[OPEN-ILS-DEV] ***SPAM*** RE: ***SPAM*** a question/comment about relevance ranking

[Soulliere, Robert]

Hi Melissa,
We were able to adjust our relevancy weights. It does require editing or adding records to a database table directly.
The table which controls this is search.relevance_adjustment
The fields in this table are:
active - set to true or false
field - this is the field which is being adjusted. You will need to refer to the table config.metabib_field to link the adjustment to appropriate filed being searched and field_class or type of search. E.g. keyword searches.
bump_type - what aspect of the field you are you adjusting. Options include: first_word, full_match and word_order.
multiplier - how much weight you are giving to the bump_type. (might require mathematical consideration or experimentation to get this perfected)
You can edit fields or add new ones to search.relevance_adjustment.
This does require understanding of postgresql and knowledge of the Evergreen database structure.
You might want to check out:
http://www.postgresql.org/docs/ for documentation on Postgresql
and
http://open-ils.org/documentation/evergreen-schema-1.4.0.2.html for information on the database schema used in Evergreen (as of version 1.4.0.2)
You could  also look at a friendly GUI front end application for postgresql such as PGAdmin or Navicat. However, some postgresql configuration is required for those tools to access the database.
Regards,
Robert
Robert Soulliere
Systems Librarian
robert.soulliere at mohawkcollege.ca<mailto:robert.soulliere at mohawkcollege.ca>
Tel: 905-575-1212 X3936

Library at Mohawk
Connect * Learn * Succeed
brain.mohawkcollege.ca<http://brain.mohawkcollege.ca/>

From: open-ils-dev-bounces at list.georgialibraries.org [mailto:open-ils-dev-bounces at list.georgialibraries.org] On Behalf Of Daniluk, Judy
Sent: January 13, 2010 5:48 PM
To: Evergreen Development Discussion List
Subject: Re: [OPEN-ILS-DEV] ***SPAM*** a question/comment about relevance ranking

Here in North Texas we are on 1.4, but we had a similar problem.  A keyword search on "wildlife" would have many pages of results, but items where "Wildlife" was the complete one-word title would be buried many pages down instead of at the top.  We knew Evergreen could do better because a similar search on the Georgia PINES catalog would correctly bring the one-word titles to the top.

Our Evergreen support is handled by Equinox. Once the problem was brought to their attention, they were able to fix the problem.  I don't know how they did it - something about adjusting relevancy ranking weights.


Judy Daniluk
Technology Consultant, North Texas Regional Library System
6320 Southwest Blvd, Suite 101, Fort Worth, TX 76109
jdaniluk at ntrls.org<mailto:jdaniluk at ntrls.org>         817-201-6778 (cell)      817-377-4440 (office)        www.ntrls.org<http://www.ntrls.org/>


________________________________
From: open-ils-dev-bounces at list.georgialibraries.org on behalf of Melissa Belvadi
Sent: Wed 1/13/2010 11:39 AM
To: open-ils-dev at list.georgialibraries.org
Subject: [OPEN-ILS-DEV] ***SPAM*** a question/comment about relevance ranking

We are now on 1.6. I still don't understand the relevance ranking system, or perhaps disagree with it, or maybe we can tweak it locally?


Here's a search I just did:  keyword: food society


The very first hit, sorted by "relevance" was:

Free radicals and oxidative stress : environment, drugs and food additives / organized and edited by C. Rice-Evans, B. Halliwell and G.G. Lunt.

Somewhere in the "local notes" was a mention of the publication being affiliated with a professional society. The word society appears in the author and that note, but not in title or subject headings.


The next several were more or less the same kind as the first.


Way down at position 9 was this gem, which is exactly the kind of book  I had in mind:

The Cultural feast : an introduction to food and society / Carol A. Bryant ... [et al.].


As a librarian, I would expect that matches to my keywords appearing in the title to have the very highest weight, then (or possibly co-equal) subject headings, and much lower down in the formula would be 5xx fields and author fields.


Is this a philosophical problem or a technical one? Can we modify the ranking algorithm locally at the level of individual MARC tags or even index groupings (eg things in the title index, subject index, etc.)?


Thanks!


Melissa


---
Melissa Belvadi
Emerging Technologies & Metadata Librarian
University of Prince Edward Island
mbelvadi at upei.ca
902-566-0581


________________________________
This E-mail contains privileged and confidential information intended
only for the individual or entity named in the message. If the reader
of this message is not the intended recipient, or the agent responsible
to deliver it to the intended recipient, you are hereby notified that
any review, dissemination, distribution or copying of this communication
is prohibited. If this communication was received in error, please
notify the sender by reply E-mail immediately, and delete and destroy
the original message.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://libmail.georgialibraries.org/pipermail/open-ils-dev/attachments/20100114/eb4a6547/attachment-0001.htm