Library Gazette

Vufind status update – October 2009

Tuesday, October 20, 2009 7:13 am

Vufind has been live for just about 2 months now. In that time we have gotten 118 feedback emails detailing bug reports, enhancement requests, and personal opinions about the new system. It has been a busy fall for the systems staff and we are just now finishing up the fixes on a few of our big Vufind issues. One of our biggest problems has been consistent record updates. I am glad to report that we are now running daily loads into Vufind. We will shortly be introducing daily deletes as well. Full index regens still take 20 hours and we have yet to figure out exactly how frequently we need to do that. The full index regen time depends more on our record export, modification, and transfer from our Voyager system at the moment (Vufind takes about 3 hours to index the 1.6M records).

Our other major issue has been system stability. I appreciate how patient everyone has been while we iron out the issues that cause Vufind to go down so frequently. We are still working on this but hopefully have allocated enough RAM to the server and enabled the system to ‘clean-up’ after itself so that Vufind can remain responsive even during moderate load (fingers crossed – we have not had any downtime since the last modifications a week ago – many thanks to Jeremy Kindy for helping us work through this!). An interesting thing that IS found recently was that Google was responsible for 1/3 of our vufind traffic (we have now blocked their robot) :).

The Vufind community as recently created a new administrative organization and is working towards fixing many of the bugs that we have listed. When the community releases the official 1.0 release we will upgrade! In the mean time we will continue to work on our end and contribute back to the community where it is valuable. The list of enhancement requests, bugs, and fixed issues below represent all of the feedback that we have gotten so far. They are broken down into three categories, unresolved enhancement requests, unresolved bug reports, and resolved enhancement requests/bug reports.

Enhancement requests

  • Would like to be able to see how many hold requests exist on an item in the new catalog
  • Would like the new catalog to explicitly state which series or version an item is (example Mi-5 season 1,2,3)
  • Add year into results listing
  • Add journal option to basic search
  • Add the ability to see 20, 40, 60 records per page
  • Improve serial current issues display – right now it shows item level detail but not summary holdings
  • Add ability to preserve certain facets (like library) when doing searching
  • Add grouping to locations (All physical reference locations for example)
  • Add the ability to click on call numbers for browsing
  • Reduce the number of clicks to get to information
  • Add data to the results screen including publisher information, dvd season info, pub place/date, etc
  • Implement Spell Check
  • Make subject headings work the same way that authors do – via listing at the top of the screen
  • Make subject hierarchy work more consistently – united states history is a good example
  • Add a new items feature to vufind, particularly by subject or call # range
  • Would like to be able to replicate all brief record info in vufind

Bug Reports (Partially resolved or Unresolved)

  • Location listing should be in alphabetical order, should be consolidated in certain cases (ref desk and reference for example) – still working on figuring this one out.
  • Advanced Searching does not work with more than 2 terms, truncation proves to be problematic, further advanced searching returns inconsistent or known to be incorrect results when compared to the old catalog. One suggestion would be to remove advanced search and have advanced search link to the old catalog. There has been alot of discussion about how appropriate this. ..any thoughts? please leave comments!
  • Item statuses in voyager not always reported as desired in Vufind (missing books showing up as lost), lost showing up as overdue. This is going to require some advanced item status processing in the Voyager driver and will take some time
  • Date sorting not working as desired
  • Recently received issues do not have a location? – We need some clarification on this
  • Endnote Export not working
  • “I hate vufind” – While a very real problem there is no specific bug fix for this. We may want to discuss re-introducing our “classic view” in a more prominent place to alleviate this issue
  • Vufind does not always return what I search for – We have lots of reports of this. Sometimes Vufind has the record but it is not on the initial screen. In some other cases the record is not in the system. There are a few things we are working on here, first daily data loads will address recent titles. Second, we have a list of 22K records that did not import that we need to troubleshoot. Finally – we may need to think about the default search algorithm.
  • Save to favorites, email functions do not have polished javascript/ajax interface, require scrolling, etc
  • Name authorities are not consistent, cary grant, shakespeare return different result counts from old catalog
  • ISBN searching does not work (looks like vufind is not parsing out the – during indexing and as such needs it for the search
  • Improve holds/recalls

Fixed Issues

  • Catalog slows down/crashes under ‘heavy’ use – Some lib100 classes of 15 people have seen some slow response times – We have worked with IS to try to resolve these issues. We have increased the amount of RAM allocated to the system, tuned SOLR settings, and searched the logs for memory leaks. Hopefully this has been resolved.
  • Call Number now shows at the top of every view of the record
  • Library links not always proxied appropriately – Kevin implemented a workaround for now
  • Ebooks now showing as available
  • Military Science added as location
  • Sometimes the 007 in items (item format) does not correspond to what the item actually is. These items should be reported when identified and will be fixed by cataloging
  • Known items not always showing up – We have a number of specific reports here. In some cases this is due to a lag in indexing (still working on getting the connection between our two servers opened up) but in others the items were kicked out due to record errors.
  • Call number searching should not include periods – makes it difficult – resolved
  • Resources without Item records in catalog show incorrect status of Checked Out – We have a workaround for this but it requires addressing each location specifically in the code. If you still see errors please send them to me
  • Collections not synchronized, items in old catalog not in new – daily updating is in place, working on daily delete. It currently takes 20 hours to re-index our catalog from scratch
  • Wake Forest University facet limit does not return records (It is in essence a useless facet since everything in the db has this tag) – item removed from list.

8 Responses to “Vufind status update – October 2009”

  1. Erik,
    Thank you for keeping us up to date.

  2. Thanks for the update Erik and all the hard work the Systems team has done so far!!

  3. “Location listing should be in alphabetical order” — I disagree with that. A search I just entered returned 28 different locations. In the “less” view, the five that would show would be: the blank location(!), Browsing, Current Periodicals, Documents CD-ROM, and Documents Website.
    Given these results, I like the current display that prioritizes by number of hits. However, one possible enhancement would be to give users the option to show the facets from A to Z.

  4. Erik, Thanks for such a detailed update; it’s helpful to see what other people are encountering with Vufind as well as what you have been able to accomplish in addressing myriad issues!

  5. Thank you for this comprehensive update and kudos for the daily loads achievement. Whoo hoo!

    Resource Services is now setting up cataloging via batch record loading for EBL e-books and have learned that our records will have 245|h computer resource, which won’t show up under the e-book facet (which would be|h computer file). I’m wondering if this is an issue or a non-issue? I wasn’t sure which Wiki page to use for more VuFind feedback at this point; could you let us know where you’d like that recorded now?

  6. Many thanks to Till Kinstler from the Vufind community who took the time to follow-up with me after coming across this blog posting!

  7. Stanford has switched to Blacklight, but a few comments:

    1. call number browsing: we contributed code to the solrmarc project that can create shelfkeys and reverse shelfkeys for call number browsing. We are probably a month away from exposing our version of this, but the indexing code exists, and I can share the Solr queries I use to make it work.

    2. seeing more records per page – that’s pretty simple — have a pulldown for selection and use it as rows=X in the Solr request.

    3. hierarchical facets – already fakeable; Solr 1.5 (?) is coming with a way to do this.

    4. spell check based on the index — Solr can provide suggestions if you use the nightly build.

    5. hierarchy of subjects – that was kind of a mess in our version of VuFind. We fixed it in blacklight. It has to do with how the code looks for subfields without matching them to the particular field occurrences.

    6. “location listing” – not sure if this is from the record view, but if so, you can index a field with both the correctly sorting version of the location (then separator) and the display version. Sort based on the sort text (parsed out), but display based on the display text (matched to each sort key)

    7. Advanced search — Stanford is actively working on a solution to this. We have solved the problem of multiple search terms. You can go to and look at what we have (will be updated to reflect our latest work by 10/22, if not sooner). I will be posting information to the Solr wikis about how to fake boolean with dismax queries, and how to use nested queries to allow different search boxes, e.g. “author”, “title”, etc. I can share the spec document I wrote for our UI code guy if you would like.

    8. Endnote export is working for us; we may have applied a fix locally to VuFind back when we were using it. I’ll bet if you ask the VuFind community, someone else has a fix … that possibly didn’t make it back into the trunk (cough cough).

    9. Search Results: For Solr dismax, I figured out that you can have unstemmed and stemmed versions of a field and weight the unstemmed field matches higher. You can probably do the same with Solr lucene request handler, but you’d have to experiment. This has improved our relevancy greatly.

    10. ISBN searching – I know I figured out how to get Solr queries working with *and* without the hyphen, but I don’t remember how right this minute. Ask me if you like.

  8. Erik, thanks for this great review!

    I’d like to second a couple of items:

    Request: “Make subjects work like authors — list at top of screen.” If this would result in the kind of hierarchical LCSH display that you get with Voyager’s “Basic Search,” then that would fix the one major worry I’ve had (and a common complaint in the literature, I’ve noticed) about faceted systems — you lose the relationships that LCSH make clear, and that keyword searching can’t do as effectively. And it’s just the kind of thing that is most helpful to students and other novice users. Whoever thought of the author-list feature as applicable to subjects, that was a neat idea! I hope it’s doable!

    BYW, taking subjects one step further: any way to integrate LCSH’s “see” references into VuFind? (I haven’t seen any evidence in the faceting that this has been done.) For instance, in Voyager, if you search the subject “Music” or “Drama,” and click on the “ref” button, you get a really valuable list of all the genres of that artform — again, something most helpful to the novice searcher. Anybody in the community working on a way to integrate those into either a top-of-screen listing or the facets?

    Request: “Make subject hierarchy work more consistently, e.g., United States — History.” I recall a LIB100 instructor remarking that if you click on the beginning of the subject string (“United States”), VuFind broadens its search to everything about the U.S. in general; you have to click on “History” in the subject string to get things that are actually on U.S. history. As for myself, I’ve been vacillating between reactions of “Horreurs!” and “Maybe this is actually an improvement.” With a little user ed, it in fact enables a user to “browse” a subject string and broaden or narrow it to the precise degree that they need — instead of the previous “take it or leave it” option, which can be inflexible when the subject string is long, i.e., very specific. Maybe this is what the VuFind folks were envisioning.

    Advanced/Boolean/Multi-term searching: I was glad to see in another comment that Stanford is working on a fix. Like you, I’m on the fence about linking to Voyager for the advanced search — it would work for now, but future generations of users won’t be familiar with the “old” Voyager, and just dumping them into a whole new interface doesn’t seem like the right way to handle it. Maybe an intermediate solution would be a page behind the link explaining the “added” features of Voyager: not only Boolean searching, but hierarchical LSCH displays, etc. I hope we can investigate the Stanford fix.

    Thanks again for all your hard work!

August 2014
July 2014
June 2014
May 2014
April 2014
March 2014
February 2014
January 2014
December 2013
November 2013
October 2013
September 2013
August 2013
July 2013
June 2013
May 2013
April 2013
March 2013
February 2013
January 2013
December 2012
November 2012
October 2012
September 2012
August 2012
July 2012
June 2012
May 2012
April 2012
March 2012
February 2012
January 2012
December 2011
November 2011
October 2011
September 2011
August 2011
July 2011
June 2011
May 2011
April 2011
March 2011
February 2011
January 2011
December 2010
November 2010
October 2010
September 2010
August 2010
July 2010
June 2010
May 2010
April 2010
March 2010
February 2010
January 2010
December 2009
November 2009
October 2009
September 2009
August 2009
July 2009
June 2009
May 2009
April 2009
March 2009
February 2009
January 2009
December 2008
November 2008
October 2008
September 2008
August 2008
July 2008
June 2008
May 2008
April 2008
March 2008
February 2008
January 2008
December 2007
November 2007
October 2007
September 2007
August 2007
July 2007
May 2007
April 2007
March 2007
February 2007
January 2007
December 2006
October 2006
September 2006
August 2006
May 2006
April 2006
February 2006
January 2006
December 2005
October 2005
August 2005
July 2005

Powered by, protected by Akismet. Blog with