static void Main(string[] args)
{
/* CAttributeClassification client = new CAttributeClassification();
client.ReadRealname(@"D:\cuiwanyun\qa\data\SatoriCategoryToRealname.txt");
client.ReadRDF(@"\\graph012\e$\v-wacui\NewSatori\qaFullShort_sampleRDF.txt");
client.OutputAttributeFrequency(@"D:\cuiwanyun\qa\data\AttributeCategoryFrequency_Satorinew.txt");
client.FindMostcategory(@"D:\cuiwanyun\qa\data\AttributeCategoryFrequency_Satorinew.txt", @"D:\cuiwanyun\qa\data\AttributeMaxcategoryFrequency_Satorinew.txt");
client.FindOccurredMaxcategory(@"D:\cuiwanyun\qa\data\AttributeMaxcategoryFrequency_Satorinew.txt", @"D:\cuiwanyun\qa\data\AttributeOccurredmaxcategoryFrequency_Satorinew.txt");*/
/* CPathClassification client = new CPathClassification();
Console.WriteLine(client.GetPathCategory("-population-year"));
Console.WriteLine(client.GetPathCategory("-population-number"));
Console.WriteLine(client.GetPathCategory("-political appointer"));*/
CClassifyQuestion client = new CClassifyQuestion();
Console.WriteLine(client.GetScategory("What is the area of south africa?"));
}
static void Main(string[] args)
{
CClassifyQuestion client = new CClassifyQuestion();
StreamReader reader = new StreamReader(@"E:\v-wacui\QuestionAnswering\ParsingData\LabeledQuestion.txt");
string currentLine;
Dictionary<string,string> categoryToScategory=new Dictionary<string,string>();
Dictionary<string,int> trueFrequency=new Dictionary<string,int>();
Dictionary<string,int> totalFrequency=new Dictionary<string,int>();
categoryToScategory.Add("ENTY", "Text");
categoryToScategory.Add("LOC", "Text");
categoryToScategory.Add("NUM", "Decimal");
categoryToScategory.Add("NUM:date", "Datetime");
categoryToScategory.Add("HUM", "Text");
while ((currentLine = reader.ReadLine()) != null)
{
int index;
index = currentLine.IndexOf(':');
string category = currentLine.Substring(0, index);
index = currentLine.IndexOf(' ');
if (category == "NUM")
{
string category2 = currentLine.Substring(0, index);
if (category2 == "NUM:date") category = category2;
}
if (categoryToScategory.ContainsKey(category) == false) continue;
string scategory = categoryToScategory[category];
if (trueFrequency.ContainsKey(scategory) == false)
{
trueFrequency.Add(scategory, 0);
totalFrequency.Add(scategory, 0);
}
string question = currentLine.Substring(index + 1);
if (client.GetScategory(question) == scategory) trueFrequency[scategory]++;
totalFrequency[scategory]++;
}
reader.Close();
foreach (string scategory in totalFrequency.Keys)
Console.WriteLine(scategory + "\t" + Convert.ToDouble(trueFrequency[scategory]) / totalFrequency[scategory]);
}
public void ReadQA(string filePath)
{
StreamReader reader = new StreamReader(filePath);
string currentLine;
int lineCount = 0;
string resourcefilePath = @"D:\cuiwanyun\qa\data\ResourceData\";
CStringTool.ReadPluralToSingual(resourcefilePath+@"StemmingVerb.txt");
CStringTool.ReadPluralToSingual2(resourcefilePath + @"StemmingNoun.txt");
//NERKBclient.ReadCategoryId(resourcefilePath + @"CategoryId_Satorinew.txt");
//NERKBclient.ReadNER(resourcefilePath + @"qaFullShort_PatternFrequencyNER.txt");
CClassifyQuestion client=new CClassifyQuestion();
while ((currentLine = reader.ReadLine()) != null)
{
lineCount++;
if (lineCount % 100000 == 0)
//break;
Console.WriteLine(lineCount);
if (lineCount == 1)
currentLine = "What year was america established? In 1776, when America signed the Declaration of Independence stating their independence from England.";
if (visitedLine.Contains(currentLine))
continue;
visitedLine.Add(currentLine);
string[] tokens = currentLine.Split('\t');
string question = CStringTool.FormalizeString(tokens[0]);
string answer = CStringTool.FormalizeStringNumber(tokens[1]);
string questionCategory = client.GetScategory(tokens[0]);
if (questionCategory == "DESC"||questionCategory=="ABBR") continue;
if (question.Contains(" your ") || question.Contains(" you "))
continue;
Understand(question, answer,questionCategory);
}
reader.Close();
}
