[OPEN-ILS-GENERAL] Activity metric for relevance

[Rogan Hamby]

Something that took into consideration how current the circs/holds were
could be valuable I think in reflecting a more ... natural effect that
titles have in relevancy.

For example, Gone Girl by Gillian Flynn.  In the first 24 hours that the
bib was in our system the total holds might be relatively small compared to
some other titles during the last 6 months or a year.  Now, by the end of a
week the holds had skyrocketed and would show up well even in a search just
for "girl" under the proposed approach.

I think taking into consideration the timeliness of those weights would
give a more natural reflection of the title rising to the top over the
course of the week (and significantly higher that first day).




On Fri, Mar 15, 2013 at 9:01 AM, Thomas Berezansky <tsbere at mvlc.org> wrote:

> The current plan would not take into account how recent the circs (or
> holds) were, just that they were within a configurable time period of the
> time the cronjob that counts them last ran (default will likely be to
> include those from within the last 6 to 12 months). If you have an
> algorithm you think would work well and are willing to share we would
> gladly include that as an option when doing the work, though.
>
> We would not, however, be able to make it a per-bump option with the way
> we currently plan on storing the circ and hold counts, so instead it would
> function as an overall modifier to the circ/hold count numbers. Though even
> as I type this email I have thoughts on how we could change that if the
> feeling is that it should be at least partially bump-to-bump configurable.
>
> Thomas Berezansky
> Merrimack Valley Library Consortium
>
>
> Quoting Mike Rylander <mrylander at gmail.com>:
>
>  Kathy,
>>
>> Have you considered allowing an aging parameter for some bumps, so that
>> newer data toward the near end of the horizon is considered more
>> important?
>> For instance, spikes in circulation might have a larger short term effect
>> on relevance, but over time, while still being factored into relevance,
>> would be less important though still considered in the bump logic.  I ask
>> because I have a simple algorithm I'm using in another project, to be
>> debuted at the conference, that may be portable to this work.
>>
>> --miker
>>
>>
>>
>> On Thu, Mar 14, 2013 at 3:53 PM, Kathy Lussier <klussier at masslnc.org>
>> wrote:
>>
>>  Hi all,
>>>
>>> MassLNC is working with our partners at MVLC to develop an activity
>>> metric
>>> (aka popularity metric) that will allow sites to rank more popular items
>>> a
>>> little higher in search results than items that don't see as much
>>> activity.
>>> I've raised this idea on the list before. Although Evergreen allows sites
>>> to adjust relevancy based on the appearance of keywords in certain
>>> fields,
>>> which is highly useful, our hope is that this additional functionality
>>> will
>>> lead to further improvement when ranking results by relevance.
>>>
>>> As an example, if a user were conducting a keyword search on "abraham
>>> lincoln,"  there are many titles in most US libraries where the words
>>> "abraham lincoln" show up in the title. There would be no way to tease
>>> out
>>> the titles that are getting the most attention by readers. In fact, a
>>> title
>>> like "Team of Rivals" ranks very low in our search results even though
>>> there is a high likelihood it is the title the patron is seeking.  By
>>> applying a metric based on activity, we might be able to see those
>>> more-recently popular titles floating higher in the search results list.
>>>
>>> I would like to share MVLC's proposal outlining the details for
>>> implementing this project. The proposal is available at
>>> http://masslnc.cwmars.org/****node/2757<http://masslnc.cwmars.org/**node/2757>
>>> <http://masslnc.**cwmars.org/node/2757<http://masslnc.cwmars.org/node/2757>
>>> >.
>>>
>>> It provides a lot of flexibility in allowing sites to define what "high
>>> activity" means to them. Circulation activity, holds activity, total
>>> copies, and publication age/bib record age can all be used as an activity
>>> metric.
>>>
>>> If you have any feedback or questions, feel free to let us know.
>>>
>>> Kathy
>>>
>>> --
>>> Kathy Lussier
>>> Project Coordinator
>>> Massachusetts Library Network Cooperative
>>> (508) 343-0128
>>> klussier at masslnc.org
>>> Twitter: http://www.twitter.com/****kmlussier<http://www.twitter.com/**kmlussier>
>>> <http://www.twitter.**com/kmlussier <http://www.twitter.com/kmlussier>>
>>>
>>>
>>>
>>
>> --
>> Mike Rylander
>>  | Director of Research and Development
>>  | Equinox Software, Inc. / Your Library's Guide to Open Source
>>  | phone:  1-877-OPEN-ILS (673-6457)
>>  | email:  miker at esilibrary.com
>>  | web:  http://www.esilibrary.com
>>
>>
>
>


-- 

Rogan Hamby, MLS, CCNP, MIA
Managers Headquarters Library and Reference Services,
York County Library System

"You can never get a cup of tea large enough or a book long enough to suit
me."
-- C.S. Lewis <http://www.goodreads.com/author/show/1069006.C_S_Lewis>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://libmail.georgialibraries.org/pipermail/open-ils-general/attachments/20130315/8caa1b16/attachment.htm>