SunYatong

感谢提醒，已修改

After training, You can run the `predict(int userIdx, itemIdx)` function with the testMatrix. // after training for (MatrixEntry me : testMatrix) { int userIdx = me.row(); // user int itemIdx...

@mvijaikumar Command line doesn't support this. You can denote the test set before training with TestSetSplitter, and add the code after training. ``` // after training for (MatrixEntry me :...

So you want to get the prediction of arbitrary user-item pairs? @mvijaikumar

So you do want to get the prediction of arbitrary user-item pairs... @mvijaikumar Just use ``` predict(userIdx, itemIdx) ```

Yes, it cost a lot of time when converting a tensor to a matrix for splitting.

You can check the configuration and testcase of FM-based methods, e.g., FMSGDTestCase and FMALSTestCase.

I am afraid not. You can check the similarity directory for all the similarity measurements we have.

Hi, @bsesar. Thanks for your interest in our work. The function of loading and saving models are not completed yet both in Java API and command line API. You can...

@bsesar user/item Id is mapped to user/item factor matrix's raw index by "userMappingData" and "itemMappingData" respectively, you can find them in AbstractRecommender.