mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sean Owen <>
Subject Re: [Taste] Sanity Check and Questions
Date Thu, 18 Jun 2009 16:37:08 GMT
Still seems a little funny. I am away from the code otherwise I would check
- forget if I ever implemented weighting in the standard correlation
similarity metrics. A pure correlation does not account for whether you
overlapped in 10 or 1000 items. This sort of weighting exists elsewhere but
I forget about here. It might help.

Also did not quite answer your other question. There is no standard notion
of 'explain' since the algorithms are quite different but look for the
recommendedBecause method.

On Jun 18, 2009 5:28 PM, "Grant Ingersoll" <> wrote:

On Jun 18, 2009, at 12:12 PM, Sean Owen wrote: > That would make sense only
if your user had alrea...
No, there is not 100% overlap between the 'philes, but there was a high
degree of overlap.  If I reduce the overlap, then I start to see some of the
'philes recommendations creeping in.  That's what was so curious.

Again, the tricky part here is to not read too much into made up data, but
it does seem interesting as a thought experiment.

> On Thu, Jun 18, 2009 at 4:54 PM, Grant Ingersoll<>
wrote: >> >> Actually, I ...

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message