LSH with Scikit in Python, evaluated by MML

[Djoels]

Hi everybody,

I've been looking at the external item recommender, for evaluating kNN with Locality-Sensitive Hashing in Python, and am wondering how I can best reintroduce the results into MML in a pragmatic way, knowing that the same data is used. Is there anyone who has done this in the past?

[Alejandro Bellogín]

Hi Djoels,

if you are only interested in evaluation, in RiVal we have isolated that part and tested with MML, LensKit and Mahout (see [1]).

RankSys also contains interesting evaluation metrics and dimensions (like novelty or diversity).

Best,

Alejandro

[1] http://dl.acm.org/citation.cfm?doid=2645710.2645746

[Djoels]

Thank you very much for these! I will have a look at them.

Best wishes

Op donderdag 23 februari 2017 18:12:56 UTC+1 schreef Alejandro Bellogín:

[Zeno Gantner]

Hi Julien, 

I'd also suggest to try those toolkits (and tell us about your experiences). 

If you want to use results from other systems, you can always use the ExternalItemRecommender. 

Let us know in case you have trouble using it, or other kinds of feedback. 

Cheers, 

   Z. 

--
You received this message because you are subscribed to the Google Groups "MyMediaLite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mymedialite+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.