Implementation

From Pigbert Wiki

Command Line Arguments

  java -dataset (s|m|n) -task TaskClassName
        
  [-testPartition m/n]
        
  [-homemode] [-debug] 

  [-noout] [-noerr] [-redirect_debug fileFromRoot] [-redirect_out fileFromRoot] [-redirect_err fileFromRoot]

The Task Package

task.reformat

• CheckNFFiles renames each datafile to remove the ur_ or mv_ suffix.
• CreateGaintFile concatenate all rating files.
• DeleteDate

Delete the date information in the baseset. Currently only run on NF dataset.

• Index1stLine
• IndexIdentity
  • IndexIdentityGreedily
  • IndexIdentityThoroughly
• PartitionNFsTestset

partitions the NF's qualifying.txt or probe.txt into N subsets, each of which is in charge of "#ratings/N" many ratings, rounded to items, and in order. The output files do not contain date information. The partitioned files are stored under dataroot/tests(N)/... Running this program will also prints out three stats.txt files, which contain information about the statistical distribution of the test data.

• ProbeDataMarker mark the big dataset to be testdata (+) or basedata (-)
• RatingFileSorter sort the rating file SubID ASC.
• ResetDelimiter reset the delimiters to \t.

task.statistics

• CheckVotingProfileShiftingShape test if there exists voting habits != linear shift, for all datasets?

Dataset

RatingDataset

• RatingDataset
  • RatingDatamap
  • RatingDatacache

• IdentityRatingSet<MainID,SubsID>
  • IdentityRatingArray
  • IdentityRatingScanner

IDRatings<SubsID> getAllRatingsOfAsArray(ID)

Map<SubsID,Rating> getAllRatingsOfAsMap(ID)

Map<SubsID,Byte> getAllRatingsOfAsByteMap(ID)

Set<SubsID> getAllIDsRelatedTo(ID)

Set<MainID> getAllInvolvedIDs()

• IDRatings<T extends ID>

• RatingHolder

IdentityDataset

• IdentityInfoSet
• IdentityInfoStore
• IdentityInfoCache
• IdentityInfoSimulator