public double Estimate(Item item)
{
Preference pref = user.GetPreferenceFor(item.ID);
if (pref == null)
{
return(Double.NaN);
}
double correlationValue = correlation.GetItemCorrelation(recommendedItem, item);
return((1.0 + correlationValue) * pref.Value);
}
public double Estimate(Item item)
{
Pair <Item, Item> pair = new Pair <Item, Item>(toItem, item);
if (rescorer.IsFiltered(pair))
{
return(Double.NaN);
}
double originalEstimate = correlation.GetItemCorrelation(toItem, item);
return(rescorer.Rescore(pair, originalEstimate));
}
public double Estimate(Item item)
{
RunningAverage average = new FullRunningAverage();
foreach (Item toItem in toItems)
{
Pair <Item, Item> pair = new Pair <Item, Item>(toItem, item);
if (rescorer.IsFiltered(pair))
{
continue;
}
double estimate = correlation.GetItemCorrelation(toItem, item);
estimate = rescorer.Rescore(pair, estimate);
average.AddDatum(estimate);
}
return(average.Average);
}
/// <summary>
/// <p>Builds a list of item-item correlations given an {@link ItemCorrelation} implementation and a
/// <see cref="DataModel">DataModel</see>, rather than a list of {@link ItemItemCorrelation}s.</p>
/// <p>It's valid to build a <see cref="GenericItemCorrelation"/> this way, but perhaps missing some of the point
/// of an item-based Recommender. Item-based recommenders use the assumption that item-item correlations
/// are relatively fixed, and might be known already independent of user preferences. Hence it is useful
/// to inject that information, using {@link GenericItemCorrelation(java.util.Collection)}.</p>
/// </summary>
/// <param name="otherCorrelation">otherCorrelation other {@link ItemCorrelation} to get correlations from</param>
/// <param name="dataModel">dataModel data Model to get {@link Item}s from</param>
public GenericItemCorrelation(ItemCorrelation otherCorrelation, DataModel dataModel)
{
List <Item> items = EnumeratorUtils.EnumerableToList <Item>(dataModel.GetItems());
int size = items.Count;
for (int i = 0; i < size; i++)
{
Item item1 = items[i];
for (int j = i + 1; j < size; j++)
{
Item item2 = items[j];
double correlation = otherCorrelation.GetItemCorrelation(item1, item2);
Dictionary <Item, Double> map = null;
if (!correlationMaps.TryGetValue(item1, out map))
{
map = new Dictionary <Item, Double>(1009);
correlationMaps.Add(item1, map);
}
map.Add(item2, correlation);
}
}
}
/// <summary>
/// <p>Builds a list of item-item correlations given an {@link ItemCorrelation} implementation and a
/// <see cref="DataModel">DataModel</see>, rather than a list of {@link ItemItemCorrelation}s.</p>
/// <p>It's valid to build a <see cref="GenericItemCorrelation"/> this way, but perhaps missing some of the point
/// of an item-based Recommender. Item-based recommenders use the assumption that item-item correlations
/// are relatively fixed, and might be known already independent of user preferences. Hence it is useful
/// to inject that information, using {@link GenericItemCorrelation(java.util.Collection)}.</p>
/// </summary>
/// <param name="otherCorrelation">otherCorrelation other {@link ItemCorrelation} to get correlations from</param>
/// <param name="dataModel">dataModel data Model to get {@link Item}s from</param>
public GenericItemCorrelation(ItemCorrelation otherCorrelation, DataModel dataModel)
{
List<Item> items = EnumeratorUtils.EnumerableToList<Item>(dataModel.GetItems());
int size = items.Count;
for (int i = 0; i < size; i++)
{
Item item1 = items[i];
for (int j = i + 1; j < size; j++)
{
Item item2 = items[j];
double correlation = otherCorrelation.GetItemCorrelation(item1, item2);
Dictionary<Item, Double> map = null;
if (!correlationMaps.TryGetValue(item1, out map))
{
map = new Dictionary<Item, Double>(1009);
correlationMaps.Add(item1, map);
}
map.Add(item2, correlation);
}
}
}