Ticket #1124 (closed defect: fixed)

Opened 13 months ago

Last modified 12 months ago

Author order incorrect in search results

Reported by: rcave Owned by: ronald
Priority: high Milestone: 0.9.2
Component: ambra Version: 0.9.1_rc1
Keywords: search Cc:

Description (last modified by ronald) (diff)

I put this (simple) search into the PLoS Genetics website (I searched for: 'vitamin D'):  http://www.plosgenetics.org/search/simpleSearch.action?query=vitamin+d&x=0&y=0

The authorship order is wrong for all of the articles found.

E.g., according to search, we have:
Expression of the Multiple Sclerosis-Associated MHC Class II Allele HLA-DRB1*1501 Is Regulated by Vitamin D Gabriele C. DeLuca?, Lahiru Handunnetthi, Sreeram V. Ramagopalan, George C. Ebers, David A. Dyment, Sarah-Michelle Orton, Blanca M. Herrera, Matthew R. Lincoln, A. Dessa Sadovnick, Narelle J. Maugeri, Michael J. Chao, Julian C. Knight

Clicking through to the article reveals the correct order:
Expression of the Multiple Sclerosis-Associated MHC Class II Allele HLA-DRB1*1501 Is Regulated by Vitamin D Sreeram V. Ramagopalan, Narelle J. Maugeri, Lahiru Handunnetthi, Matthew R. Lincoln, Sarah-Michelle Orton, David A. Dyment, Gabriele C. DeLuca?, Blanca M. Herrera, Michael J. Chao, A. Dessa Sadovnick, George C. Ebers, Julian C. Knight

So you can see that the authorship order in search doesn't match the order in the article.

This might be related to #661 (?)

Change History

Changed 13 months ago by ronald

  • description modified (diff)

Changed 13 months ago by ronald

This is due to the fact that the creators in DublinCore is a Set rather than a List. Fixing this would require a data migration, as the creators would need be stored as an Rdf:Seq (or Rdf:List) rather than as multiple statements with the same predicate.

Another option would be to pull the list of authors from article.dublinCore.bibliographicCitation.authors.realName - however, that may have a noticeable performance impact as this will force the loading all the author's profiles.

This problem was introduced in 0.9.1 because we're now using the Article object to get our meta information, rather than pulling it out of the lucene article document which had this information duplicated.

Changed 13 months ago by rcave

  • milestone changed from 0.9.1 RC2 to 0.9.2

Since the fix requires a data migration, moving into 0.9.2

Changed 12 months ago by ronald

  • status changed from new to closed
  • resolution set to fixed

(In [7456]) Display authors in search results in the order they appear in the article.

DublinCore?.creators is currently unordered; while it would be possible to change it to be ordered, that would not come for free in terms of query performance. Using the already ordered author list from the citation is a bit slower, but not hugely so: it slows down retrieval of the article info for a search hit by 10% (i.e. a page full of results now takes 10s instead of 9s).

This closes #1124.

Note: See TracTickets for help on using tickets.