/// <summary>
/// Returns a distance-based similarity score for person1 and person2
/// </summary>
/// <param name="user1"></param>
/// <param name="user2"></param>
/// <returns></returns>
public double CalculateSimilarity(UserRating user1, UserRating user2)
{
var sharedItems = new List <uint>();
// Get the list of shared movies
foreach (var movieId in user1.GetRatedMovies())
{
if (user2.HasRating(movieId))
{
sharedItems.Add(movieId);
}
}
// if they have no ratings in common, return 0
if (sharedItems.Count == 0)
{
return(0);
}
// Add up the squares of all the differences
var sumOfSquares = 0.0;
foreach (var movieID in sharedItems)
{
sumOfSquares += Math.Pow(user1.GetRating(movieID) - user2.GetRating(movieID), 2);
}
return(1 / (1 + sumOfSquares));
}
/// <summary>
/// Gets recommendations for a person by using a weighted average
/// of every other user's rankings
/// </summary>
/// <param name=""></param>
/// <param name=""></param>
/// <param name=""></param>
/// <returns></returns>
public List <Tuple <double, uint> > GetRecommendations(UserRating person)
{
var totals = new Dictionary <uint, double>();
var simSums = new Dictionary <uint, double>();
foreach (var other in UserRatings.Values)
{
if (other.UserID == person.UserID)
{
continue;
}
var similarity = CalculateSimilarity(person, other);
//# ignore scores of zero or lower
if (similarity <= 0)
{
continue;
}
foreach (var item in other.GetRatedMovies())
{
// only score movies I haven't seen yet
if (!person.HasRating(item) || person.GetRating(item) == 0)
{
// Similarity * Score
if (totals.ContainsKey(item))
{
totals[item] += other.GetRating(item) * similarity;
simSums[item] += similarity;
}
else
{
totals.Add(item, other.GetRating(item) * similarity);
simSums.Add(item, similarity);
}
}
}
}
var rankings = new List <Tuple <double, uint> >();
// Create the normalized list
foreach (var item in totals.Keys)
{
rankings.Add(new Tuple <double, uint>(totals[item] / simSums[item], item));
}
// Return the sorted list
rankings.Sort();
rankings.Reverse();
return(rankings);
}
/// <summary>
/// Returns the best matches for person from the prefs dictionary.
/// Number of results and similarity function are optional params.
/// </summary>
/// <param name="user"></param>
/// <returns></returns>
public List <Tuple <double, uint> > GetTopMatches(UserRating user, int n)
{
var scores = new List <Tuple <double, uint> >();
foreach (var other in UserRatings.Values)
{
if (other.UserID != user.UserID)
{
scores.Add(new Tuple <double, uint>(CalculateSimilarity(user, other), other.UserID));
}
}
// Sort the list so the highest scores appear at the top
scores.Sort();
scores.Reverse();
return(scores.GetRange(0, n));
}