public static Dictionary<int, Movie> ParseFromFolder(string folder, int limit = NOLIMIT)
{
if (!Directory.Exists(folder)) return null;
Dictionary<int, Movie> movies = new Dictionary<int, Movie>();
foreach (var file in Directory.EnumerateFiles(folder))
{
if (limit > -1 && movies.Count == limit)
break;
string[] lines = File.ReadAllLines(file);
int movieId = Int32.Parse(lines[0].Replace(":", string.Empty));
Movie movie = new Movie(movieId);
for (int i = 1; i < lines.Length; i++)
{
string[] fields = lines[i].Split(',');
int userId = Int32.Parse(fields[0]);
int rating = Int32.Parse(fields[1]);
movie.AddOrUpdate(userId, rating);
}
movies.Add(movieId, movie);
}
return movies;
}
static void Main(string[] args)
{
Control.UseNativeMKL();
bool readFromFile = true;
if (!readFromFile)
{
Dictionary<int, Movie> allTestMovies = MovieParser.ParseFromFolder(@"C:\Users\Frederik\Desktop\download\download\training_set\training_set", 1000);
List<Movie> rawProbes = ProbeParser.Parse(@"C:\Users\Frederik\Desktop\download\download\probe.txt", 1000);
List<Movie> tempProbes = new List<Movie>();
foreach (var probe in rawProbes)
{
if (allTestMovies.ContainsKey(probe.Id))
{
Movie movie = new Movie(probe.Id);
foreach (var userId in probe.Ratings.Keys)
{
int rating = allTestMovies[probe.Id].GetRating(userId);
movie.AddOrUpdate(userId, rating);
allTestMovies[probe.Id].RemoveRating(userId);
}
tempProbes.Add(movie);
}
}
rawProbes = null;
GC.Collect();
//WriteListToFile("testMovies.txt", allTestMovies.Values.ToList());
allTestMovies = null;
GC.Collect();
WriteListToFile("probeMovie.txt", tempProbes);
tempProbes = null;
GC.Collect();
}
//List<Movie> testMovies = MovieParser.Parse("testMovies.txt");
List<Movie> probes = MovieParser.Parse("probeMovie.txt");
Preprocesser preprocesser = new Preprocesser();
preprocesser.BuildRatingMatrix(probes);
SparseMatrix matrix = preprocesser.GetMeanMatrix();
Factorization factorization = new Factorization();
factorization.Factorize(matrix);
//double sum = matrix.SumMatrix();
//Console.WriteLine(sum);
Console.ReadLine();
}
public static List<Movie> Parse(string file, int limit = NOLIMIT)
{
if (!File.Exists(file)) return null;
List<Movie> movies = new List<Movie>();
int previousId = default(int);
int movieId = default(int);
Dictionary<int, int> ratings = new Dictionary<int, int>();
using (StreamReader reader = new StreamReader(file))
{
while (!reader.EndOfStream)
{
string line = reader.ReadLine();
if (line.Contains(":"))
{
if (movieId != previousId)
{
Movie movie = new Movie(movieId);
movie.AddOrUpdate(ratings);
movies.Add(movie);
previousId = movieId;
ratings = new Dictionary<int, int>();
}
movieId = Int32.Parse(line.Replace(":", string.Empty));
}
else
{
string[] fields = line.Split(',');
int userId = Int32.Parse(fields[0]);
int rating = Int32.Parse(fields[1]);
ratings.Add(userId, rating);
}
}
Movie m = new Movie(movieId);
m.AddOrUpdate(ratings);
movies.Add(m);
}
return movies;
}
public static List<Movie> Parse(string file, int limit = NOLIMIT)
{
if (!File.Exists(file)) return null;
List<Movie> movies = new List<Movie>();
int movieId = default(int);
List<int> userIds = new List<int>();
using (StreamReader reader = new StreamReader(file))
{
while (!reader.EndOfStream)
{
if (limit > -1 && movies.Count == limit)
break;
string line = reader.ReadLine();
if (line.Contains(":"))
{
if (movieId != default(int))
{
Movie movie = new Movie(movieId);
foreach (var userId in userIds)
{
movie.AddOrUpdate(userId, 0);
}
movies.Add(movie);
userIds = new List<int>();
}
movieId = Int32.Parse(line.Replace(":", string.Empty));
}
else
{
int userId = Int32.Parse(line);
userIds.Add(userId);
}
}
}
return movies;
}
