-
Notifications
You must be signed in to change notification settings - Fork 15
Usage
Takahiko Ito edited this page Feb 4, 2015
·
2 revisions
Likelike input files have the following two components split by tab,
- instance id
- features - features for the instance. Each feature is split by tab
The following is the sample input.
0 776 2919 4576 4485 2380 4261 3622 1456 2109 3227
1 4358 4350 559 1136 2848 1345 1428 4212 2681 120
2 3614 3035 4592 1191 441 1408 734 1005 4826 33
3 691 878 1938 3381 127 4283 2269 3814 2908 1864
4 3781 603 1262 1988 1309 864 3511 729 1454 2845
5 3183 3575 180 4367 3616 4072 824 4976 4021 3361
Extract related (or similar) examples.
$bin/likelike lsh
-input INPUT use INPUT as input resource
-output OUTPUT use OUTPUT as outupt prefix
[-depth DEPTH] use DEPTH as size of concatinations (default 1)
[-iterate ITERATE] use ITERATE as the number of hash keys (default 1)
[-maxRecommend SIZE] use SIZE as the maximum number of recommendation for one example
[-help] show this message
Extract features from related examples.
$bin/likelike featureExtraction
-input INPUT use INPUT as input resource (output file of lsh)
-output OUTPUT use OUTPUT as outupt prefix
-feature FEATURE use feature as input resource (input file of lsh)
[-help] show this message
Validate the output from LSH.
$bin/likelike validate
-input INPUT use INPUT as input resource (output file of lsh)
-output OUTPUT use OUTPUT as outupt prefix
-feature FEATURE use FEATURE as an input dir (input file of lsh)
-threshold VALUE use VALUE as the threshold to output pairs
[-help] show this message