Stuart Peterson

There is a SwapChangefunction in all the classes under metric. Do you guys know what is it for? It is quite confusing.

I have trained a ListNet model, and when tested on a test data, the model did nothing. Specifically, I am using -indri to check on the ranked list. But -indri returned a list with an exact same order as my test data ranked list. Here is a snapshot of my model file: (##) ListNet (##) Epochs = 1500 (##) No. of features = 57 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 0 0 0 -0.02878251201717241...

I have trained a ListNet model, and when tested on a test data, the model did nothing. Specifically, I am using -indri to check on the ranked list. But -indri returned a list with an exact same order as my test data ranked list. Here is a snapshot of my model file: (##) ListNet (##) Epochs = 1500 (##) No. of features = 57 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 0 0 0 -0.02878251201717241...

if that is the case, then what will the impacts of randomized partition on my training data?

Do we need to order the training data file by query type and relevancy? I got some errors if I do not order the data by query type, is it something normal?

Should be "0 qid:001 1:0.5 2:0.7 #docid"?

-tc is the parameter used for number of threshold candidates for tree spliting. I knew from the documentation that the default value is 256. However, if we have less than 256 features, will the -1 parameter limits the features used? How the algorithm will deal with such situation?

May I know what should we put in the target value column for test data as the relevancy of the document pair is unknow? Standard file format: "<target><qid> <feature>:<value> ... # <info>"</info></value></feature></qid></target> Note: 'target' represents relevancy label but the label is not available for test data.