttributes = fvWekaAttributes;
return(_fvWekaAttributes);
}
public Identifier(Boolean isAnnotated, WhyTrainer wt)
{
this.isAnnotated = isAnnotated;
this.wt = wt;
listWhoCandidates = new List <Candidate>();
listWhenCandidates = new List <Candidate>();
listWhereCandidates = new List <Candidate>();
listWhatCandidates = new List <List <Token> >();
listWhyCandidates = new List <List <Token> >();
listSecondaryWhyCandidates = new List <Candidate>();
fvPOS = new FastVector(Token.PartOfSpeechTags.Length);
foreach (String POS in Token.PartOfSpeechTags)
{
fvPOS.addElement(POS);
}
whoClassifier = (Classifier)SerializationHelper.read(@"..\..\IdentifierModels\who.model");
whenClassifier = (Classifier)SerializationHelper.read(@"..\..\IdentifierModels\when.model");
whereClassifier = (Classifier)SerializationHelper.read(@"..\..\IdentifierModels\where.model");
whyClassifier = (Classifier)SerializationHelper.read(@"..\..\IdentifierModels\why.model");
initializeAnnotations();
}
private static Instances DefineBucketAttributes(int numBuckets)
{
FastVector attributes = new FastVector();
FastVector numberOfTimesPregnant = new FastVector();
FastVector plasmaGlucoseConcentration = new FastVector();
FastVector diastolicBloodPressure = new FastVector();
FastVector tricepsSkinFoldThickness = new FastVector();
FastVector twoHourSerumInsulin = new FastVector();
FastVector bmi = new FastVector();
FastVector diabetesPedigreeFunction = new FastVector();
FastVector age = new FastVector();
for (int i = 0; i < numBuckets; i++)
{
string attributeValueName = i.ToString();
numberOfTimesPregnant.addElement(attributeValueName);
plasmaGlucoseConcentration.addElement(attributeValueName);
diastolicBloodPressure.addElement(attributeValueName);
tricepsSkinFoldThickness.addElement(attributeValueName);
twoHourSerumInsulin.addElement(attributeValueName);
bmi.addElement(attributeValueName);
diabetesPedigreeFunction.addElement(attributeValueName);
age.addElement(attributeValueName);
}
attributes.addElement(new weka.core.Attribute("numberOfTimesPregnant", numberOfTimesPregnant));
attributes.addElement(new weka.core.Attribute("plasmaGlucoseConcentration", plasmaGlucoseConcentration));
attributes.addElement(new weka.core.Attribute("diastolicBloodPressure", diastolicBloodPressure));
attributes.addElement(new weka.core.Attribute("tricepsSkinFoldThickness", tricepsSkinFoldThickness));
attributes.addElement(new weka.core.Attribute("twoHourSerumInsulin", twoHourSerumInsulin));
attributes.addElement(new weka.core.Attribute("bmi", bmi));
attributes.addElement(new weka.core.Attribute("diabetesPedigreeFunction", diabetesPedigreeFunction));
attributes.addElement(new weka.core.Attribute("age", age));
FastVector diabetes = new FastVector();
diabetes.addElement("0"); // negative
diabetes.addElement("1"); // positive
attributes.addElement(new weka.core.Attribute("diagnosis", diabetes));
Instances instances = new Instances("diabetes", attributes, 0);
return(instances);
}
private static FastVector PrepareFastVector()
{
var fastVector = new FastVector(_attributes.Count);
foreach (var attribute in _attributes)
{
fastVector.addElement(attribute);
}
return(fastVector);
}
/// <summary>
/// Creates and returns empty weka data set </summary>
/// <param name="numOfAttr"> int Number of attributes without class attribute </param>
/// <param name="capacity"> int Capacity of sample </param>
/// <returns> empty weka data set </returns>
private static Instances createEmptyWekaDataSet(int numOfAttr, int capacity, IDictionary <double[], string> classValues)
{
//Vector for class attribute possible values
FastVector fvClassVal = new FastVector();
//Map double value for every possible class value
Hashtable classVals = new Dictionary <string, double?>();
//Map class label with double key value
Hashtable classValsDoubleAsKey = new Dictionary <double?, string>();
//ind represents double value for class attribute
int ind = 0;
//loop through possible class values
foreach (KeyValuePair <double[], string> values in classValues)
{
//add value to vector
fvClassVal.addElement(values.Value);
//map double value for class value
classVals[values.Value] = new double?(ind);
//map class label for double key value
classValsDoubleAsKey[new double?(ind)] = values.Value;
ind++;
}
//Class attribute with possible values
Attribute classAttribute = new Attribute("theClass", fvClassVal, classValues.Count);
//Creating attribute vector for Instances class instance
FastVector fvWekaAttributes = new FastVector(numOfAttr + 1);
//Fill vector with simple attributes
for (int i = 0; i < numOfAttr; i++)
{
fvWekaAttributes.addElement(new Attribute(i + "", i));
}
//Add class attribute to vector
fvWekaAttributes.addElement(classAttribute);
//newDataSet as Instances class instance
Instances newDataSet = new Instances("newDataSet", fvWekaAttributes, capacity);
return(newDataSet);
}
///
/// <summary> * Sets the format of output instances. </summary>
///
private void setOutputFormat()
{
// give names to the new attributes
FastVector newAtts = new FastVector();
string foName = null;
for (int i = 0; i < getInputFormat().numAttributes(); i++)
{
string attName = getInputFormat().attribute(i).name();
foName = "'FT " + attName.Replace('\'', ' ').Trim() + " (re)\'";
Attribute newAttribX = new Attribute(foName);
newAtts.addElement(newAttribX);
foName = "'FT " + attName.Replace('\'', ' ').Trim() + " (im)\'";
Attribute newAttribY = new Attribute(foName);
newAtts.addElement(newAttribY);
}
setOutputFormat(new Instances(getInputFormat().relationName(), newAtts, getNumCoeffs()));
}
//code taken from here http://stackoverflow.com/questions/9616872/classification-of-instances-in-weka/14876081#14876081
// This creates the data set's attributes vector
public static FastVector CreateFastVector(int size)
{
var fv = new FastVector();
weka.core.Attribute att;
foreach (int key in TrainingTesting_SharedVariables._trainTopIGFeatures)
{
if (key != TrainingTesting_SharedVariables._trainTopIGFeatures[TrainingTesting_SharedVariables._trainTopIGFeatures.Length - 1])
{
att = new weka.core.Attribute("att_" + (key + 1).ToString());
fv.addElement(att);
}
}
{
var classValues = new FastVector(1); //it doesnt matter if its 3 or 1, when addElement is used the fastvector grows.
List <string> labels = GuiPreferences.Instance.getLabels();
GuiPreferences.Instance.setLog("automatically! adding " + (labels.Count - 1).ToString() + " classes 2 -> " + (labels.Count + 1).ToString() + " to fast vector, based on protocol labels");
//baseline is ignored, we start from the second event in the protocol
for (int l = 1; l < labels.Count; l++)
{
classValues.addElement((l + 1).ToString());
}
//classValues.addElement("2");
//classValues.addElement("3");
//classValues.addElement("4");
//classValues.addElement("5");
var classAttribute = new weka.core.Attribute("class", classValues);
fv.addElement(classAttribute);
}
return(fv);
}
// Define all the attributes for the diabetes dataset
private static Instances DefineCategoricalAttributes()
{
FastVector attributes = new FastVector();
FastVector numberOfTimesPregnant = new FastVector();
numberOfTimesPregnant.addElement("zero"); // 0
numberOfTimesPregnant.addElement("low"); // 1-4
numberOfTimesPregnant.addElement("medium"); // 5-9
numberOfTimesPregnant.addElement("high"); // 10+
attributes.addElement(new weka.core.Attribute("numberOfTimesPregnant", numberOfTimesPregnant));
FastVector plasmaGlucoseConcentration = new FastVector();
plasmaGlucoseConcentration.addElement("normal"); // < 140
plasmaGlucoseConcentration.addElement("high"); // >= 140
attributes.addElement(new weka.core.Attribute("plasmaGlucoseConcentration", plasmaGlucoseConcentration));
FastVector diastolicBloodPressure = new FastVector();
diastolicBloodPressure.addElement("low"); // < 60
diastolicBloodPressure.addElement("normal"); // 60 to 79
diastolicBloodPressure.addElement("pre-high"); // 80 to 89
diastolicBloodPressure.addElement("high"); // 90+
attributes.addElement(new weka.core.Attribute("diastolicBloodPressure", diastolicBloodPressure));
FastVector tricepsSkinFoldThickness = new FastVector();
tricepsSkinFoldThickness.addElement("low"); // < 4.5
tricepsSkinFoldThickness.addElement("normal"); // 4.5 to 36.5
tricepsSkinFoldThickness.addElement("high"); // > 36.5
attributes.addElement(new weka.core.Attribute("tricepsSkinFoldThickness", tricepsSkinFoldThickness));
FastVector twoHourSerumInsulin = new FastVector();
twoHourSerumInsulin.addElement("normal"); // < 166
twoHourSerumInsulin.addElement("high"); // >= 166
attributes.addElement(new weka.core.Attribute("twoHourSerumInsulin", twoHourSerumInsulin));
FastVector bmi = new FastVector();
bmi.addElement("underweight"); // < 18.5
bmi.addElement("normal"); // 18.5 to 25
bmi.addElement("overweight"); // 25 to 30
bmi.addElement("obese"); // 30+
attributes.addElement(new weka.core.Attribute("bmi", bmi));
FastVector diabetesPedigreeFunction = new FastVector();
diabetesPedigreeFunction.addElement("low"); // < 0.2
diabetesPedigreeFunction.addElement("normal"); // 0.2 to 0.7
diabetesPedigreeFunction.addElement("high"); // > 0.7
attributes.addElement(new weka.core.Attribute("diabetesPedigreeFunction", diabetesPedigreeFunction));
FastVector age = new FastVector();
age.addElement("young"); // under 30
age.addElement("middle"); // 30 to 50 inclusive
age.addElement("old"); // over 50
attributes.addElement(new weka.core.Attribute("age", age));
FastVector diabetes = new FastVector();
diabetes.addElement("0"); // negative
diabetes.addElement("1"); // positive
attributes.addElement(new weka.core.Attribute("diagnosis", diabetes));
Instances instances = new Instances("diabetes", attributes, 0);
return(instances);
}
