I'm not sure that I would agree that people producing the text should try to make the author name they are submitting "exactly the same" as what already appears in the PG catalog. There has never been an agreement (that I am aware of) on what the metadata in the PG header is supposed to be... An exact transcription of material on the title page? A "working title" (and author), used as the file is being processed? A LoC-style heading? All three of these are different things. In my experience having someone who is not used to the dealing with the catalog trying to make cataloging descions can only lead to more confusion. Andrew On Wed, 12 Apr 2006, Marcello Perathoner wrote:

1. You should direct your catalog questions to catalog@pglaf.org

2. Aforementioned phenomenon happens because the automagic catalog script just reads the ebook header to get at the author. If the spelling is different from the spelling already in the database a duplicate author entry will be created. To avoid this, just search the gutenberg catalog before posting your ebook and if you find your author make sure your spelling is *exactly* the same.

3. I will fix this right now and tomorrow new pages will be created.

On Wed, 12 Apr 2006, Marcello Perathoner wrote:

Andrew Sly wrote:

...

I'm not sure that I would agree that people producing the text should try to make the author name they are submitting "exactly the same" as what already appears in the PG catalog.

Why not? It seems the logical thing to do. If they don't, somebody on the catalog team will have to edit the author name later.

Well, if we're going to have any reasonable kind of catalog, every record has to be checked manually anyway.

...

In my experience having someone who is not used to the dealing with the catalog trying to make cataloging descions can only lead to more confusion.

Nobody asked to make decisions. Only to copy the author name exactly as in the catalog, of course after making sure that it is indeed the same author.

I'm afraid it's not quite that simple. Anyone who types out some author or title, which is then copied into the PG header, which is then automatically copied into the catalog, is making cataloging descisions whether they know it or not. I've seen a few cases where it seems that information could be getting garbled as it passes through the hands of different people involved in getting a text into PG. For an example from today, here is part of an email I sent just ten minutes ago: Ok, I'm a little confused here. As near as I can make out, this is an account by Henry Charles Mahoney of his time as a German prisoner-of-war. His papers were then edited into book form or, as the text says, "Chronicled by" Frederick Talbot. When this text was submitted, Frederick Arthur Ambrose Talbot was listed as creator, and Henry Charles Mahoney as Editor. Would there be any objection if I switched those around? Date: Sun, 9 Apr 2006 11:52:47 -0700 Subject: [posted] Posted (#18134, Talbot) ! To: posted@lists.pglaf.org Message-ID: <20060409185247.GA7225@pglaf.org> Content-Type: text/plain; charset=us-ascii Sixteen Months in Four German Prisons,by Frederick Arthur Ambrose Talbot 18134 [Editor: Henry Charles Mahoney] [Subtitle: Wesel, Sennelager, Klingelputz, Ruhleben] [Link: http://www.gutenberg.org/1/8/1/3/18134 ] [Files: 18134.txt; 18134-8.txt; 18134-h.htm; ] [Clearance: 20051019064205talbot]

Jeroen wrote:

I think it is good practice to list the author name in the catalog the same way it appears on the title page. Easy, straightforward, and a pain in the * for people who like normalized databases....

One approach is OEBPS' use of Dublin Core, where for each creator/contributor, both may be included: the name as it appears in the publication, and a normalized name using the "file-as" attribute. As an aside, the OEBPS approach of creator/contributor, and the use of the "role"/"file-as" attributes should be looked at for the PG catalog/database. See: http://www.openebook.org/oebps/oebps1.2/download/oeb12-xhtml.htm#sec2.2.6

We should still link to a record that carries details for the author, and actually tells us that the author of Book A is the same as that of Book B, even though the name is different.

Interesting. Of course, using a normalized name (along with the birth/death years) is an excellent way to remove almost all ambiguity and result in something that is eminently machine readable. I suppose one could use some established catalogs (like the LoC) for coming up with normalized names. Since the PG corpus is still relatively small (compared to most libraries), it would not take too long for a few librarians to once and for all normalize the author/contributor names for each Work. It's simply a matter of deciding to do so, and making sure the effort is not sabotaged later on. (Of course, MARC records from established catalogs may also be imported and used.) Jon Noring