static void wordsSvm()
{//http://www.svm-tutorial.com/2014/10/svm-tutorial-classify-text-csharp/
var index = 11;
var rooms11 = File.ReadAllText(string.Format("../../room{0}.txt", index));
var rooms22 = File.ReadAllText(string.Format("../../room{0}.txt", ++index));
var rooms1 = rooms11.Split(';');
var rooms2 = rooms22.Split(';');
var termFreqWeight = new PanGuTermFreqWeight();
termFreqWeight.computerFW(rooms1, rooms2);
var problem = new SVMProblem(); //SVM.SetPrintStringFunction(mySVMPrintFunction);
var lableFeaturesBuilder = new LableFeaturesBuilder();
var segment = new PanGuSegment();
foreach (var room in rooms2)
{
var words = termFreqWeight.TermsByDoc[1, room];
words = words.Where(t => t.Length > 1 && termFreqWeight.TermFWByGlobal.ContainsKey(t) && termFreqWeight.TermFWByType.ContainsKey(0, t)).ToArray();
//lableFeaturesBuilder.AddToProblem(problem, room, words.Select(t => new KeyValuePair<string, double>(t,termFreqWeight.TermFWByDoc[1, room, t].Freq*GetW(t))));
lableFeaturesBuilder.AddToProblem(problem, room, words.Select(t => new KeyValuePair <string, double>(t, Math.Max(termFreqWeight.TermFWByType[0, t].Freq, termFreqWeight.TermFWByType[1, t].Freq))));
}
SVMParameter parameter = new SVMParameter();
parameter.Type = SVMType.C_SVC;
parameter.Kernel = SVMKernelType.LINEAR;
parameter.C = 1;
parameter.Probability = true;
//parameter.
//parameter.WeightLabels = lableFeaturesBuilder.CreateWeightFeatures("大", "双", "套", "大床", "双床").ToArray();
//parameter.Weights = new double[] { 1.90, 1.90, 1.90, 1.99, 1.99 };
//parameter.WeightLabels = lableFeaturesBuilder.CreateWeightLables<int>("行政湖景双床房").ToArray();
//parameter.Weights = new double[] { 1.90, 1.90, 1.90, 1.99, 1.99 };
problem = problem.Normalize(SVMNormType.L1);
problem.CheckParameter(parameter);
var model2 = LibSVMsharp.SVM.Train(problem, parameter);
model2.SaveModel("roomMatching.model");
var model = LibSVMsharp.SVM.LoadModel("roomMatching.model");
foreach (var room in rooms1)
{
var words = termFreqWeight.TermsByDoc[0, room];
words = words.Where(t => t.Length > 1 && termFreqWeight.TermFWByGlobal.ContainsKey(t) && termFreqWeight.TermFWByType.ContainsKey(1, t)).ToArray();
//var nodes = lableFeaturesBuilder.CreateNodes(words.Select(t => new KeyValuePair<string, double>(t, termFreqWeight.TermFWByDoc[0, room, t].Freq * GetW(t))));
var nodes = lableFeaturesBuilder.CreateNodes(words.Select(t => new KeyValuePair <string, double>(t, Math.Max(termFreqWeight.TermFWByType[0, t].Freq, termFreqWeight.TermFWByType[1, t].Freq))));
if (nodes.Length > 0)
{
nodes = nodes.Normalize(SVMNormType.L1);
double predictedY = 0;
predictedY = LibSVMsharp.SVM.Predict(model, nodes);
double[] values = null; double probabilityValue = 0;
probabilityValue = LibSVMsharp.SVM.PredictValues(model, nodes, out values);
double[] est = null; double probability = 0;
probability = LibSVMsharp.SVM.PredictProbability(model, nodes, out est);
Console.WriteLine("{0,22}\t{1},{2},{3},{4},{5}", room, lableFeaturesBuilder.GetLable(predictedY), predictedY, probabilityValue, probability, string.Empty);
}
}
Console.WriteLine(new string('=', 80));
}