|
||||||||||
PREV CLASS NEXT CLASS | FRAMES NO FRAMES | |||||||||
SUMMARY: NESTED | FIELD | CONSTR | METHOD | DETAIL: FIELD | CONSTR | METHOD |
java.lang.Objectorg.apache.mahout.cf.taste.impl.recommender.slopeone.MemoryDiffStorage
public final class MemoryDiffStorage
An implementation of DiffStorage
that merely stores item-item diffs in memory. It is fast, but can
consume a great deal of memory.
Constructor Summary | |
---|---|
MemoryDiffStorage(DataModel dataModel,
Weighting stdDevWeighted,
boolean compactAverages,
long maxEntries)
Creates a new . |
Method Summary | |
---|---|
RunningAverage |
getAverageItemPref(long itemID)
|
RunningAverage |
getDiff(long itemID1,
long itemID2)
|
RunningAverage[] |
getDiffs(long userID,
long itemID,
PreferenceArray prefs)
|
FastIDSet |
getRecommendableItemIDs(long userID)
|
void |
refresh(java.util.Collection<Refreshable> alreadyRefreshed)
Triggers "refresh" -- whatever that means -- of the implementation. |
java.lang.String |
toString()
|
void |
updateItemPref(long itemID,
float prefDelta,
boolean remove)
Updates internal data structures to reflect an update in a preference value for an item. |
Methods inherited from class java.lang.Object |
---|
clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait |
Constructor Detail |
---|
public MemoryDiffStorage(DataModel dataModel, Weighting stdDevWeighted, boolean compactAverages, long maxEntries) throws TasteException
Creates a new .
See SlopeOneRecommender
for the meaning of
stdDevWeighted
. If compactAverages
is set, this uses alternate data structures
(CompactRunningAverage
versus FullRunningAverage
) that use almost 50% less memory but
store item-item averages less accurately. maxEntries
controls the maximum number of
item-item average preference differences that will be tracked internally. After the limit is reached, if
a new item-item pair is observed in the data it will be ignored. This is recommended for large datasets.
The first maxEntries
item-item pairs observed in the data are tracked. Assuming that item
ratings are reasonably distributed among users, this should only ignore item-item pairs that are very
infrequently co-rated by a user. The intuition is that data on these infrequently co-rated item-item
pairs is less reliable and should be the first that is ignored. This parameter can be used to limit the
memory requirements of SlopeOneRecommender
, which otherwise grow as the square of the number of
items that exist in the DataModel
. Memory requirements can reach gigabytes with only about 10000
items, so this may be necessary on larger datasets.
stdDevWeighted
- see SlopeOneRecommender
compactAverages
- if true
, use CompactRunningAverage
instead of FullRunningAverage
internallymaxEntries
- maximum number of item-item average preference differences to track internally
java.lang.IllegalArgumentException
- if maxEntries
is not positive or dataModel
is null
TasteException
Method Detail |
---|
public RunningAverage getDiff(long itemID1, long itemID2)
getDiff
in interface DiffStorage
RunningAverage
encapsulating the average difference in preferences between items
corresponding to itemID1
and itemID2
, in that direction; that is, it's
the average of item 2's preferences minus item 1's preferencespublic RunningAverage[] getDiffs(long userID, long itemID, PreferenceArray prefs)
getDiffs
in interface DiffStorage
userID
- user ID to get diffs foritemID
- itemID to assessprefs
- user's preferendces
RunningAverage
s for that user's item-item diffspublic RunningAverage getAverageItemPref(long itemID)
getAverageItemPref
in interface DiffStorage
RunningAverage
encapsulating the average preference for the given itempublic void updateItemPref(long itemID, float prefDelta, boolean remove)
DiffStorage
Updates internal data structures to reflect an update in a preference value for an item.
updateItemPref
in interface DiffStorage
itemID
- item to update preference value forprefDelta
- amount by which preference value changed (or its old value, if being removedremove
- if true
, operation reflects a removal rather than change of preferencepublic FastIDSet getRecommendableItemIDs(long userID) throws TasteException
getRecommendableItemIDs
in interface DiffStorage
TasteException
public void refresh(java.util.Collection<Refreshable> alreadyRefreshed)
Refreshable
Triggers "refresh" -- whatever that means -- of the implementation. The general contract is that any should always leave itself in a consistent, operational state, and that the refresh atomically updates internal state from old to new.
refresh
in interface Refreshable
alreadyRefreshed
- s that are known to have already been
refreshed as a result of an initial call to a method on some
object. This ensure that objects in a refresh dependency graph aren't refreshed twice
needlessly.public java.lang.String toString()
toString
in class java.lang.Object
|
||||||||||
PREV CLASS NEXT CLASS | FRAMES NO FRAMES | |||||||||
SUMMARY: NESTED | FIELD | CONSTR | METHOD | DETAIL: FIELD | CONSTR | METHOD |