private List <string> recommendNN(RecommendationMethod sAlgorithm, string sUserId, int cRecommendations)
{
Dictionary <string, double> sumWmovies = new Dictionary <string, double>();
Dictionary <string, double> similarUsers = calcSimilarUsers(sAlgorithm, sUserId);
foreach (string movie in m_trainMovieToUser.Keys)
{
if (m_trainMovieToUser[movie].Contains(sUserId))
{
continue;
}
foreach (string user in similarUsers.Keys)
{
if (m_trainMovieToUser[movie].Contains(user))
{
if (!sumWmovies.ContainsKey(movie))
{
sumWmovies.Add(movie, 0);
}
sumWmovies[movie] += similarUsers[user];
}
}
}
List <string> moviesSorted = sumWmovies.Keys.ToList();
moviesSorted.Sort((a, b) => sumWmovies[a].CompareTo(sumWmovies[b]));
return(moviesSorted.Take(cRecommendations).ToList());
}
public List <string> Recommend(RecommendationMethod sAlgorithm, string sUserId, int cRecommendations)
{
List <string> result = new List <string>();
switch (sAlgorithm)
{
case (RecommendationMethod.Popularity):
result = GetPopularItems(sUserId, cRecommendations);
break;
case (RecommendationMethod.Pearson):
result = GetTopItems(predictionEngine.getModel(PredictionMethod.Pearson), sUserId, cRecommendations);
break;
case (RecommendationMethod.Cosine):
result = GetTopItems(predictionEngine.getModel(PredictionMethod.Cosine), sUserId, cRecommendations);
break;
case (RecommendationMethod.BaseModel):
result = GetTopItems(predictionEngine.getModel(PredictionMethod.BaseModel), sUserId, cRecommendations);
break;
case (RecommendationMethod.Stereotypes):
result = GetTopItems(predictionEngine.getModel(PredictionMethod.Stereotypes), sUserId, cRecommendations);
break;
case (RecommendationMethod.NNPearson):
result = GetTopItemsBasedNN(new PearsonMethod(), sUserId, cRecommendations);
break;
case (RecommendationMethod.NNCosine):
result = GetTopItemsBasedNN(new CosineMethod(), sUserId, cRecommendations);
break;
case (RecommendationMethod.NNBaseModel):
result = GetTopItemsBasedNN(new BaseModelMethod(predictionEngine.getModel(PredictionMethod.BaseModel)), sUserId, cRecommendations);
break;
case (RecommendationMethod.NNJaccard):
result = GetTopItemsBasedNN(new JaccardMethod(), sUserId, cRecommendations);
break;
case (RecommendationMethod.CP):
result = GetTopItemsCp(sUserId, cRecommendations);
break;
case (RecommendationMethod.Jaccard):
result = GetTopItemsJaccard(sUserId, cRecommendations);
break;
}
return(result);
}
private Dictionary <string, double> calcSimilarUsers(RecommendationMethod sAlgorithm, string sUserId)
{
Dictionary <string, double> userW = new Dictionary <string, double>();
List <int> usedLocations = new List <int>();
double sumOfW = 0;
int numOfUsers = 0;
Random r = new Random();
while (numOfUsers < 500) //not going over all the users to save time
{
if (numOfUsers >= 20 && (sumOfW / (double)numOfUsers) > 0.8) //!!
{
break;
}
int location = (int)((m_train.Count - 1) * r.NextDouble());
if (usedLocations.Contains(location))
{
continue;
}
string user = m_train.Keys.ToList()[location];
if (user.Equals(sUserId) /*|| userW.ContainsKey(user)*/)
{
continue;
}
double w = 0;
if (sAlgorithm == RecommendationMethod.NNCosine)
{
w = calcWCosine(sUserId, user, "");
}
else if (sAlgorithm == RecommendationMethod.NNJaccard)
{
w = calcJaccardSimilarity(sUserId, user);
}
else if (sAlgorithm == RecommendationMethod.BaseModel)
{
w = calcBaseModelSimilarity(sUserId, user);
}
else
{
w = calcWPearson(sUserId, user, "");
}
if (w >= 0.5)//!!
{
sumOfW += w;
numOfUsers++;
userW.Add(user, w);
}
}
return(userW.OrderBy(x => x.Value).Take(20).ToDictionary(x => x.Key, x => x.Value));
}
public List <string> Recommend(RecommendationMethod sAlgorithm, string sUserId, int cRecommendations)
{
if (sAlgorithm == RecommendationMethod.Popularity)
{
return(recommendPopularity(sUserId, cRecommendations));
}
else if (sAlgorithm.ToString().StartsWith("NN"))
{
return(recommendNN(sAlgorithm, sUserId, cRecommendations));
}
else if (sAlgorithm == RecommendationMethod.Jaccard)
{
return(recommendCP(sUserId, cRecommendations, true));
}
else if (sAlgorithm == RecommendationMethod.CP)
{
return(recommendCP(sUserId, cRecommendations, false));
}
else //prediction
{
PredictionMethod predictionMethod = PredictionMethod.Pearson;
if (sAlgorithm == RecommendationMethod.BaseModel)
{
predictionMethod = PredictionMethod.BaseModel;
}
else if (sAlgorithm == RecommendationMethod.Cosine)
{
predictionMethod = PredictionMethod.Cosine;
}
else if (sAlgorithm == RecommendationMethod.Stereotypes)
{
predictionMethod = PredictionMethod.Stereotypes;
}
return(recommendPredictions(sUserId, cRecommendations, predictionMethod));
}
}
public List <string> Recommend(RecommendationMethod sAlgorithm, string sUserId, int cRecommendations)
{
throw new NotImplementedException();
}
