lucenenet-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Robert Oschler <robert.osch...@gmail.com>
Subject Re: Does the latest version of lucene.net contain the Lucene 4.0 speed improvements for fuzzy queries?
Date Thu, 09 Jul 2015 09:13:37 GMT
Hello Simon,

Ok.  I got excited when I saw the following using statement in the 3.0.3
build:

using Lucene.Net.Codecs.Lucene40;

But from what you are saying I take it the 4.0 label does not refer to
Lucene 4.0.

>> You could probably take the FuzzyQuery class from Lucene 4.0 and port it
to the Lucene 3.0.3 code base. That's the only way you can get those
improvements while still using a stable version of Lucene.net.

Thanks.  I'll try grabbing the FuzzyQuery class and converting the Java
code to C#.  Hopefully there aren't too many dependencies in that unit that
I'll have to drag in.

>> Regarding the Azure library... are you using Azure? If so, are you using
several worker machines that need to share an index? You've not mentioned
anything about your current setup to help you evaluate if switching to
AzureDirectory will be an improvement or not.

Yes I am using Azure, but I am just getting started so I have not set up
Lucene yet.  I'm trying to decide my current setup right now.  Given my
expected usage profile (The index will probably receive a few hundred new
updates over the course of the day.  Over time, there could well be a
hundred thousand sentences or so.), do you have any suggestions?  I'll want
the lowest latency I can give my users that I can get when searching the
index.  Note, I am more concerned with the performance of index lookups.
Update/modifications can take a few seconds if needed since they will be
much less frequent.

Thanks,
Robert



On Thu, Jul 9, 2015 at 5:01 AM, Simon Svensson <sisve@devhost.se> wrote:

> Hi,
>
> The latest stable version of Lucene.net is v3.0.3, and does not contain
> any code changes from the the higher versioned java code. There's currently
> a 4.8 port in progress, but it's not stable enough yet.
>
> You could probably take the FuzzyQuery class from Lucene 4.0 and port it
> to the Lucene 3.0.3 code base. That's the only way you can get those
> improvements while still using a stable version of Lucene.net.
>
> Regarding the Azure library... are you using Azure? If so, are you using
> several worker machines that need to share an index? You've not mentioned
> anything about your current setup to help you evaluate if switching to
> AzureDirectory will be an improvement or not.
>
> // Simon
>
>
>
> On 09/07/15 10:51, Robert Oschler wrote:
>
>> Hello,
>>
>> Does the latest version of lucene.net contain the Lucene 4.0 speed
>> improvements for fuzzy queries?  If not, is there any way to get those
>> improvements?  I saw this experimental Lucene.net 4.0 branch on
>> Apache.org,
>> but it seems to be inactive now and I don't know how stable it is:
>>
>> https://svn.apache.org/repos/asf/lucene.net/branches/Lucene.Net_4e/
>>
>> If so, what is the exact version of lucene.net I should be using to have
>> those improvements?
>>
>> Also, I saw this post on using Azure blobs to speed up server side
>> processing:
>>
>> https://code.msdn.microsoft.com/windowsazure/Azure-Library-for-83562538
>>
>> Are the improvements using this technique substantial enough to warrant
>> using it?  My "documents" are average sized sentences.  The index will
>> probably receive a few hundred new updates over the course of the day.
>> Over time, there could well be a hundred thousand sentences or so.
>>
>>
>


-- 
Thanks,
Robert Oschler
Twitter -> http://twitter.com/roschler
http://www.RobotsRule.com/
http://www.Robodance.com/

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