private TextClassificationResult PredictClass(string file, string exp)
{
var row = C45TextClassify.GetDataRowFeatureFile(file, null, SelectedFeatures);
var inputdata = GetInputData(row);
var output = c45Tree.Decide(inputdata);
var found = c45Tree.ClassArray[output];
return(new TextClassificationResult()
{
DocumentName = Path.GetFileName(file),
DocumentPath = file,
ExpectedClass = exp,
FoundClass = found
});
}
private List <TextClassificationResult> testClassification(string featureDirectoryPath, string resultFile)
{
DirectoryInfo directoryInfo = new DirectoryInfo(featureDirectoryPath + "/TESTINGDATA");
string resultFilePath = string.Format("{0}/{1}", featureDirectoryPath, resultFile);
//List<string> lstFeatures = C45TextClassify.GetListOfFeatures(featureFilePath);
var data = C45TextClassify.GetFeatureInTable(SelectedFeatures, directoryInfo.FullName, true);
var lstClasses = c45Tree.ClassArray;
if (File.Exists(resultFilePath))
{
File.Delete(resultFilePath);
}
var ff = File.Create(resultFilePath);
ff.Close();
List <TextClassificationResult> results = new List <TextClassificationResult>();
foreach (DataRow row in data.Rows)
{
var inputdata = GetInputData(row);
var output = c45Tree.Decide(inputdata);
var found = "UNKNOWN";
if (output > -1 && output < lstClasses.Count())
{
found = lstClasses[output];
}
var expected = row["ResultClass"].ToString();
var docPath = row["DocumentPath"].ToString();
File.AppendAllLines(resultFilePath, new string[] { "Doc:" + Path.GetFileName(docPath) + "Found:" + found + "| Expected: " + expected });
var result = new TextClassificationResult()
{
DocumentName = Path.GetFileName(docPath),
DocumentPath = docPath,
ExpectedClass = expected,
FoundClass = found
};
results.Add(result);
}
return(results);
}
public async Task <DecisionTree> Execute()
{
C45TextClassify c45 = new C45TextClassify();
return(await Task.Run(() => c45.CreateTextClassificationTrainingData(TrainingDataSetFolderLocation, lstSelectedFeature)));
}
