public void CreateLdaInput(string outputFile)
{
var map = new Mapping();
var outputLines = File.ReadAllLines(TagFile).Select(l =>
{
if (l.StartsWith("ERROR"))
return "";
var tokens = l.Split('\t');
int tagCount = Convert.ToInt32(tokens[1]);
string output = "";
if (tagCount > 0)
{
var tags = tokens[2].Split(new string[] { "##" }, StringSplitOptions.RemoveEmptyEntries);
output += map.ToInternalID(tokens[0]);
foreach (var tag in tags)
{
var parts = tag.Split(new string[] { "::" }, StringSplitOptions.None);
output += " " + map.ToInternalID(parts[0]) + ":" + parts[1];
}
}
return output;
}).Where(o => !String.IsNullOrEmpty(o));
File.WriteAllLines(outputFile, outputLines);
}
/// <summary>Load entity mappings from a file</summary>
/// <param name='filename'>the name of the file</param>
/// <returns>an object of type EntityMapping</returns>
public static IMapping LoadMapping(this string filename)
{
var mapping = new Mapping();
using ( var reader = new StreamReader(filename) )
{
string line;
while ( (line = reader.ReadLine()) != null )
{
if (line.Length == 0)
continue;
string[] tokens = line.Split('\t');
if (tokens.Length != 2)
throw new FormatException("Expected exactly 2 columns: " + line);
int internal_id = int.Parse(tokens[0]);
string external_id = tokens[1];
if (internal_id != mapping.NumberOfEntities)
throw new FormatException(string.Format("Expected ID {0}, not {1}, in line '{2}'", mapping.NumberOfEntities, internal_id, line));
mapping.internal_to_original.Add(external_id);
mapping.original_to_internal[external_id] = internal_id;
}
}
return mapping;
}
public override string ToLibFmFeatureVector(Mapping usersItemsMap)
{
string userClusterFeature = "", itemClusterFeature = "", aUserClusterFeature = "", aItemClusterFeature = "";
if (!string.IsNullOrEmpty(UserCluster))
userClusterFeature = usersItemsMap.ToInternalID(UserCluster).ToString();
if (!string.IsNullOrEmpty(ItemCluster))
itemClusterFeature = usersItemsMap.ToInternalID(ItemCluster).ToString();
if (!string.IsNullOrEmpty(AuxUserCluster))
aUserClusterFeature = usersItemsMap.ToInternalID(AuxUserCluster).ToString();
if (!string.IsNullOrEmpty(AuxItemCluster))
aItemClusterFeature = usersItemsMap.ToInternalID(AuxItemCluster).ToString();
string featVector = string.Format("{0} {1}:1 {2}:1",
Rating,
usersItemsMap.ToInternalID(User.Id),
usersItemsMap.ToInternalID(Item.Id));
if (userClusterFeature != "")
featVector += " " + userClusterFeature + ":1";
if (itemClusterFeature != "")
featVector += " " + itemClusterFeature + ":1";
if (aUserClusterFeature != "")
featVector += " " + aUserClusterFeature + ":1";
if (aItemClusterFeature != "")
featVector += " " + aItemClusterFeature + ":1";
return featVector;
}
public int GetMappedId(Mapping mapper)
{
if (_mappedId == -1)
_mappedId = mapper.ToInternalID(Id);
return _mappedId;
}
[Test()] public void TestComputeCorrelations2()
{
// load data from disk
var user_mapping = new Mapping();
var item_mapping = new Mapping();
var ratings = RatingData.Read("../../../../data/ml-100k/u1.base", user_mapping, item_mapping);
var p = new Pearson(ratings.AllUsers.Count, 200f);
Assert.AreEqual(-0.02788301f, p.ComputeCorrelation(ratings, EntityType.ITEM, 45, 311), 0.00001);
}
private static ITimedRatings readDataMapped(string data_file, ref MyMediaLite.Data.Mapping user_mappings, ref MyMediaLite.Data.Mapping item_mappings)
{
Console.WriteLine("Dataset: " + data_file);
Console.WriteLine(DateTime.Now);
var all_data = TimedRatingData.Read(data_file, user_mappings, item_mappings, TestRatingFileFormat.WITHOUT_RATINGS, true);
Console.Write(all_data.Statistics());
Console.WriteLine("Finished dataset read");
return(all_data);
}
public MovieLensCrossDomainContainer(int numDomains, bool randomClusters = false)
: base()
{
_mapper = new Mapping();
NumDomains = numDomains;
RandomClusters = randomClusters;
_itemsCluster = new Dictionary<string, int>();
for (int i = 0; i < numDomains; i++)
{
string dId = "ml" + i;
Domains.Add(dId, new Domain(dId));
}
}
public void CreateNmfClustersAmazon()
{
var reader = new AmazonReader(Paths.AmazonBooksRatings);
var dataset = new Dataset<ItemRating>(reader);
var userMapping = new Mapping();
var itemMapping = new Mapping();
var data = dataset.AllSamples.Select(ir => new
{
UserId = userMapping.ToInternalID(ir.User.Id),
ItemId = itemMapping.ToInternalID(ir.Item.Id),
Rating = Convert.ToDouble(ir.Rating)
}).ToList();
// users
int i = 0;
var uOut = File.ReadAllLines(Paths.AmazonBooksUsersCluster + ".lf").Select(l =>
{
var values = l.Split(new string[] { " " }, StringSplitOptions.RemoveEmptyEntries)
.Select(v => Convert.ToDouble(v)).ToList();
var maxIndex = values.IndexOf(values.Max());
return new { UserId = userMapping.ToOriginalID(i++), ClusterId = maxIndex };
}).Select(uc => string.Format("{0},{1}", uc.UserId, uc.ClusterId));
File.WriteAllLines(Paths.AmazonBooksUsersCluster + ".nmf.u", uOut);
// items
int j = 0;
var iOut = File.ReadAllLines(Paths.AmazonBooksUsersCluster + ".rf").Select(l =>
{
var values = l.Split(new string[] { " " }, StringSplitOptions.RemoveEmptyEntries)
.Select(v => Convert.ToDouble(v)).ToList();
var maxIndex = values.IndexOf(values.Max());
return new { ItemId = itemMapping.ToOriginalID(j++), ClusterId = maxIndex };
}).Select(ic => string.Format("{0},{1}", ic.ItemId, ic.ClusterId));
File.WriteAllLines(Paths.AmazonBooksUsersCluster + ".nmf.i", iOut);
}
private static void startUserKNN(string data)
{
MyMediaLite.Data.Mapping user_mapping = new MyMediaLite.Data.Mapping();
MyMediaLite.Data.Mapping item_mapping = new MyMediaLite.Data.Mapping();
ITimedRatings all_data = readDataMapped(data, ref user_mapping, ref item_mapping);
removeUserThreshold(ref all_data);
Console.WriteLine("Start iteration Test UserKNN");
//for (int i = 0; i < 5; i++) {
ITimedRatings validation_data = new TimedRatings(); // 10%
ITimedRatings test_data = new TimedRatings(); // 20%
ITimedRatings training_data = new TimedRatings(); // 70%
readAndSplitData(all_data, ref test_data, ref training_data, ref validation_data);
IPosOnlyFeedback training_data_pos = new PosOnlyFeedback <SparseBooleanMatrix> (); // 80%
for (int index = 0; index < training_data.Users.Count; index++)
{
training_data_pos.Add(training_data.Users [index], training_data.Items [index]);
}
MyMediaLite.ItemRecommendation.UserKNN recommender = new MyMediaLite.ItemRecommendation.UserKNN();
recommender.K = 80;
recommender.Q = 1;
recommender.Weighted = false;
recommender.Alpha = 0.5f;
recommender.Correlation = MyMediaLite.Correlation.BinaryCorrelationType.Jaccard;
recommender.Feedback = training_data_pos;
DateTime start_time = DateTime.Now;
recommender.Train();
Console.Write("Total Training time needed:");
Console.WriteLine(((TimeSpan)(DateTime.Now - start_time)).TotalMilliseconds);
Console.WriteLine("Final results in this iteration:");
var results = MyMediaLite.Eval.ItemsWeatherItemRecommender.EvaluateTime(recommender, validation_data, training_data, "VALIDATION ", false);
results = MyMediaLite.Eval.ItemsWeatherItemRecommender.EvaluateTime(recommender, test_data, training_data, "TEST ", false);
//}
}
public LibFmFeatureBuilder()
{
Mapper = new Mapping();
}
public new string ToLibFmFeatureVector(Mapping usersItemsMap)
{
return string.Format("{0} {1}:1 {2}:1", Rating, usersItemsMap.ToInternalID(User.Id), usersItemsMap.ToInternalID(Item.Id));
}
public Clusterer(IDataset<ItemRating> dataset)
{
_dataset = dataset;
_userMapping = new Mapping();
_itemMapping = new Mapping();
}
private static void TrainForItemPrediction(Mapping userMapping, Mapping itemMapping, String[] args)
{
var training_data = ItemData.Read(trainingCompactFileForItems, userMapping, itemMapping);
itemRecommender = new CustomBPRMF();
if (File.Exists(Path.Combine(args[2], "model")))
{
Console.WriteLine("Skipping training, Loading saved model");
itemRecommender.LoadModel(Path.Combine(args[2], "model"));
itemRecommender.Feedback = training_data;
return;
}
Console.WriteLine("Training model for Item Prediction, this may take a while...");
itemRecommender.Feedback = training_data;
itemRecommender.NumFactors = 50;
itemRecommender.NumIter = 100;
itemRecommender.Train();
itemRecommender.SaveModel(Path.Combine(args[2], "model"));
}
/// <summary>
/// Predict the rating of the item by users
/// </summary>
private static void PredictRating(string[] args)
{
Console.WriteLine("Predicting ratings for Users...");
String outputFile = Path.Combine(args[2], outputFileForRatings);
testingFile = Path.Combine(args[1], testingFile);
testingFileModified = Path.Combine(args[2], testingFileModified);
ModifyTestingFileForRating(testingFile, testingFileModified);
float minRating = 1;
float maxRating = 5;
var userMapping = new Mapping();
var itemMapping = new Mapping();
var trainingData = StaticRatingData.Read(trainingCompactFileForRatings, userMapping, itemMapping, RatingType.FLOAT, TestRatingFileFormat.WITH_RATINGS, false);
var testUsers = trainingData.AllUsers; // users that will be taken into account in the evaluation
var candidate_items = trainingData.AllItems; // items that will be taken into account in the evaluation
var testData = StaticRatingData.Read(testingFileModified, userMapping, itemMapping, RatingType.FLOAT, TestRatingFileFormat.WITH_RATINGS, false);
var recommender = new BiasedMatrixFactorization();
recommender.MinRating = minRating;
recommender.MaxRating = maxRating;
recommender.Ratings = trainingData;
recommender.NumFactors = 30;
recommender.NumIter = 100;
recommender.RegI = 0.04F;
recommender.RegU = 0.04F;
//recommender.BiasReg = 0.09F;
recommender.FrequencyRegularization = true;
recommender.BoldDriver = true;
recommender.LearnRate = 0.07F;
Stopwatch timeKeeper = new Stopwatch();
timeKeeper.Start();
recommender.Train();
timeKeeper.Stop();
Console.WriteLine("time passed for training rating prediction model: " + timeKeeper.ElapsedMilliseconds);
// measure the accuracy on the test data set
timeKeeper = new Stopwatch();
timeKeeper.Start();
var results = recommender.Evaluate(testData);
timeKeeper.Stop();
Console.WriteLine("time passed for rating prediction: " + timeKeeper.ElapsedMilliseconds);
Console.WriteLine("RMSE={0}", results["RMSE"]);
recommender.WritePredictions(testData, outputFile, userMapping, itemMapping, "{0}-{1},{2}", "userID-itemID,rating");
}
private static void PredictPurchase(string user, Dictionary<string, List<string>> itemsPurchases, Mapping userMapping, Mapping itemMapping, String[] args)
{
// Get the master set - feedback
string logFile = Path.Combine(args[2], "ItemPrediction.log");
StreamWriter writer = null;
try
{
writer = new StreamWriter(logFile, true);
float result1 = 0.0F;
//Stopwatch timeKeeper = new Stopwatch();
//timeKeeper.Start();
rankedPurchase = new SortedDictionary<float, List<MeanAverPrecisionDetails>>();//Comparer<float>.Create((x, y) => y.CompareTo(x)));
// this user's average ratings - does he buy items which are low rated?
float[] userAverageAndNumberOfRatings = GetAverageRatingAndNumberOfRatingsGivenByUser(userMapping.ToInternalID(user));
float averageRatingByUser = userAverageAndNumberOfRatings[0];
float numberOfRatingsByUser = userAverageAndNumberOfRatings[1];
//bool isUserActive = UserActivity(userMapping.ToInternalID(user));
Dictionary<string, int> userCategories = GetUserPurchaseCategories(userMapping.ToInternalID(user));
foreach (String item in itemsPurchases.Keys)
{
float resultForItemSimilarity = itemRecommender.GetNearestItemSimilarity(userMapping.ToInternalID(user), itemMapping.ToInternalID(item));
//float resultForUserSimilarity = itemRecommender.GetNearestUserSimilarity(userMapping.ToInternalID(user), itemMapping.ToInternalID(item));
float averageRatingForItem = GetAverageRatingForItem(item);
bool predictedValueOfPurchase = false;
float resultFromItemCategories = CompareUserItemCategories(userCategories, GetItemCategories(item));
float result = resultForItemSimilarity;
//if (averageRatingForItem > userAverageRating)
// result += ((averageRatingForItem - userAverageRating)/ averageRatingForItem);
if (averageRatingForItem > 4.0)
{
resultFromItemCategories = +((averageRatingForItem - 4.0F) / averageRatingForItem);
}
////result += userActivity;
//if (!isUserActive)
//{
// result -= 0.2f;//(0.1F * result);
//}
if (resultFromItemCategories > 0.45)
{
result = result - (1.0F * resultFromItemCategories);
}
//if (numberOfRatingsByUser > 10)
//{
// result = result - 0.1F;
//}
writer.WriteLine("User: {0}, Item: {1}, hasBought: {2}, result: {3}, rating for Item {4}", user, item, itemsPurchases[item][0], result, averageRatingForItem);
result1 +=result;
if (result < 0.4)
{
predictedValueOfPurchase = true;
}
List<MeanAverPrecisionDetails> elementsToAddToRankedPurchaseList = new List<MeanAverPrecisionDetails>();
elementsToAddToRankedPurchaseList.Add(new MeanAverPrecisionDetails(user, item, predictedValueOfPurchase ? 1 : 0, itemsPurchases[item][0].Equals("1") ? 1 : 0, result, Int32.Parse(itemsPurchases[item][1])));
if (!rankedPurchase.ContainsKey(result))
{
rankedPurchase.Add(result, elementsToAddToRankedPurchaseList);
}
else
{
List<MeanAverPrecisionDetails> exisitingValues = rankedPurchase[result];
exisitingValues.AddRange(elementsToAddToRankedPurchaseList);
rankedPurchase[result] = exisitingValues;
}
// User has actually bought the item
if (itemsPurchases[item][0].Equals("1"))
{
totalActualPurchases++;
if (predictedValueOfPurchase)
{
// We have correctly identified that the user has purchased item
truePositives++;
}
else
{
// Missed predicting the user's purchase
falseNegatives++;
}
}
else
{
// User has actually not bought this item
if (predictedValueOfPurchase)
{
// Wrongly predicted that the user bought the item
falsePositives++;
}
else
{
// Correctly predicted that the user did not buy item
trueNegatives++;
}
}
// Find out if this user has bought items from the same category(ies)
}
CalculateAveragePrecision(args);
writer.Close();
//timeKeeper.Stop();
//Console.WriteLine("time passed PredictPurchase: " + timeKeeper.ElapsedMilliseconds);
}
catch (Exception) {}
finally
{
if (writer != null)
{
writer.Close();
}
}
}
public static void RecommentItems(Mapping userMapping, Mapping itemMapping, String[] args)
{
Console.WriteLine("Predicting Items for Users...");
testingFileForItems = Path.Combine(args[1], testingFileForItems);
if (File.Exists(testingFileForItems))
{
StreamReader reader = null;
try
{
reader = new StreamReader(testingFileForItems);//(args[0]);
String line = null;
int userCount = 0;
if (reader != null)
{
line = reader.ReadLine();
line = reader.ReadLine();
while (line != null)
{
String[] labels = line.Split(',');
String[] user_item = labels[0].Split('-');
string user = user_item[0];
Dictionary<string, List<string>> itemsPurchase = new Dictionary<string, List<string>>();
List<string> itemPurchaseDetails = new List<string>();
itemPurchaseDetails.Add(labels[1]); // bought or not, 0 or 1
itemPurchaseDetails.Add(labels[2]); // rank of this item
itemsPurchase.Add(user_item[1], itemPurchaseDetails);
line = reader.ReadLine();
while (line != null && user.Equals(line.Split(',')[0].Split('-')[0]))
{
labels = line.Split(',');
user_item = labels[0].Split('-');
if (!user.Equals(user_item[0]))
break;
itemPurchaseDetails = new List<string>();
itemPurchaseDetails.Add(labels[1]); // bought or not, 0 or 1
itemPurchaseDetails.Add(labels[2]);
itemsPurchase.Add(user_item[1], itemPurchaseDetails);
line = reader.ReadLine();
}
PredictPurchase(user, itemsPurchase, userMapping, itemMapping, args);
userCount++;
}
}
double precision = (double) truePositives / (truePositives + falsePositives);
double recall = (double)truePositives / (truePositives + falseNegatives);//(totalActualPurchases);
double f1Measure = 2 * ((precision * recall) / (precision + recall));
double meanAveragePrecision = averagePrecisionSum / userCount;
Console.WriteLine("precision : {0}, recall : {1}, f1 : {2}, MAP : {3} ", precision, recall, f1Measure, meanAveragePrecision);
}
catch (IOException e)
{
Console.WriteLine("exception reading the training file : {0}", e.Message);
}
finally
{
if (reader != null)
{
reader.Close();
}
}
}
}
/// <summary>Read movie data from a file</summary>
/// <param name="filename">the name of the file to be read from</param>
/// <param name="item_mapping">ID mapping for the movies</param>
public void Read(string filename, Mapping item_mapping)
{
using ( var reader = new StreamReader(filename) )
Read(reader, item_mapping);
}
/// <summary>Read movie data from a StreamReader</summary>
/// <param name="reader">a StreamReader to be read from</param>
/// <param name="item_mapping">ID mapping for the movies</param>
public void Read(StreamReader reader, Mapping item_mapping)
{
movie_list = new List<Movie>();
IMDB_KEY_To_ID = new Dictionary<string, int>();
var separators = new string[] { "::" };
string line;
while (!reader.EndOfStream)
{
line = reader.ReadLine();
string[] tokens = line.Split(separators, StringSplitOptions.None);
if (tokens.Length != 3)
throw new FormatException("Expected exactly 3 columns: " + line);
int movie_id = item_mapping.ToInternalID(tokens[0]);
string movie_imdb_key = tokens[1];
//string[] movie_genres = tokens[2].Split('|');
int movie_year = 1900;
string movie_title = movie_imdb_key;
movie_list.Add(new Movie(movie_id, movie_title, movie_year, movie_imdb_key));
IMDB_KEY_To_ID[movie_imdb_key] = movie_id;
}
}
