From: Andrew N. <and...@vi...> - 2007-11-09 17:45:28
|
> > Just a general question for you all. Right now from what I can tell, > > VUFind doesn't really use authority records at all, just process the > > MARC records and normalizes the author names. Are there any plans on > > changing this? I don't see anything on the road map. I actually > > have a couple of reasons for asking. > > > You are correct. VuFind indexes MARC records in a general purpose > fashion. Currently, there is no integration with respect to authority > records and I'm not aware of any plans to do so. I suppose it might > be possible to create a process of merging authority records with MARC > records and perhaps use this information to create an authority-based > author search. However, I couldn't answer that question (right now, > anyway). Someone else on this list might have a better grasp on how > this might be accomplished in our Voyager system. Yes, and I would say the main reason is due to the fact that we do not use = Authority records here at VU. I would love to get someone to volunteer to = improve VuFind to utilize authority control. Any volunteers? > > > > > > Right now I'm on a consortium committee for CARLI that's getting a > > chance to view some of the excellent work Chris Delis (Hi Chris) has > > done with VUFind. But in playing around it became clear that names > > are just being normalized and leading to weird situations. One of my > > favorite examples is Michael Jackson. There's actually several > > writers that are named Michael Jackson, including a children book or > > two by the actually famous Michael Jackson. Searching right now > lumps > > them all together as one person and always shows the Wikipedia > article > > for the "King of Pop". > > > > When a person actually decides to look for books by an author, I've > > notice they tend to actually like authority records. It's true > facets > > might be able to help here, but in the case of looking for authors > one > > very dominate author can "hide" works by other authors. Of course, I > > haven't gotten too many people's reactions on the current system, > > could be I'm making a mountain of a molehill. > > > > Generally speaking, I tend to think of searches in two categories: > general searches, where a user doesn't know exactly what he or she is > looking for, e.g., a topic or subject search; and targeted searches, > where a user knows more precisely what he or she is looking for, e.g., > a specific author or ISBN. It may very well be that VuFind is more > geared towards the former category. However, I don't find it lacking > too much in the latter, either. If I wanted to find materials by a > different Michael Jackson (not the pop star), how difficult is it to > do this? In our test environment of 1,000,000 bib records, an author > search for Michael Jackson yields 14 results. And you are correct to > mention how facets can assist in this scenario. If, say, I was > interested in Michael Jackson, the author in the field of Social > Science, I could simply narrow my search based on this subject > category (in this particular case, it will narrow my search results to > 1). So, although I think you bring about a valid point in this > discussion, I personally do not find it to be a "mountain" nor a > "molehill." Perhaps, it lies somewhere in between? Maybe I'm just > not understanding something (very possible :-). Or maybe a different > example will help me understand better? I agree that it is not a major problem, because the facet options do help w= ith this as you point out. But lets say there are 2 people in a similar fi= eld that publish with the same name. Chances of that happening im sure are= rare - but probably do happen. Also, if I were to search for Samuel Clemens - I should see (or it would be= nice) records where Mark Twain is listed as the author. > > Actually, in this example, I find that I-find (pun intended :-) > simplifies my search. Here are WebVoyage results for Michael Jackson, > the author: > > http://i-share.carli.illinois.edu/uc/cgi- > bin/Pwebrecon.cgi?Search_Arg=3DJackson%2C+Michael&Search_Code=3DNAME_&CNT= =3D2 > 0&HIST=3D1 > > If I wanted materials in the field of Social Science, which one do I > choose? Yes, I don't think this solution is any good either. Are you expecting a p= oor freshman to know the birth year of the author they are looking for :) > > > Up to now, I haven't addressed the Wikipedia ambiguities. I agree > that we should try to find a solution to this issue. Have you figured > out a way to reliably access Wikipedia's "disambiguation page"? If we > can do this, that might be a good solution (make the user pick). Hmmmmmm..... Lots of possiblities on improvement ... vufind needs more developers and mo= re adoption to get these issues solved. Join in the fight! Andrew |