mahout-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sean Owen" <sro...@gmail.com>
Subject Re: PriorityQueue in NearestNUserNeighborhood
Date Fri, 07 Nov 2008 10:13:49 GMT
The only catch there as I remember is the need to know the smallest
element -- but also the need to limit the size and therefore know the
largest element too. So I think that's why I couldn't use the data
structure. I could be missing something.

On Thu, Nov 6, 2008 at 10:47 PM, Otis Gospodnetic
<otis_gospodnetic@yahoo.com> wrote:
> Hi,
>
> I see this in the NearestNUserNeighborhood:
>
>          ListIterator<UserCorrelationPair> iterator = queue.listIterator(queue.size());
>          while (iterator.hasPrevious()) {
>            if (theCorrelation <= iterator.previous().theCorrelation) {
>              iterator.next();
>              break;
>            }
>          }
>
> This looks like a priority queue.... if so, I wonder if it would be clearer to just use
the java.util.PriorityQueue or maybe even a variation of the one in Lucene?  Not sure if there's
a need for this change, just thinking our loud as I browse Taste code.
>
> Oh, I see Hadoop has its own PQ, too:
>  http://hadoop.apache.org/core/docs/current/api/org/apache/hadoop/util/PriorityQueue.html
> which seems to be a slightly modified variation of the Lucene version:
>  http://hudson.zones.apache.org/hudson/job/Lucene-trunk/javadoc/org/apache/lucene/util/PriorityQueue.html
>
> Otis
> --
> Sematext -- http://sematext.com/ -- Lucene - Solr - Nutch
>
>

Mime
View raw message