static void Assumption()
{
//int[] X1 = new int[] { 1, 1, 2, 2, 2, 2, 1, 2 };
//int[] X2 = new int[] { 2, 1, 1, 2, 1, 1, 2, 2 };
int[] X1 = new int[] { 1, 1, 1, 1, 2, 2, 2, 2 };
int[] X2 = new int[] { 1, 1, 1, 1, 2, 2, 2, 2 };
int numberOfX1Is1 = 0;
int numberOfX1Is2 = 0;
int numberOfX2Is1 = 0;
int numberOfX2Is2 = 0;
int numberOfX1Is1X2Is1 = 0;
int numberOfX1Is2X2Is1 = 0;
int numberOfX1Is1X2Is2 = 0;
int numberOfX1Is2X2Is2 = 0;
double N = Convert.ToDouble(X1.Length);
for (int i = 0; i < X1.Length; ++i)
{
if (X1[i] == 1)
{
++numberOfX1Is1;
if (X2[i] == 1)
{
++numberOfX2Is1;
++numberOfX1Is1X2Is1;
}
else
{
++numberOfX2Is2;
++numberOfX1Is1X2Is2;
}
}
else
{
++numberOfX1Is2;
if (X2[i] == 1)
{
++numberOfX2Is1;
++numberOfX1Is2X2Is1;
}
else
{
++numberOfX2Is2;
++numberOfX1Is2X2Is2;
}
}
}
double MI = (numberOfX1Is1X2Is1 / N) * Math.Log((numberOfX1Is1X2Is1 * N) / (numberOfX1Is1 * numberOfX2Is1), 2)
//+ (numberOfX1Is1X2Is2 / N) * Math.Log((numberOfX1Is1X2Is2 * N) / (numberOfX1Is1 * numberOfX2Is2), 2)
//+ (numberOfX1Is2X2Is1 / N) * Math.Log((numberOfX1Is2X2Is1 * N) / (numberOfX1Is2 * numberOfX2Is1), 2)
+ (numberOfX1Is2X2Is2 / N) * Math.Log((numberOfX1Is2X2Is2 * N) / (numberOfX1Is2 * numberOfX2Is2), 2);
double X = SpecialFunction.chisqc(1, 2 * N * MI); //（右尾面积）越小越有联系（不独立），越大越独立
}
public bool IsAWitnessByMI(IList <Label> witness, int group, double threshold)//BN用，PCAlgorithm
{
double MI = 0;
double m = 0;
for (int i = 0; i < Math.Pow(2, witness.Count); ++i)
{
//初始化
int numberOfXTrue = 0;
int numberOfXFalse = 0;
int numberOfYTrue = 0;
int numberOfYFalse = 0;
int numberOfXTrueYTrue = 0;
int numberOfXTrueYFalse = 0;
int numberOfXFalseYTrue = 0;
int numberOfXFalseAYFalse = 0;
int N = 0;
int numberOfZ = 0;
Labelset labelset = new Labelset(witness, i);
//统计数量信息
foreach (Sentence sentence in Variable.Sentences)
{
foreach (Annotation annotation in sentence.AnnotaitonGroups[group].AnnotatorAnnotationDic.Values)
{
++N;
if (annotation.IsAccordingToLabelset(labelset))
{
++numberOfZ;
if (annotation.Labels[First])
{
++numberOfXTrue;
if (annotation.Labels[Second])
{
++numberOfYTrue;
++numberOfXTrueYTrue;
}
else
{
++numberOfYFalse;
++numberOfXTrueYFalse;
}
}
else
{
++numberOfXFalse;
if (annotation.Labels[Second])
{
++numberOfYTrue;
++numberOfXFalseYTrue;
}
else
{
++numberOfYFalse;
++numberOfXFalseAYFalse;
}
}
}
}
}
MI += getConditionalMI(numberOfXTrue, numberOfXFalse, numberOfYTrue, numberOfYFalse,
numberOfXTrueYTrue, numberOfXTrueYFalse, numberOfXFalseYTrue, numberOfXFalseAYFalse, numberOfZ, N);
m = N;
}
return(SpecialFunction.chisqc(Math.Pow(2, witness.Count), 2 * m * MI) >= threshold);//（右尾面积）越小越有联系（不独立），越大越独立
}