The singular values on my recommender vectors come out: 90, 10, 1.2,
1.1, 1.0, 0.95. This was playing with your R code. Based on this,
I'm adding the QR stuff to my visualization toolkit.
Lance
On Sat, Jul 2, 2011 at 10:15 PM, Lance Norskog goks...@gmail.com wrote:
All pairwise distances
On Sat, Jul 2, 2011 at 11:34 AM, Sean Owen sro...@gmail.com wrote:
Yes that's well put. My only objection is that this sounds like you're
saying that there is a systematic problem with the ordering, so it
will usually help to pick any different ordering than the one you
thought was optimal.
That is the point of the exponential in the example that I gave you. The
top few recommendations are nearly stable. It is the lower ranks that are
really churned up. This has the property that you state.
On Sat, Jul 2, 2011 at 12:45 PM, Salil Apte sa...@offlinelabs.com wrote:
I really like
I would be very surprised if java.lang.Random exhibited this behavior. It
isn't *that* bad.
On Sat, Jul 2, 2011 at 6:49 PM, Lance Norskog goks...@gmail.com wrote:
For full Random Projection, a lame random number generator
(java.lang.Random) will generate a higher standard deviation than a
I wasn't thinking when I typed that post. An orthonormal projection always
preserves distances since it is just a generalized reflection/rotation.
Preserving all dot products (including to self) also implies distances are
preserved because |x-y|_2 = x \dot x - 2 x \dot y + y \dot y.
On Sat, Jul
On Sun, Jul 3, 2011 at 8:05 AM, Ted Dunning ted.dunn...@gmail.com wrote:
For instance, if the recommendation engine recommends B if you have seen A
and there is little other way to discover C which is ranked rather low (and
thus never seen), then there is no way for the engine to even get
Please, what I am doing wrong?
trunk$ bin/mahout seq2sparse -i wikipedia-all-cat -o
wikipedia-vectors-analysed -seq -a org.apache.lucene.analysis.Analyzer -ow
Running on hadoop, using HADOOP_HOME=/usr/local/hadoop
No HADOOP_CONF_DIR set, using /usr/local/hadoop/src/conf
11/07/03 19:52:50 INFO
Oh, some code works, some not (these stack overflows). I'm confused.
Seems that I need to run everything in the fresh isolated environment
On 3 July 2011 21:58, Ted Dunning ted.dunn...@gmail.com wrote:
Sean,
Any ideas how that tiny little commit caused this?
On Sun, Jul 3, 2011 at 5:07 AM,
Insightful and interesting. But it seems that quantitative measure of
gain/loss from different methods would help.
The question is how you measure the gain?
One example: suppose recommendations are ignored by 99% of the users and
there is some measurable action (now or later) from 1% of the
I don't see why one would believe that the randomly selected items
farther down the list are more likely to engage a user. If anything,
the recommender says they are less likely to be engaging.
(Or put another way, by this reasoning, we ought to pick
recommendations at random.)
I do think that
I whipped up a MurmurHash Random and the Gaussian plot is much
cleaner. The MH version is exactly as fast as the java.util.Random-
j.u.R makes a new int in every cycle, and MH makes a new long, so does
half as many cycles.
On Sun, Jul 3, 2011 at 12:12 AM, Ted Dunning ted.dunn...@gmail.com wrote:
On Sun, Jul 3, 2011 at 1:08 PM, Sean Owen sro...@gmail.com wrote:
I don't see why one would believe that the randomly selected items
farther down the list are more likely to engage a user. If anything,
the recommender says they are less likely to be engaging.
There are two issues with this
Roughly.
But remember, a single recommendation isn't the end of the game. If this is
the last recommendation to ever be made, dithering doesn't help at all.
On Sun, Jul 3, 2011 at 1:02 PM, Konstantin Shmakov kshma...@gmail.comwrote:
It seems that as long as recommenders are dealing with the
13 matches
Mail list logo