org.grouplens.lenskit

Interface RatingPredictor

All Superinterfaces:

ItemScorer

All Known Implementing Classes:

AbstractRatingPredictor, QuantizedRatingPredictor, SimpleRatingPredictor

public interface RatingPredictor
extends ItemScorer

Predict user ratings. A rating predictor is like an ItemScorer, but its output will be predicted ratings.

Note: The fact that this interface extends ItemScorer is deprecated.

Author:

GroupLens Research

Compatibility: Public

Method Summary

Methods

Modifier and Type	Method and Description
boolean	canUseHistory() Query whether this predictor can actually use user history.
SparseVector	predict(long user, Collection<Long> items) Predict the user's preference for a collection of items.
double	predict(long user, long item) Prdict a user's rating for a single item.
void	predict(long user, MutableSparseVector predictions) Predict for items in a vector.
SparseVector	predict(UserHistory<? extends Event> profile, Collection<Long> items) Predict a collection of items for the user using a history.
double	predict(UserHistory<? extends Event> profile, long item) Predict a user's preference for an item using a history.
void	predict(UserHistory<? extends Event> profile, MutableSparseVector predictions) Predict for items in a vector.
SparseVector	score(long user, Collection<Long> items) Deprecated. Use predict(long, Collection)
double	score(long user, long item) Deprecated. Use predict(long, long)
void	score(long user, MutableSparseVector scores) Deprecated. Use predict(long, MutableSparseVector)
SparseVector	score(UserHistory<? extends Event> profile, Collection<Long> items) Deprecated. Use predict(UserHistory, Collection)
double	score(UserHistory<? extends Event> profile, long item) Deprecated. Use predict(UserHistory, long)
void	score(UserHistory<? extends Event> profile, MutableSparseVector scores) Deprecated. Use predict(UserHistory, MutableSparseVector)

Method Detail

canUseHistory

boolean canUseHistory()

Query whether this predictor can actually use user history.

Specified by:

canUseHistory in interface ItemScorer

Returns:

true if the history passed to one of the history-based methods may be used, and false if it will be ignored.

predict

double predict(long user,
             long item)

Prdict a user's rating for a single item.

Parameters:

user - The user ID for whom to generate a prediction.

item - The item ID whose rating is to be predicted.

Returns:

The predicted preference, or Double.NaN if no preference can be predicted.

See Also:

predict(UserHistory, MutableSparseVector)

predict

@Nonnull
SparseVector predict(long user,
                           @Nonnull
                           Collection<Long> items)

Predict the user's preference for a collection of items.

Parameters:

user - The user ID for whom to generate predicts.

items - The items to predict for.

Returns:

A mapping from item IDs to predicted preference. This mapping may not contain all requested items.

See Also:

predict(UserHistory, MutableSparseVector)

predict

void predict(long user,
           @Nonnull
           MutableSparseVector predictions)

Predict for items in a vector. The key domain of the provided vector is the items whose predictions are requested, and the predict method sets the values for each item to its predict (or unsets it, if no prediction can be provided). The previous values are discarded.

Parameters:

user - The user ID.

predictions - The prediction output vector. Its key domain is the items to score.

See Also:

predict(UserHistory, MutableSparseVector)

predict

double predict(@Nonnull
             UserHistory<? extends Event> profile,
             long item)

Predict a user's preference for an item using a history. If possible, the provided history is used instead of whatever history may be in the database or model.

Parameters:

profile - The user's profile.

item - The item to predict.

Returns:

The predict, or Double.NaN if no predict can be computed.

See Also:

predict(UserHistory, MutableSparseVector)

predict

@Nonnull
SparseVector predict(@Nonnull
                           UserHistory<? extends Event> profile,
                           @Nonnull
                           Collection<Long> items)

Predict a collection of items for the user using a history. If possible, the provided history is used instead of whatever history may be in the database or model.

Parameters:

profile - The user's profile

items - The items to predict.

Returns:

A mapping from item IDs to predicts. This mapping may not contain all requested items — ones for which the predictr cannot compute a predict will be omitted.

See Also:

predict(UserHistory, MutableSparseVector)

predict

void predict(@Nonnull
           UserHistory<? extends Event> profile,
           @Nonnull
           MutableSparseVector predictions)

Predict for items in a vector. The key domain of the provided vector is the items whose ratings should be predicted, and the predict method sets the values for each item to its predict (or unsets it, if no prediction can be provided). The previous values are discarded.

If the user has rated any items to be predicted, the algorithm should not just use their rating as the predict — it should compute a predict in the normal fashion. If client code wants to substitute ratings, it is easy to do so as a separate step or wrapper interface.

Parameters:

profile - The user history.

predictions - The prediction output vector.

score

@Deprecated
double score(long user,
                      long item)

Deprecated. Use predict(long, long)

Score a single item.

Specified by:

score in interface ItemScorer

Parameters:

user - The user ID for whom to generate a score.

item - The item ID to score.

Returns:

The score, or Double.NaN if no score can be generated.

See Also:

ItemScorer.score(UserHistory, MutableSparseVector)

score

@Deprecated
@Nonnull
SparseVector score(long user,
                                    @Nonnull
                                    Collection<Long> items)

Deprecated. Use predict(long, Collection)

Score a collection of items.

Specified by:

score in interface ItemScorer

Parameters:

user - The user ID for whom to generate scores.

items - The item to score.

Returns:

A mapping from item IDs to predicted preference. This mapping may not contain all requested items.

See Also:

ItemScorer.score(UserHistory, MutableSparseVector)

score

@Deprecated
void score(long user,
                    @Nonnull
                    MutableSparseVector scores)

Deprecated. Use predict(long, MutableSparseVector)

Score items in a vector. The key domain of the provided vector is the items to score, and the score method sets the values for each item to its score (or unsets it, if no score can be provided). The previous values are discarded.

Specified by:

score in interface ItemScorer

Parameters:

user - The user ID.

scores - The score vector.

See Also:

ItemScorer.score(UserHistory, MutableSparseVector)

score

@Deprecated
double score(@Nonnull
                      UserHistory<? extends Event> profile,
                      long item)

Deprecated. Use predict(UserHistory, long)

Score an item for the user using a history. If possible, the provided history is used instead of whatever history may be in the database or model.

Specified by:

score in interface ItemScorer

Parameters:

profile - The user's profile.

item - The item to score.

Returns:

The score, or Double.NaN if no score can be computed.

See Also:

ItemScorer.score(UserHistory, MutableSparseVector)

score

@Deprecated
@Nonnull
SparseVector score(@Nonnull
                                    UserHistory<? extends Event> profile,
                                    @Nonnull
                                    Collection<Long> items)

Deprecated. Use predict(UserHistory, Collection)

Score a collection of items for the user using a history. If possible, the provided history is used instead of whatever history may be in the database or model.

Specified by:

score in interface ItemScorer

Parameters:

profile - The user's profile

items - The items to score.

Returns:

A mapping from item IDs to scores. This mapping may not contain all requested items — ones for which the scorer cannot compute a score will be omitted.

See Also:

ItemScorer.score(UserHistory, MutableSparseVector)

score

@Deprecated
void score(@Nonnull
                    UserHistory<? extends Event> profile,
                    @Nonnull
                    MutableSparseVector scores)

Deprecated. Use predict(UserHistory, MutableSparseVector)

Score items in a vector. The key domain of the provided vector is the items to score, and the score method sets the values for each item to its score (or unsets it, if no score can be provided). The previous values are discarded.

If the user has rated any items to be scored, the algorithm should not just use their rating as the score — it should compute a score in the normal fashion. If client code wants to substitute ratings, it is easy to do so as a separate step or wrapper interface.

Specified by:

score in interface ItemScorer

Parameters:

profile - The user history.

scores - The score vector.