static public bool CalculatePdataAndSij(ref Sij sij, Pj pj, PAkjl pakjl, ref Pdata pdata)
{
bool isFinished = false;
pdata = new Pdata(++pdata.Time, pdata.Value);
sij = new Sij(++sij.Time);
double[,] numerator = new double[Variable.Sentences.Count, (int)Math.Pow(2, Variable.LabelArray.Length)];
for (int i = 0; i < Variable.Sentences.Count; ++i)
{
for (int j = 0; j < Math.Pow(2, Variable.LabelArray.Length); ++j)
{
numerator[i, j] = 1;
}
}
foreach (Sentence sentence in Variable.Sentences)
{
for (int j = 0; j < Math.Pow(2, Variable.LabelArray.Length); ++j) //正确标签
{
foreach (Annotator annotator in Variable.Annotators) //人
{
foreach (Annotation annotation in Variable.Data[annotator][sentence])
{
numerator[sentence.ID, j] *= pakjl.Value[annotator][new Labelset(Variable.LabelArray, j)][new Labelset(Variable.LabelArray, annotation.IntLabel)];
}
}
numerator[sentence.ID, j] *= pj.Value[new Labelset(Variable.LabelArray, j)];
}
}
double[] denominator = new double[Variable.Sentences.Count];
for (int i = 0; i < Variable.Sentences.Count; ++i)
{
for (int q = 0; q < Math.Pow(2, Variable.LabelArray.Length); ++q)
{
denominator[i] += numerator[i, q];
}
}
//计算Pdata和Sij
foreach (Sentence sentence in Variable.Sentences)
{
for (int j = 0; j < Math.Pow(2, Variable.LabelArray.Length); ++j)
{
sij.Value[sentence][new Labelset(Variable.LabelArray, j)] = numerator[sentence.ID, j] / denominator[sentence.ID];
}
pdata.Value += -Math.Log10(denominator[sentence.ID]);
}
if (pdata.MondifiedValue == 0 || pdata.Time == 10)
{
isFinished = true;
}
//Variable.OutputFile.WriteLine(pdata.ToString());
//Variable.OutputFile.WriteLine(sij.ToString(DependentVariable.NumberOfIntlabel));
return(isFinished);
}
static public void CalculatePAkjl(Sij sij, ref PAkjl pakjl)
{
pakjl = new PAkjl(++pakjl.Time);
IDictionary <Annotator, double[, ]> numerator = new Dictionary <Annotator, double[, ]>(); //分子
IDictionary <Annotator, double[]> denominator = new Dictionary <Annotator, double[]>(); //分母
foreach (Annotator annotator in Variable.Annotators) //人
{
numerator.Add(annotator, new double[(int)Math.Pow(2, Variable.LabelArray.Length), (int)Math.Pow(2, Variable.LabelArray.Length)]);
denominator.Add(annotator, new double[(int)Math.Pow(2, Variable.LabelArray.Length)]);
}
//计算分子分母
foreach (Annotator annotator in Variable.Annotators)
{
for (int j = 0; j < Math.Pow(2, Variable.LabelArray.Length); ++j)//正确标签
{
foreach (Sentence sentence in Variable.Sentences)
{
foreach (Annotation annotation in Variable.Data[annotator][sentence])
{
numerator[annotator][j, annotation.IntLabel] += sij.Value[sentence][new Labelset(Variable.LabelArray, j)];
denominator[annotator][j] += sij.Value[sentence][new Labelset(Variable.LabelArray, j)];
}
}
}
}
//计算π
foreach (Annotator annotator in Variable.Annotators) //人
{
for (int j = 0; j < (int)Math.Pow(2, Variable.LabelArray.Length); ++j) //正确标签
{
if (denominator[annotator][j] != 0) //某些结果就是在所有句子中都没出现过
{
for (int l = 0; l < (int)Math.Pow(2, Variable.LabelArray.Length); ++l) //人标的标签
{
pakjl.Value[annotator][new Labelset(Variable.LabelArray, j)][new Labelset(Variable.LabelArray, l)] = numerator[annotator][j, l] / denominator[annotator][j];
}
}
}
}
//Variable.OutputFile.WriteLine(pajl.ToString(DependentVariable.NumberOfIntlabel));
}
static public void RunPeTM(PorSForJointje PorS, Smoothing SmoothingBE, BnOrNot bnOrNot)
{
double[] accuracyOfPersonalityForEachGroup = new double[GroupVariable.AnnotatorGroups.Length];
for (int groupIndex = 0; groupIndex < GroupVariable.AnnotatorGroups.Length; ++groupIndex)
{
IDictionary <Annotator, IDictionary <Character, IDictionary <Will, double> > > okcx = PersonalityFunction.CalculateOkcx(groupIndex);//模拟人对角色个性的标注，计算一次就不变了
Mce mce = InitializeMce(groupIndex);
Sije sije = null;
if (bnOrNot == BnOrNot.Yes)
{
sije = InitializeSijeWithBN(mce, groupIndex);
}
else
{
sije = InitializeSije(mce, groupIndex); //old new
}
Pje pje = null; //p(t|e)
Pdata pdata = null;
IList <double> Pdatas = new List <double>();
for (int convergeTime = 1; convergeTime <= Variable.ConvergeTimeThreshold; ++convergeTime)
{
mce = CalculateMce(sije, groupIndex);
PAkjl pakjl = CoreFunction.CalculatePAkjl(Variable.LabelArray, sije.ToSij, convergeTime, groupIndex);
BEkef bekef = PersonalityFunction.CalculateBExy(mce, okcx, SmoothingBE, convergeTime, groupIndex);
if (PorS == PorSForJointje.P)
{
pje = CalculatePje(sije, convergeTime);
}
if (CalculatePdataAndSije(ref sije, pakjl, bekef, pje, mce, okcx, ref pdata, Pdatas, groupIndex))//old/new
{
break;
}
}
IDictionary <Sentence, IDictionary <Will, double> > sic = ObtainBinaryResult(sije, mce, groupIndex);
WriteBinaryResultFile(sic, mce, groupIndex);
PersonalityFunction.WriteMVResultFile(mce, groupIndex);
accuracyOfPersonalityForEachGroup[groupIndex] = PersonalityPaperFunction.AccuracyOfPersonalityForEachGroup(PersonalityVariable.TruePersonality, mce.EstimatedPersonality);
}
Function.ConsoleWriteLine("Accuracy Of PeTM: " + PersonalityPaperFunction.AccuracyOfPersonality(accuracyOfPersonalityForEachGroup));
}
//废弃
static private bool CalculatePdataAndSijeOld(ref Sije sije, PAkjl pakjl, BEkef bekef, Pje pje, Mce mce, IDictionary <Annotator, IDictionary <Character, IDictionary <Will, double> > > okxc, ref Pdata pdata, IList <double> pdatas, int groupIndex)
{
bool isFinished = false;
//sije的分子
IDictionary <Sentence, IDictionary <Labelset, IDictionary <Will, double> > > numerator = new Dictionary <Sentence, IDictionary <Labelset, IDictionary <Will, double> > >();
//sij的分母（P(data on i)）
IDictionary <Sentence, double> denominator = new Dictionary <Sentence, double>();
//计算分子
foreach (Sentence sentence in sije.Value.Keys)
{
numerator.Add(sentence, new Dictionary <Labelset, IDictionary <Will, double> >());
//开始计算
foreach (Labelset labelsetj in sije.Value[sentence].Keys)//j: true label
{
double valueOfNumeratorForStrong = 1;
double valueOfNumeratorForWeak = 1;
foreach (Annotator annotator in sentence.AnnotaitonGroups[groupIndex].AnnotatorAnnotationDic.Keys)
{
Labelset labelsetl = sentence.AnnotaitonGroups[groupIndex].AnnotatorAnnotationDic[annotator].ToLabelset(Variable.LabelArray);
if (pakjl.Value[annotator].ContainsKey(labelsetj))
{
if (pakjl.Value[annotator][labelsetj].ContainsKey(labelsetl))
{
valueOfNumeratorForStrong *= pakjl.Value[annotator][labelsetj][labelsetl];
valueOfNumeratorForWeak *= pakjl.Value[annotator][labelsetj][labelsetl];
}
else
{
valueOfNumeratorForStrong = 0;
valueOfNumeratorForWeak = 0;
break;
}
}
else
{
break;
}
//β部分的值（Personality新增）
valueOfNumeratorForStrong *= Math.Pow(bekef.Value[annotator][Tuple.Create(Will.strong, Will.strong)], okxc[annotator][sentence.Character][Will.strong])
* Math.Pow(bekef.Value[annotator][Tuple.Create(Will.strong, Will.weak)], okxc[annotator][sentence.Character][Will.weak]);
valueOfNumeratorForWeak *= Math.Pow(bekef.Value[annotator][Tuple.Create(Will.weak, Will.strong)], okxc[annotator][sentence.Character][Will.strong])
* Math.Pow(bekef.Value[annotator][Tuple.Create(Will.weak, Will.weak)], okxc[annotator][sentence.Character][Will.weak]);
}
//if (valueOfNumerator == 0) continue;
//乘以P(t|c)*P(c)
valueOfNumeratorForStrong *= sije.Value[sentence][labelsetj][Will.strong] * mce.Value[sentence.Character][Will.strong] * Math.Pow(labelsetj.HowStrong, 1);
valueOfNumeratorForWeak *= sije.Value[sentence][labelsetj][Will.weak] * mce.Value[sentence.Character][Will.weak] * Math.Pow(labelsetj.HowWeak, 1);
if (valueOfNumeratorForStrong != 0 || valueOfNumeratorForWeak != 0)
{
numerator[sentence].Add(labelsetj, new Dictionary <Will, double>());
numerator[sentence][labelsetj].Add(Will.strong, valueOfNumeratorForStrong);
numerator[sentence][labelsetj].Add(Will.weak, valueOfNumeratorForWeak);
}
}
double valueOfDenominator = 0;
foreach (Labelset Labelsetq in numerator[sentence].Keys)//因为是加，故只需遍历numerator里有的标注，不需遍历所有标注
{
valueOfDenominator += numerator[sentence][Labelsetq][Will.strong];
valueOfDenominator += numerator[sentence][Labelsetq][Will.weak];
}
denominator.Add(sentence, valueOfDenominator);
}
//计算Pdata和Sij
pdata = pdata != null ? new Pdata(++pdata.Time, pdata.Value) : new Pdata(1, 0);
sije = new Sije(++sije.Time);
foreach (Sentence sentence in Variable.Sentences)
{
sije.Value.Add(sentence, new Dictionary <Labelset, IDictionary <Will, double> >());
foreach (Labelset labelset in numerator[sentence].Keys)
{
sije.Value[sentence].Add(labelset, new Dictionary <Will, double>());
sije.Value[sentence][labelset][Will.strong] = numerator[sentence][labelset][Will.strong] / denominator[sentence];//Dic赋值时没有的元素会自动加
sije.Value[sentence][labelset][Will.weak] = numerator[sentence][labelset][Will.weak] / denominator[sentence];
}
pdata.Value += -Math.Log(denominator[sentence]);
}
if (pdatas.Contains(pdata.Value) || (Math.Abs(pdata.MondifiedValue) <= Variable.ConvergeValueThreshold))
{
isFinished = true;
}
else
{
pdatas.Add(pdata.Value);
}
if (Variable.OutputPdata)
{
Variable.OutputFile.WriteLine(pdata.ToString());
}
return(isFinished);
}
static private bool CalculatePdataAndSije(ref Sije sije, PAkjl pakjl, BEkef bekef, Pje pje, Mce mce, IDictionary <Annotator, IDictionary <Character, IDictionary <Will, double> > > okxc, ref Pdata pdata, IList <double> pdatas, int groupIndex)
{
bool isFinished = false;
//sije的分子
IDictionary <Sentence, IDictionary <Labelset, IDictionary <Will, double> > > numerator = new Dictionary <Sentence, IDictionary <Labelset, IDictionary <Will, double> > >();
//sij的分母（P(data on i)）
IDictionary <Sentence, double> denominator = new Dictionary <Sentence, double>();
//计算分子
foreach (Sentence sentence in sije.Value.Keys)
{
numerator.Add(sentence, new Dictionary <Labelset, IDictionary <Will, double> >());
#region 联合概率P(t,m)
IDictionary <Labelset, IDictionary <Will, double> > jointje;
if (pje == null)//PorS == PorSForJointje.S
{
//求后验概率P(t|e)，新增
jointje = new Dictionary <Labelset, IDictionary <Will, double> >();
IDictionary <Will, double> willDenominator = new Dictionary <Will, double>();
willDenominator.Add(Will.strong, 0);
willDenominator.Add(Will.weak, 0);
foreach (Labelset labelset in sije.Value[sentence].Keys)
{
//double valueOfStrong = sije.Value[sentence][labelset][Will.strong] * labelset.HowStrong;//这两个不科学
//double valueOfWeak = sije.Value[sentence][labelset][Will.weak] * labelset.HowWeak;
double valueOfStrong = sije.Value[sentence][labelset][Will.strong];
double valueOfWeak = sije.Value[sentence][labelset][Will.weak];
if (jointje.ContainsKey(labelset))
{
jointje[labelset][Will.strong] += valueOfStrong;
jointje[labelset][Will.weak] += valueOfWeak;
}
else
{
jointje.Add(labelset, new Dictionary <Will, double>());
jointje[labelset].Add(Will.strong, valueOfStrong);
jointje[labelset].Add(Will.weak, valueOfWeak);
}
willDenominator[Will.strong] += valueOfStrong;
willDenominator[Will.weak] += valueOfWeak;
}
//p(t|e)
foreach (Labelset labelset in jointje.Keys.ToArray())
{
if (willDenominator[Will.strong] != 0)
{
jointje[labelset][Will.strong] /= willDenominator[Will.strong];
}
if (willDenominator[Will.weak] != 0)
{
jointje[labelset][Will.weak] /= willDenominator[Will.weak];
}
}
//p(t|e)*p(x)
if (willDenominator[Will.strong] != 0 && willDenominator[Will.weak] != 0)//有一个等于0就不用再算了
{
foreach (Labelset labelset in jointje.Keys.ToArray())
{
jointje[labelset][Will.strong] *= mce.Value[sentence.Character][Will.strong];
jointje[labelset][Will.weak] *= mce.Value[sentence.Character][Will.weak];
}
}
}
else//PorS == PorSForJointje.P
{
jointje = pje.Value;
double valueOfStrong = 0;
double valueOfWeak = 0;
foreach (Labelset labelset in jointje.Keys.ToArray())
{
valueOfStrong += jointje[labelset][Will.strong];
valueOfWeak += jointje[labelset][Will.weak];
}
if (valueOfStrong != 0 && valueOfWeak != 0)
{
foreach (Labelset labelset in jointje.Keys.ToArray())
{
jointje[labelset][Will.strong] *= mce.Value[sentence.Character][Will.strong];
jointje[labelset][Will.weak] *= mce.Value[sentence.Character][Will.weak];
}
}
}
#endregion
foreach (Labelset labelsetj in jointje.Keys)//j: true label
{
#region P({n}|t,e)
double valueOfNumeratorForStrong = 1;
double valueOfNumeratorForWeak = 1;
foreach (Annotator annotator in sentence.AnnotaitonGroups[groupIndex].AnnotatorAnnotationDic.Keys)
{
Labelset labelsetl = sentence.AnnotaitonGroups[groupIndex].AnnotatorAnnotationDic[annotator].ToLabelset(Variable.LabelArray);
if (pakjl.Value[annotator].ContainsKey(labelsetj))
{
if (pakjl.Value[annotator][labelsetj].ContainsKey(labelsetl))
{
valueOfNumeratorForStrong *= pakjl.Value[annotator][labelsetj][labelsetl];
valueOfNumeratorForWeak *= pakjl.Value[annotator][labelsetj][labelsetl];
}
else
{
valueOfNumeratorForStrong = 0;
valueOfNumeratorForWeak = 0;
break;
}
}
else
{
break;
}
//β部分的值（Personality新增）
valueOfNumeratorForStrong *= Math.Pow(bekef.Value[annotator][Tuple.Create(Will.strong, Will.strong)], okxc[annotator][sentence.Character][Will.strong])
* Math.Pow(bekef.Value[annotator][Tuple.Create(Will.strong, Will.weak)], okxc[annotator][sentence.Character][Will.weak]);
valueOfNumeratorForWeak *= Math.Pow(bekef.Value[annotator][Tuple.Create(Will.weak, Will.strong)], okxc[annotator][sentence.Character][Will.strong])
* Math.Pow(bekef.Value[annotator][Tuple.Create(Will.weak, Will.weak)], okxc[annotator][sentence.Character][Will.weak]);
}
#endregion
//乘以(P(t|e)*P(e))
valueOfNumeratorForStrong *= jointje[labelsetj][Will.strong];
valueOfNumeratorForWeak *= jointje[labelsetj][Will.weak];
if (valueOfNumeratorForStrong != 0 || valueOfNumeratorForWeak != 0)
{
numerator[sentence].Add(labelsetj, new Dictionary <Will, double>());
numerator[sentence][labelsetj].Add(Will.strong, valueOfNumeratorForStrong);
numerator[sentence][labelsetj].Add(Will.weak, valueOfNumeratorForWeak);
}
}
double valueOfDenominator = 0;
foreach (Labelset Labelsetq in numerator[sentence].Keys)//因为是加，故只需遍历numerator里有的标注，不需遍历所有标注
{
valueOfDenominator += numerator[sentence][Labelsetq][Will.strong];
valueOfDenominator += numerator[sentence][Labelsetq][Will.weak];
}
denominator.Add(sentence, valueOfDenominator);
}
//计算Pdata和Sij
pdata = pdata != null ? new Pdata(++pdata.Time, pdata.Value) : new Pdata(1, 0);
sije = new Sije(++sije.Time);
foreach (Sentence sentence in Variable.Sentences)
{
sije.Value.Add(sentence, new Dictionary <Labelset, IDictionary <Will, double> >());
foreach (Labelset labelset in numerator[sentence].Keys)
{
sije.Value[sentence].Add(labelset, new Dictionary <Will, double>());
sije.Value[sentence][labelset][Will.strong] = numerator[sentence][labelset][Will.strong] / denominator[sentence];//Dic赋值时没有的元素会自动加
sije.Value[sentence][labelset][Will.weak] = numerator[sentence][labelset][Will.weak] / denominator[sentence];
}
pdata.Value += -Math.Log(denominator[sentence]);
}
if (pdatas.Contains(pdata.Value) || (Math.Abs(pdata.MondifiedValue) <= Variable.ConvergeValueThreshold))
{
isFinished = true;
}
else
{
pdatas.Add(pdata.Value);
}
if (Variable.OutputPdata)
{
Variable.OutputFile.WriteLine(pdata.ToString());
}
return(isFinished);
}
