/// <summary>
/// The entry point of the program, where the program control starts and ends.
/// </summary>
/// <param name="args">The command-line arguments.</param>
public static void Main(string[] args)
{
CheckArguments(args); // check arguments
/* file names of training file and test file. */
string trainingFile = args[0];
string testFile = args[1];
string resultFile = args[0] + "_prediction.txt";
/* instances of DataSet class representing training and test set. */
var trainingData = new RatingData(trainingFile);
var testData = new RatingData(testFile);
Console.WriteLine("rating matrices for both files are created.");
var recommender = new UserKNN();
recommender.Ratings = trainingData;
recommender.K = 60;
recommender.Correlation = CorrelationType.Pearson;
recommender.Aggregation = AggregationType.WeightedScaledSum;
recommender.Training(trainingData);
recommender.WriteResult(resultFile, testData);
PrintRMSE(trainingFile);
}
public void WriteResult(string resultFilePath, RatingData testData)
{
StreamWriter writer = new StreamWriter(resultFilePath);
for (int index = 0; index < testData.Count; ++ index)
{
int user = testData.Users[index];
int item = testData.Items[index];
float predictedScore = Predict(ratings, user, item);
StringBuilder sb = new StringBuilder();
sb.Append(user.ToString()).Append('\t').
Append(item.ToString()).Append('\t').
Append(predictedScore.ToString());
writer.WriteLine(sb.ToString());
}
writer.Close();
}
public float Predict(RatingData ratings, int user, int item)
{
int count = 0; // # of similar items
float prediction = 0; // result of prediction
float normFactor = 0; // normalizing factor
float sum = 0; // linear sum, weighted sum, or weighted scaled sum
uint kThreshold = 80; // threshold k (Top-K similarities)
var simItems = corrMatrix.CorrelatedEntities(item); // similar item group
foreach (int simItem in simItems)
{
if (booleanRatings[user, simItem])
{
float rating = ratings[user, simItem];
float average = ratings.ItemAverage(simItem);
float weight = (float) corrMatrix[simItem, item];
count ++;
normFactor += weight;
switch (aggregation)
{
case AggregationType.NormalSum:
sum += rating;
break;
case AggregationType.WeightedSum:
sum += weight * rating;
break;
case AggregationType.WeightedScaledSum:
sum += weight * (rating - average);
break;
}
if (--kThreshold == 0)
break;
}
}
if (count == 0 || normFactor.Equals(0))
{
prediction = 0;
}
else
{
switch (aggregation)
{
case AggregationType.NormalSum:
prediction = sum / count;
break;
case AggregationType.WeightedSum:
prediction = sum / normFactor;
break;
case AggregationType.WeightedScaledSum:
prediction = ratings.UserAverage(user);
prediction += sum / normFactor;
break;
}
}
if (prediction > MaxRating)
prediction = MaxRating;
if (prediction < MinRating)
prediction = MinRating;
return prediction;
}
public void Training(RatingData trainingData)
{
corrMatrix = new CorrelationMatrix(ratings.MaxItem);
corrMatrix.Correlation = correlation;
corrMatrix.Construct(trainingData, Globals.itemSim);
}
