search

ualbertalib / discovery

Discovery is the University of Alberta Libraries' catalogue interface, built using Blacklight
http://search.library.ualberta.ca
12 stars 3 forks source link

Call number search working? #854

Closed fculpepper closed 8 years ago

fculpepper commented 8 years ago

When I search for KF 2925 S27 2011 I get 2 results. The first one is the right book, wrong year, the second is just the wrong book. The same search in NEOS brings up the correct record, so it's there, just not coming up in my search.

merisjames commented 8 years ago

I wonder if that has something to do with getting a "something went wrong" when trying to view the actual book you want... https://www.library.ualberta.ca/catalog/4844347 Something funky with the record?

fculpepper commented 8 years ago

Oh drat. No that was me trying to link to my search, I assumed the address bar would permalink to it. I just searched the call #, kf 2925 s27 2011, in the main search and got the two results.

On Fri, Jan 8, 2016 at 12:58 PM, merisjames notifications@github.com wrote:

I wonder if that has something to do with getting a "something went wrong" when trying to view the actual book you want... https://www.library.ualberta.ca/catalog/4844347 Something funky with the record?

— Reply to this email directly or view it on GitHub https://github.com/ualbertalib/discovery/issues/854#issuecomment-170107424 .

merisjames commented 8 years ago

No, not at all, your link was fine! I followed it, then looked at other books by the author. The dead link is the one for the 2011 book, which I assume is the one you were expected to see in your results?

fculpepper commented 8 years ago

Oh gotcha! Yep, we were looking for the 2011 edition. So not a call number search issue but a no-show item issue. Thanks!

On Fri, Jan 8, 2016 at 1:06 PM, merisjames notifications@github.com wrote:

No, not at all, your link was fine! I followed it, then looked at other books by the author. The dead link is the one for the 2011 book, which I assume is the one you were expected to see in your results?

— Reply to this email directly or view it on GitHub https://github.com/ualbertalib/discovery/issues/854#issuecomment-170109724 .

theLinkResolver commented 8 years ago

Re. the two results you do get, I wonder if these are coming back because most of the search terms (i.e. bits of the call number) match. (Which suggests, in this relevance-based system, that using quotes around call numbers might be a necessary step when looking up a known call number.)

However, the non-retrieval is concerning. The search should also retrieve this record - https://www.library.ualberta.ca/catalog/3973940 (again, matching on most bits of the call number). In this case, I'm wondering if only the first 090 (call number) field is part of the search index. (see also #846)

As to why https://www.library.ualberta.ca/catalog/4844347 doesn't load, I don't see anything on the cataloguing end. Hoping @redlibrarian can figure that one out.

theLinkResolver commented 8 years ago

for 3973940, the first-listed call number (090) field in the record is for C-FER, just to clarify what the heck I'm talking about in my previous comment.

As is the case in the Classic Catalogue, keyword searching for a call number is looking at the call numbers listed in the bib record, rather than the call numbers in the items. Hence the need to make sure all the 090 fields are indexed, as there can be quite a few among us and our NEOS pals.

ghost commented 8 years ago

@theLinkResolver Are the 090 call numbers correct? I thought one of the reasons we don't use them for display is that they could differ from what we actually have on the items. It should be possible to index all the 090s (Blacklight is probably assuming only 1 per record), but if it would be better to index the item call numbers, then we'll need to discuss that (as we currently don't index item level data).

theLinkResolver commented 8 years ago

@redlibrarian Good to clarify. The 090s do match the items. It's the 050s that don't, but are used for the call number facet.

ghost commented 8 years ago

@theLinkResolver OK, thanks - that's great. I should be able to index multiple 090s pretty easily.

fculpepper commented 8 years ago

Just to clarify, if the record does not have an 050 searching the call number won't retrieve it? A coworker just brought a non-retrieving call number to me (HC 79 E5 G373 2011) and I want to make sure I give her the right answer :)

ghost commented 8 years ago

@fculpepper That's currently true. When I've indexed the 090s, there should, I think, be better coverage. @theLinkResolver is that fair to say?

theLinkResolver commented 8 years ago

@redlibrarian @fculpepper That's correct.

ghost commented 8 years ago

This is done (see #846 ). Please test in next release.

theLinkResolver commented 8 years ago

Tested here and with issue #846. Working great.

@ualbertalib/discovery-testing @ualbertalib/serials-cataloguing An important takeaway here is that you will want to put quotes around the call number to increase precision (and to replicate the experience you are used to with the old system).

Then again, you may discover another edition or closely related work if you don't use the quotes, which might not be a bad thing (and the perfect match seems to always be the first hit in my tests).

@linds37 I see you have covered call number searches in the search guide. Looks good - and, in light of the above, you might consider adding a note about using quotes to find only that call number.

linds37 commented 8 years ago

thanks - done!!

On Wed, Apr 27, 2016 at 4:25 PM, Scott Davies notifications@github.com wrote:

Tested here and with issue #846 https://github.com/ualbertalib/discovery/issues/846. Working great.

@ualbertalib/discovery-testing https://github.com/orgs/ualbertalib/teams/discovery-testing @ualbertalib/serials-cataloguing https://github.com/orgs/ualbertalib/teams/serials-cataloguing An important takeaway here is that you will want to put quotes around the call number to increase precision (and to replicate the experience you are used to with the old system).

Then again, you may discover another edition or closely related work if you don't use the quotes, which might not be a bad thing (and the perfect match seems to always be the first hit in my tests).

@linds37 https://github.com/linds37 I see you have covered call number searches in the search guide. Looks good - and, in light of the above, you might consider adding a note about using quotes to find only that call number.

— You are receiving this because you were mentioned. Reply to this email directly or view it on GitHub https://github.com/ualbertalib/discovery/issues/854#issuecomment-215249987