[Catalog-sig] Search engine relevance

Donald Stufft donald at stufft.io
Fri Mar 8 16:22:39 CET 2013


On Mar 8, 2013, at 9:51 AM, Jacob Kaplan-Moss <jacob at jacobian.org> wrote:

> Hi Antoine -
> 
> Yes, PyPI's search engine is rather simplistic, I think that's a
> pretty well-known problem.
> 
> For the time being you might try Crate instead (crate.io); I've found
> its search engine to be much much better.

Crate's search uses ElasticSearch whereas I believe PyPI is just using SQL against the DB.

That being said Crate's search could be a lot better still :/ But I'm not an expert on how to get the best search results.

> 
> Jacob
> 
> On Fri, Mar 8, 2013 at 8:00 AM, Antoine Pitrou <solipsis at pitrou.net> wrote:
>> 
>> Hello,
>> 
>> It seems the PyPI search engine is quite crude and doesn't try to make the
>> results relevant at all.
>> For example, if I'm trying to search "agi" in the hope of finding modules
>> relevant to the Asterisk Gateway Interface (nicknamed "AGI"), I get the
>> following results:
>> 
>> https://pypi.python.org/pypi?%3Aaction=search&term=agi&submit=search
>> 
>> As you can see, a large number of results pop up simply because they contain
>> the word "magic", which apparently is considered to match the "agi" request.
>> Clearly either the selection or the weighting algorithm isn't very efficient
>> here.
>> 
>> Regards
>> 
>> Antoine.
>> 
>> 
>> _______________________________________________
>> Catalog-SIG mailing list
>> Catalog-SIG at python.org
>> http://mail.python.org/mailman/listinfo/catalog-sig
> _______________________________________________
> Catalog-SIG mailing list
> Catalog-SIG at python.org
> http://mail.python.org/mailman/listinfo/catalog-sig


-----------------
Donald Stufft
PGP: 0x6E3CBCE93372DCFA // 7C6B 7C5D 5E2B 6356 A926 F04F 6E3C BCE9 3372 DCFA

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://mail.python.org/pipermail/catalog-sig/attachments/20130308/a4bf773f/attachment.pgp>


More information about the Catalog-SIG mailing list