You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@bloodhound.apache.org by Andrej Golcov <an...@digiverse.si> on 2012/11/16 18:12:22 UTC
Re: [Apache Bloodhound] #260: Search improvements
+ 1 to improve search relevance. I'm also interested in helping on subject.
IMHO, we should also consider Woosh [4] as possible solution as python-only
library to bypass external dependency on java.
[1] and [2] are interesting links on implementation of Trac Search plugin
with Woosh.
[1]
https://groups.google.com/forum/?fromgroups=#!msg/trac-dev/sbU-g0C6kvk/1_juL29aAtQJ
[2]http://trac.edgewall.org/wiki/AdvancedSearch
[3]http://trac.edgewall.org/wiki/SearchRefactoring
[4]http://pypi.python.org/pypi/Whoosh/
On 16 November 2012 16:19, Apache Bloodhound <
bloodhound-dev@incubator.apache.org> wrote:
> #260: Search improvements
> --------------------------+----------------------
> Reporter: jdreimann | Owner: gjm
> Type: enhancement | Status: accepted
> Priority: major | Milestone:
> Component: trac core | Version:
> Resolution: | Keywords:
> --------------------------+----------------------
> Changes (by gjm):
>
> * owner: nobody => gjm
> * status: new => accepted
>
>
> Comment:
>
> Search is an area that I do want to look into more generally. I am
> thinking of looking into solr integration. There appear to be a few
> plugins for this that might be worth checking out. I will see if I can do
> a quick investigation so I will assign this to myself, at least for the
> short term.
>
> --
> Ticket URL: <https://issues.apache.org/bloodhound/ticket/260#comment:3>
> Apache Bloodhound <https://issues.apache.org/bloodhound/>
> The Apache Bloodhound (incubating) issue tracker
>
Re: [Apache Bloodhound] #260: Search improvements
Posted by Olemis Lang <ol...@gmail.com>.
It seems to me that results are sorted by date . So I initially I was
thinking of filtering search results stream provided by search
contributors and tag each one with meta-data (e.g. page rank) . All
this by adding an interface and maybe some request filters ...
This is not contradictory with the ideas mentioned before , as e.g.
Woosh and Solr (... I was also thinking of remote hosted solutions
like Google PageRank and alike ...) integration may come based on a
similar mechanism together with an ExtensionOption to select search
results tagging approach .
Just a few ideas .
PS: I have not studied previous suggestions in detail . I'm just
mentioning what came in to my mind in first place for you to consider
.
;)
On 11/16/12, Gary Martin <ga...@wandisco.com> wrote:
> Excellent! Solr may well complicate setup so I was interested in looking
> for python solutions. I was wondering if there was anything that made some
> use of something like http://nltk.org but that may well be overkill.
>
> A few relevant plugins at trac-hacks:
> http://trac-hacks.org/wiki/FullTextSearchPlugin
> http://trac-hacks.org/wiki/TracAdvancedSearchPlugin
>
> It should be noted that there are also some simpler things that might help
> to improve search. For instance, I am not sure if there is a ticket for
> this but I note that a refinement in search terms in the search box results
> in any changes in the filters being removed. It would be very good if that
> was fixed!
>
> Cheers,
> Gary
>
>
> On 16 November 2012 17:12, Andrej Golcov <an...@digiverse.si> wrote:
>
>> + 1 to improve search relevance. I'm also interested in helping on
>> subject.
>>
>> IMHO, we should also consider Woosh [4] as possible solution as
>> python-only
>> library to bypass external dependency on java.
>>
>> [1] and [2] are interesting links on implementation of Trac Search plugin
>> with Woosh.
>>
>> [1]
>>
>> https://groups.google.com/forum/?fromgroups=#!msg/trac-dev/sbU-g0C6kvk/1_juL29aAtQJ
>> [2]http://trac.edgewall.org/wiki/AdvancedSearch
>> [3]http://trac.edgewall.org/wiki/SearchRefactoring
>> [4]http://pypi.python.org/pypi/Whoosh/
>>
>> On 16 November 2012 16:19, Apache Bloodhound <
>> bloodhound-dev@incubator.apache.org> wrote:
>>
>> > #260: Search improvements
>> > --------------------------+----------------------
>> > Reporter: jdreimann | Owner: gjm
>> > Type: enhancement | Status: accepted
>> > Priority: major | Milestone:
>> > Component: trac core | Version:
>> > Resolution: | Keywords:
>> > --------------------------+----------------------
>> > Changes (by gjm):
>> >
>> > * owner: nobody => gjm
>> > * status: new => accepted
>> >
>> >
>> > Comment:
>> >
>> > Search is an area that I do want to look into more generally. I am
>> > thinking of looking into solr integration. There appear to be a few
>> > plugins for this that might be worth checking out. I will see if I can
>> do
>> > a quick investigation so I will assign this to myself, at least for
>> > the
>> > short term.
>> >
>> > --
>> > Ticket URL: <https://issues.apache.org/bloodhound/ticket/260#comment:3>
>> > Apache Bloodhound <https://issues.apache.org/bloodhound/>
>> > The Apache Bloodhound (incubating) issue tracker
>> >
>>
>
>
>
> --
> Gary Martin
> Lead Developer | WANdisco <http://www.wandisco.com/>
> *
> *
> **
> *Join us this October at Subversion Live
> 2012<http://www.wandisco.com/svn-live-2012>
> *
>
--
Regards,
Olemis.
Blog ES: http://simelo-es.blogspot.com/
Blog EN: http://simelo-en.blogspot.com/
Featured article:
Re: [Apache Bloodhound] #260: Search improvements
Posted by Olemis Lang <ol...@gmail.com>.
On 11/18/12, Andrej Golcov <an...@digiverse.si> wrote:
> I suggest we define functionality that we want to achieve with improved
> search and then select an appropriate technical solution.
+1
[...]
>
> What do you think about new BEP for Search and following discussion on the
> mail list?
>
If you think it's the right way to go ... just add it . It should
easier to write by selecting Proposals templates when creating wiki
page ;)
--
Regards,
Olemis.
Blog ES: http://simelo-es.blogspot.com/
Blog EN: http://simelo-en.blogspot.com/
Featured article:
Re: [Apache Bloodhound] #260: Search improvements
Posted by Andrej Golcov <an...@digiverse.si>.
I suggest we define functionality that we want to achieve with improved
search and then select an appropriate technical solution. Otherwise it is
hard to select the best approach.
What do you think about new BEP for Search and following discussion on the
mail list?
Regards, Andrej
On 17 November 2012 06:57, Olemis Lang <ol...@gmail.com> wrote:
> On 11/16/12, Gary Martin <ga...@wandisco.com> wrote:
> > On 16 November 2012 17:12, Andrej Golcov <an...@digiverse.si> wrote:
> >
> >> + 1 to improve search relevance. I'm also interested in helping on
> >> subject.
> >>
> >> IMHO, we should also consider Woosh [4] as possible solution as
> >> python-only
> >> library to bypass external dependency on java.
> >>
> >> [1] and [2] are interesting links on implementation of Trac Search
> plugin
> >> with Woosh.
> >>
> >> [1]
> >>
> >>
> https://groups.google.com/forum/?fromgroups=#!msg/trac-dev/sbU-g0C6kvk/1_juL29aAtQJ
> >> [2]http://trac.edgewall.org/wiki/AdvancedSearch
> >> [3]http://trac.edgewall.org/wiki/SearchRefactoring
> >> [4]http://pypi.python.org/pypi/Whoosh/
> >>
> >
> > Excellent! Solr may well complicate setup so I was interested in looking
> > for python solutions. I was wondering if there was anything that made
> some
> > use of something like http://nltk.org but that may well be overkill.
> >
>
> If you ask me instead of Solr (which looks to me like a big-thing
> standalone J2EE web application) we should be thinking of using Lucene
> directly at a lower level
>
> "Solr uses the Lucene Java search library at its core for full-text
> indexing and search" [1]_
>
> We could interface with it via PyLucene [2]_
>
> > A few relevant plugins at trac-hacks:
> > http://trac-hacks.org/wiki/FullTextSearchPlugin
> > http://trac-hacks.org/wiki/TracAdvancedSearchPlugin
> >
> [...]
>
> FWIW I prefer to move forward with the second . It already offers Solr
> backend and claims to be designed with extensions points to add more .
> So it seems to be more appropriate to create either Whoosh or PyLucene
> backend ... or maybe both ...
>
> .. [1] Apache Solr
> (http://lucene.apache.org/solr/)
>
> .. [2] Welcome to PyLucene
> (http://lucene.apache.org/pylucene/)
>
> --
> Regards,
>
> Olemis.
>
> Blog ES: http://simelo-es.blogspot.com/
> Blog EN: http://simelo-en.blogspot.com/
>
> Featured article:
>
Re: [Apache Bloodhound] #260: Search improvements
Posted by Olemis Lang <ol...@gmail.com>.
On 11/16/12, Gary Martin <ga...@wandisco.com> wrote:
> On 16 November 2012 17:12, Andrej Golcov <an...@digiverse.si> wrote:
>
>> + 1 to improve search relevance. I'm also interested in helping on
>> subject.
>>
>> IMHO, we should also consider Woosh [4] as possible solution as
>> python-only
>> library to bypass external dependency on java.
>>
>> [1] and [2] are interesting links on implementation of Trac Search plugin
>> with Woosh.
>>
>> [1]
>>
>> https://groups.google.com/forum/?fromgroups=#!msg/trac-dev/sbU-g0C6kvk/1_juL29aAtQJ
>> [2]http://trac.edgewall.org/wiki/AdvancedSearch
>> [3]http://trac.edgewall.org/wiki/SearchRefactoring
>> [4]http://pypi.python.org/pypi/Whoosh/
>>
>
> Excellent! Solr may well complicate setup so I was interested in looking
> for python solutions. I was wondering if there was anything that made some
> use of something like http://nltk.org but that may well be overkill.
>
If you ask me instead of Solr (which looks to me like a big-thing
standalone J2EE web application) we should be thinking of using Lucene
directly at a lower level
"Solr uses the Lucene Java search library at its core for full-text
indexing and search" [1]_
We could interface with it via PyLucene [2]_
> A few relevant plugins at trac-hacks:
> http://trac-hacks.org/wiki/FullTextSearchPlugin
> http://trac-hacks.org/wiki/TracAdvancedSearchPlugin
>
[...]
FWIW I prefer to move forward with the second . It already offers Solr
backend and claims to be designed with extensions points to add more .
So it seems to be more appropriate to create either Whoosh or PyLucene
backend ... or maybe both ...
.. [1] Apache Solr
(http://lucene.apache.org/solr/)
.. [2] Welcome to PyLucene
(http://lucene.apache.org/pylucene/)
--
Regards,
Olemis.
Blog ES: http://simelo-es.blogspot.com/
Blog EN: http://simelo-en.blogspot.com/
Featured article:
Re: [Apache Bloodhound] #260: Search improvements
Posted by Gary Martin <ga...@wandisco.com>.
Excellent! Solr may well complicate setup so I was interested in looking
for python solutions. I was wondering if there was anything that made some
use of something like http://nltk.org but that may well be overkill.
A few relevant plugins at trac-hacks:
http://trac-hacks.org/wiki/FullTextSearchPlugin
http://trac-hacks.org/wiki/TracAdvancedSearchPlugin
It should be noted that there are also some simpler things that might help
to improve search. For instance, I am not sure if there is a ticket for
this but I note that a refinement in search terms in the search box results
in any changes in the filters being removed. It would be very good if that
was fixed!
Cheers,
Gary
On 16 November 2012 17:12, Andrej Golcov <an...@digiverse.si> wrote:
> + 1 to improve search relevance. I'm also interested in helping on subject.
>
> IMHO, we should also consider Woosh [4] as possible solution as python-only
> library to bypass external dependency on java.
>
> [1] and [2] are interesting links on implementation of Trac Search plugin
> with Woosh.
>
> [1]
>
> https://groups.google.com/forum/?fromgroups=#!msg/trac-dev/sbU-g0C6kvk/1_juL29aAtQJ
> [2]http://trac.edgewall.org/wiki/AdvancedSearch
> [3]http://trac.edgewall.org/wiki/SearchRefactoring
> [4]http://pypi.python.org/pypi/Whoosh/
>
> On 16 November 2012 16:19, Apache Bloodhound <
> bloodhound-dev@incubator.apache.org> wrote:
>
> > #260: Search improvements
> > --------------------------+----------------------
> > Reporter: jdreimann | Owner: gjm
> > Type: enhancement | Status: accepted
> > Priority: major | Milestone:
> > Component: trac core | Version:
> > Resolution: | Keywords:
> > --------------------------+----------------------
> > Changes (by gjm):
> >
> > * owner: nobody => gjm
> > * status: new => accepted
> >
> >
> > Comment:
> >
> > Search is an area that I do want to look into more generally. I am
> > thinking of looking into solr integration. There appear to be a few
> > plugins for this that might be worth checking out. I will see if I can
> do
> > a quick investigation so I will assign this to myself, at least for the
> > short term.
> >
> > --
> > Ticket URL: <https://issues.apache.org/bloodhound/ticket/260#comment:3>
> > Apache Bloodhound <https://issues.apache.org/bloodhound/>
> > The Apache Bloodhound (incubating) issue tracker
> >
>
--
Gary Martin
Lead Developer | WANdisco <http://www.wandisco.com/>
*
*
**
*Join us this October at Subversion Live
2012<http://www.wandisco.com/svn-live-2012>
*
Re: [Apache Bloodhound] #260: Search improvements
Posted by Olemis Lang <ol...@gmail.com>.
On 11/16/12, Andrej Golcov <an...@digiverse.si> wrote:
> On 16 November 2012 16:19, Apache Bloodhound <
> bloodhound-dev@incubator.apache.org> wrote:
>
>> #260: Search improvements
>> --------------------------+----------------------
>> Reporter: jdreimann | Owner: gjm
>> Type: enhancement | Status: accepted
>> Priority: major | Milestone:
>> Component: trac core | Version:
>> Resolution: | Keywords:
>> --------------------------+----------------------
>> Changes (by gjm):
>>
>> * owner: nobody => gjm
>> * status: new => accepted
>>
>>
>> Comment:
>>
>> Search is an area that I do want to look into more generally. I am
>> thinking of looking into solr integration. There appear to be a few
>> plugins for this that might be worth checking out. I will see if I can
>> do
>> a quick investigation so I will assign this to myself, at least for the
>> short term.
>>
>
> + 1 to improve search relevance. I'm also interested in helping on subject.
>
> IMHO, we should also consider Woosh [4] as possible solution as python-only
> library to bypass external dependency on java.
>
> [1] and [2] are interesting links on implementation of Trac Search plugin
> with Woosh.
>
> [1]
> https://groups.google.com/forum/?fromgroups=#!msg/trac-dev/sbU-g0C6kvk/1_juL29aAtQJ
> [2]http://trac.edgewall.org/wiki/AdvancedSearch
> [3]http://trac.edgewall.org/wiki/SearchRefactoring
> [4]http://pypi.python.org/pypi/Whoosh/
>
+1 for Whoosh ... AWESOME !
Seems to be exactly what we need .
--
Regards,
Olemis.
Blog ES: http://simelo-es.blogspot.com/
Blog EN: http://simelo-en.blogspot.com/
Featured article: