Jump to navigation Jump to search

90061 OPEN Multiple attempts to change the name of one of the co-authors of this book from all uppercase to mixed case have been unsuccessful. The submission goes through the motions, but re-displaying the record returns the same all uppercase name. Presumably, the software does a lookup on existing entries in the Authors table in order to determine if the submitted Author name is new or not and the match is case-insensitive. Therefore as long as there is an uppercase version of the name in the database, it will be used for all new records. Ahasuerus 15:34, 5 Jun 2006 (CDT)

In this particular case, I would argue that it should work this way. Let's say that there is an uppercase author who has 20 records in the database. I don't think that updating a single title record should fork a lowercase version of the author for that particular title, potentially splitting the bibliography between Author J. Lowercase and AUTHOR J. LOWERCASE. The right thing to do is modify the author record for AUTHOR J. LOWERCASE, which will then fix the name for all 20 records, not just the one title. (We're going to have a similar conceptual problem with editing magazine contents - should modifying the title change the canonical title, or fork a variant title?) Alvonruff 18:22, 5 Jun 2006 (CDT)
After sleeping on it, I suspect that the underlying problem is that the current model provides for no data validation up front. Instead, we try to reconcile what the submitter entered (titles, author names, etc) with various types of existing data on the back end by guessing what the submitter meant. That's always tricky. Ideally, we would have a way to perform validation and disambiguation up front, presumably by giving the submitter all the options that we would normally have on the back end and having him/her select the right one(s). Something along the lines of "You have submitted a new title, Compleat Mcandrew, by Charles SHEFFIELD. There is a similar title, The Compleat McAndrew by Charles Sheffield, currently on file. The title/author names that you submitted differ in capitalization/punctuation/etc [we can add more choices as we beef up the logic up to and including a soundex check]. For the title, do you want to (a) change the existing title to the new title, (b) keep the title currently on file, or (c) create a variant title of the file on file? [this may look similar to the existing "keep/drop" tables]. For the author, do you want to (a)/(b)/(c) [similar to the choices above]."
Does this make sense? Is it implementable? Ahasuerus 08:41, 6 Jun 2006 (CDT)
I agree totally. Similar titles, similar authors, or authors names that are obviously mangled in some way. Error checking is too minimal at present. It may require a lot of user/server interaction to properly implement (AJAX), which means it probably won't show up tomorrow, but it is going to be necessary to minimize errors. Alvonruff 09:33, 6 Jun 2006 (CDT)