public MainForm()
{
InitializeComponent();
IRater rate = new LinearRater(-4, 2, 3, 1);
IComparer compare = new CorrelationUserComparer();
recommender = new UserCollaborativeFilterRecommender(compare, rate, 50);
if (File.Exists(savedModel))
{
try
{
recommender.Load(savedModel);
rtbOutput.Text = "Loaded model from file";
EnableForm(true);
}
catch
{
rtbOutput.Text = "Saved model is corrupt";
}
}
//RecommenderTests.TestAllRecommenders();
//RecommenderTests.FindBestRaterWeights();
}
public static void FindBestRaterWeights()
{
using (StreamWriter w = new StreamWriter("rater-weights.csv", true))
{
w.WriteLine("down,up,view,dl,rmse,solved,total,precision,rank");
}
var once = new User(0, "");
var options = new List <dynamic>();
for (double up = 0.0; up < 5.0; up += 1)
{
for (double down = -5.0; down <= 0.0; down += 1)
{
for (double dl = 0.0; dl < 5.0; dl += 1)
{
for (double view = 0.0; view < 5.0; view += 1)
{
options.Add(new { up, down, dl, view });
}
}
}
}
var dbp = new UserBehaviorDatabaseParser();
var db = dbp.LoadUserBehaviorDatabase("UserBehaviour.txt");
var sp = new DaySplitter(db, 3);
var cp = new CorrelationUserComparer();
Parallel.ForEach(options, set =>
{
try
{
var rate = new LinearRater(set.down, set.up, set.view, set.dl);
var mfr = new MatrixFactorizationRecommender(20, rate);
//var mfr = new UserCollaborativeFilterRecommender(cp, rate, set.features);
mfr.Train(sp.TrainingDB);
var score = mfr.Score(sp.TestingDB, rate);
var results = mfr.Test(sp.TestingDB, 100);
lock (once)
{
using (StreamWriter w = new StreamWriter("rater-weights.csv", true))
{
w.WriteLine(set.down + "," + set.up + "," + set.view + "," + set.dl + "," + score.RootMeanSquareDifference + "," + results.UsersSolved + "," + results.TotalUsers + "," + results.AveragePrecision + "," + results.AverageRecall);
}
}
}
catch (Exception ex)
{
File.WriteAllText("errors.txt", ex.ToString());
}
});
}
public static void TestAllRecommenders()
{
UserBehaviorDatabaseParser dbp = new UserBehaviorDatabaseParser();
UserBehaviorDatabase db = dbp.LoadUserBehaviorDatabase("UserBehaviour.txt");
//var ubt = new UserBehaviorTransformer(db);
//var uart = ubt.GetUserArticleRatingsTable();
//uart.SaveSparcityVisual("sparcity.bmp");
//uart.SaveUserRatingDistribution("distrib.csv");
//uart.SaveArticleRatingDistribution("distriba.csv");
var rate = new LinearRater();
var sp = new DaySplitter(db, 5);
var uc = new CorrelationUserComparer();
//var rr = new RandomRecommender();
//rr.Train(sp.TrainingDB);
//ScoreResults scores5 = rr.Score(sp.TestingDB, rate);
//TestResults results5 = rr.Test(sp.TestingDB, 30);
var ubc = new UserCollaborativeFilterRecommender(uc, rate, 30);
var mfr = new MatrixFactorizationRecommender(30, rate);
var icf = new ItemCollaborativeFilterRecommender(uc, rate, 30);
var hbr = new HybridRecommender(ubc, mfr, icf);
hbr.Train(sp.TrainingDB);
ScoreResults scores1 = hbr.Score(sp.TestingDB, rate);
TestResults results1 = hbr.Test(sp.TestingDB, 30);
ubc = new UserCollaborativeFilterRecommender(uc, rate, 30);
mfr = new MatrixFactorizationRecommender(30, rate);
icf = new ItemCollaborativeFilterRecommender(uc, rate, 30);
ubc.Train(sp.TrainingDB);
ScoreResults scores2 = ubc.Score(sp.TestingDB, rate);
TestResults results2 = ubc.Test(sp.TestingDB, 30);
mfr.Train(sp.TrainingDB);
ScoreResults scores3 = mfr.Score(sp.TestingDB, rate);
TestResults results3 = mfr.Test(sp.TestingDB, 30);
icf.Train(sp.TrainingDB);
ScoreResults scores4 = icf.Score(sp.TestingDB, rate);
TestResults results4 = icf.Test(sp.TestingDB, 30);
using (StreamWriter w = new StreamWriter("results.csv"))
{
w.WriteLine("model,rmse,users,user-solved,precision,recall");
w.WriteLine("UCF," + scores2.RootMeanSquareDifference + "," + results2.TotalUsers + "," + results2.UsersSolved + "," + results2.AveragePrecision + "," + results2.AverageRecall);
w.WriteLine("SVD," + scores3.RootMeanSquareDifference + "," + results3.TotalUsers + "," + results3.UsersSolved + "," + results3.AveragePrecision + "," + results3.AverageRecall);
w.WriteLine("ICF," + scores4.RootMeanSquareDifference + "," + results4.TotalUsers + "," + results4.UsersSolved + "," + results4.AveragePrecision + "," + results4.AverageRecall);
w.WriteLine("HR," + scores1.RootMeanSquareDifference + "," + results1.TotalUsers + "," + results1.UsersSolved + "," + results1.AveragePrecision + "," + results1.AverageRecall);
}
}
public List <Suggestion> GetSuggest(UserBehavior db, long userId)
{
IRater rater = new SimpleRater();
IComparer comparer = new CorrelationUserComparer();
recommender = new ItemCollaborativeFilterRecommender(comparer, rater, 50);
recommender.Train(db);
var suggestion = recommender.GetSuggestions(userId, 500);
return(suggestion);
}
public ActionResult Index(string search = "")
{
if (search == "")
{
int id = Convert.ToInt32(Session["id"].ToString());
string email = context.login.Where(m => m.Id == id).FirstOrDefault().Email;
int realid = context.students.Where(m => m.Email == email).FirstOrDefault().Id;
IRater rate = new LinearRater(-4, 2, 0.5, 1);
IComparer compare = new CorrelationUserComparer();
recommender = new UserCollaborativeFilterRecommender(compare, rate, 200);
UserBehaviorDatabaseParser parser = new UserBehaviorDatabaseParser();
UserBehaviorDatabase db1 = parser.LoadUserBehaviorDatabase("/Data/NewBehavior.txt");
UserBehaviorTransformer ubt = new UserBehaviorTransformer(db1);
recommender.Train(db1);
int userId;
int ratings;
userId = realid;
ratings = 2;
List <Suggestion> result = new List <Suggestion>();
List <RecomendedArticles> rem = new List <RecomendedArticles>();
List <Suggestion> result2 = new List <Suggestion>();
RecomendedArticles recom;
if (ratings >= 1 && ratings <= 100)
{
new GetRecommendation {
UserID = userId, Ratings = ratings
};
result = recommender.GetSuggestions(userId, ratings);
result2 = recommender.GetSuggestions(userId, 6);
}
foreach (Suggestion suggestion in result)
{
var ye = context.ufiles.Where(m => m.Id == suggestion.ArticleID).FirstOrDefault();
recom = new RecomendedArticles()
{
Name = ye.Name,
UpdatedFileName = ye.UpdatedFileName,
UplodedBy = ye.UplodedBy,
Description = ye.Description,
Filename = ye.Filename,
imagepath = ye.imagepath,
UplodedDate = ye.UplodedDate,
Rating = suggestion.Rating,
Id = ye.Id,
};
rem.Add(recom);
}
NRViewModel recomendedArticles = new NRViewModel();
recomendedArticles.uplodedFiles = context.ufiles.OrderByDescending(m => m.Id).Take(6).ToList();
recomendedArticles.RecomendedArticles = rem;
return(View(recomendedArticles));
}
else
{
NRViewModel recomendedArticles = new NRViewModel();
recomendedArticles.uplodedFiles = context.ufiles.OrderByDescending(m => m.Id).Where(m => m.Name.Contains(search)).Take(6).ToList();
if (recomendedArticles.uplodedFiles == null)
{
ViewBag.messagea = "no item found";
}
ViewBag.message = "search";
return(View(recomendedArticles));
}
}