ISFDB talk:Serial Cleanup

Revision as of 16:11, 2 September 2009 by Ahasuerus (talk | contribs) (→‎Missing serial variants: What the change will do)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Process description, please

It would be nice if this page explained exactly what an editor had to do to Fix a particular title. -DES Talk 14:55, 9 August 2009 (UTC)

Good point, let me give it a shot... Ahasuerus 15:12, 9 August 2009 (UTC)
An explanation has been added -- hope it makes sense! Ahasuerus 15:29, 9 August 2009 (UTC)
Thought I'd try to fix some of these, but sorry, the explanation didn't help me. Perhaps a few pointers for the denser among us. I was able to do some of the more obvious errors in the malformed titles list. MHHutchins 07:14, 28 August 2009 (UTC)
What I've been doing is this.
  • I look at the title link in the left hand set of columns. I then follow the links to the stated author and canonical author pages.
  • If, on the canonical author page, there is a non-serial version of the work, i make the serial a variant of that work, delete the placeholder serial title linked on the right-hand side of the table (which will now have no pubs), and mark the entry "Fixed".
  • If, on the stated author page, but not on the canonical author page, there is a non-serial version of the work, I create a placeholder non-serial version under the canonical author name. Then I make the actual publications, both serial and non-serial, variants of this. Then I delete the placeholder serial title linked on the right-hand side of the table, and mark the entry "Fixed".
  • If there is no non-serial version of the work on either author page, I mark the entry "OK".
I hope that procedure is sufficiently well described to be helpful. -DES Talk 14:49, 28 August 2009 (UTC)
Thanks for the step-by-step. I'll try to work on some. MHHutchins 16:19, 28 August 2009 (UTC)

Malformed serial titles

Have these titles appeared on this list simply because the "p" in "(Part x of y)" or the "c" and "n" in "(Complete Novel)" isn't capitalized? Will capitalizing be enough to remove the "error" in those cases? Same thing with the use of brackets instead of parentheses? Also, I recall creating the record for In Polar Seas (and its pseudonymous variant) just as a placeholder for the title (it appeared on a list of important lost race novels). Without creating 19 pub records (!) for a serial in a non-genre magazine, what is the best way to handle situations like this? Thanks. MHHutchins 04:50, 28 August 2009 (UTC)

In theory, we will want to create a Publication record for each magazine issue so that we could verify them individually and enter other Publication-specific information when available. Unfortunately, we don't always have a whole lot of information to go on, especially when dealing with 19th century magazines. Much of the time all we know is what secondary sources tell us, which is often limited to something like "12 installments between January and March 1884", although things have been slowly improving with the rise of the Fictionmags project.
There are also occasional cases where a story was originally published pseudonymously and then re-serialized under the author's real name, but no book publication ever appeared. We don't have a good way of handling these case short of creating placeholder Novel Titles, but at least they are (thankfully) rare. Ahasuerus 16:39, 28 August 2009 (UTC)
What would happen if we implemented the software change and the title records in the Malformed serial titles list had not been changed? Or would there be a problem to temporarily change the serials in non-genres magazines to NOVEL type? Or if we temporarily change those Grantville Gazettes pieces to SHORTFICTION type? MHHutchins 15:54, 2 September 2009 (UTC)
For example "Under the Moons of Mars" as by "Norman Bean" had been fixed. Now it is listed as being "malformed". Is the only problem that the "p" in "part" is not "P"? -DES Talk 14:38, 28 August 2009 (UTC)
The script that generated the current list of malformed Serial titles was looking for any titles that didn't contain "(Complete Novel)" or "(Part X of Y)" where X and Y were 1-2 digits (or punctuation, to account for "Part 1 of ?" and such.) The lower case versions of the same strings, i.e. "(complete novel)" and "(part X of Y)", didn't pass the filter. Ahasuerus 16:39, 28 August 2009 (UTC)
Is this really such a vital matter? Moving the goal posts like this is unfortunate. -DES Talk 14:38, 28 August 2009 (UTC)
Help (Help:Screen:NewPub) advises to use the capitalized version, so no goal posts were moved or otherwise molested. Thankfully, there are only a few dozen titles that will need to be fixed out of the 3,000+ Serial Titles that we have. Ahasuerus 16:39, 28 August 2009 (UTC)
Not to bring an R&S discussion here (please move this comment if I should have posted this on R&S), but are all of these titles "malformed"? Must serial titles be worded as one of "Part n of m" or "Complete Novel"? It's not SF, but to capture the first printing of a Poe work, a multi-scene play, I just entered the two-part magazine publication, where the Dec publication was three of the scenes (IV, VI, VII), and the subsequent Jan publication was scenes at either end (III, IX). The rest of the scenes were not published as part of this "serialization". I used the titles Scenes from an Unpublished Drama (Scenes IV, VI, and VII) and Scenes from an Unpublished Drama (Scenes III and IX). I suppose I could have used Scenes from an Unpublished Drama (Scenes III - IX) (Part 1 of 2) and Scenes from an Unpublished Drama (Scenes III - IX) (Part 2 of 2) -- or simply included "(Part n of 2)" on the titles I did use, but that seems a bit contorted. Still, I will go clean them up if that's the way they ought to be. --MartyD 10:39, 29 August 2009 (UTC)
The primary reason to use "(Complete Novel") and "(Part X or Y)" was to make the notorious "lexical match" logic work. The secondary reason was to ensure consistent presentation of Serial data. Now that we are moving away from lexical matching, the primary reason is no longer operative and we may want to discuss whether the remaining reason, consistency, should trump individual circumstances (and there are a number of unusual cases like the one described above.)
However, as you said, that's fodder for the Rules page. The purpose of the "Malformed" section of the Serial Cleanup page was to identify all Serial records that do not follow the current standard and therefore will not be found by the automated "VT all Serials!" conversion script that I am working on. The "Malformed" list can be used to clean up obvious errors/misspellings and set up VTs where applicable so that they are properly associated with their parent Novel/Shortfiction Titles. Ahasuerus 14:45, 29 August 2009 (UTC)

Serials in Grantville Gazette

Many of the malformed serial titles are from the Grantville Gazettes, of which I readily admit I know nothing about. It's strange that they're typed as serials as none of them have parts. If they are just a continuing series (which is not a serial) shouldn't they just be typed as shortfiction and placed into a subseries of "1632"? Look at the mess it's caused this author's page. This may have been discussed before when the Gazettes were being entered, and I may have skipped the discussion because it involved ebooks which are outside my area of knowledge. (And I'm not going to debate the difference between a serial and a series. We all should know that by now.) MHHutchins 06:57, 28 August 2009 (UTC)

I believe Kevin worked on the Grantville Gazettes early on, so we may want to ask him about it. Ahasuerus 16:41, 28 August 2009 (UTC)
I entered some of these, and own copies of most of them. I will try to respond more fully to this issue over this weekend. -DES Talk

Serials Matching VTs -- what to do

I would like a clearer explanation of what sort of fix the entries in this section need, please. -DES Talk 20:52, 31 August 2009 (UTC)

Sorry, I keep forgetting that the relationships between different Titles may not be as clear as they appear to me after writing the scripts that create these lists :) I have expanded the introductory section at the top of the main Project page; hopefully it will help. Ahasuerus 21:22, 31 August 2009 (UTC)
It helps me, thank you. -DES Talk 21:30, 31 August 2009 (UTC)

When a title is a serial variant that was later published with the same title

What do we do about a title which was used in the serial, but not the first book publication, and was later reprinted as a book with the serial title? For instance The Domains of Koryphon by Vance. The title has already been set up as a variant of The Gray Prince. Should we then make "The Domains of Koryphon (Part 1 of 2)" and "The Domains of Koryphon (Part 2 of 2)" variants of The Gray Prince because that's been established as the parent title? Thanks. MHHutchins 03:22, 1 September 2009 (UTC)

That's right, we will want to make this Serial into a VT of the canonical Novel Title, i.e. The Gray Prince. Ahasuerus 14:44, 1 September 2009 (UTC)

Are these really OK?

I've checked a number of the ones that were marked OK on the archived page, and I find many of them aren't in the same shape as the ones that I did. For example, Anita Allen[1]'s "The Spark of Allah" (sorry I couldn't create a link to her page because of those brackets in her name, and if anyone can, please let me know how). In the case of a serial that never had book publication I created a title record of just the name (no parts), and then made all of the serial parts variants of that single title record. Was this incorrect or unnecessary because the software change would do it for us? Thanks. MHHutchins 04:03, 2 September 2009 (UTC)

I couldn't get it to work through the {{A}} template, but in a plain link you can encode the braces:
[ Anita Allen<nowiki>[1]</nowiki>]
to get Anita Allen[1] -- "[" is %5B and "]" is %5D. The web server will interpret the encoded characters the same way as the plain characters. --MartyD 11:12, 2 September 2009 (UTC)
When there are special characters in an author's name, you need to use the "altname" parameter with {{A}}. If the characters are brackets they need to be encoded also. For example {{A|altName=Anita_Allen%5B1%5D|name=Anita Allen}} gives Anita Allen. -DES Talk 14:23, 2 September 2009 (UTC)
Here's a better example (one I can link to): look at "Crack of Doom" under Serials on the John Brunner summary page. It just doesn't look right. Then look at how I handled "Black World" on the Raymond A. Palmer page. MHHutchins 04:16, 2 September 2009 (UTC)
Another example: "Minotaur in a Mushroom Maze" by Richard and Nancy Carrigan. Only the second part of the serial was on the worklist and someone marked it OK, but it doesn't appear to be. MHHutchins 04:44, 2 September 2009 (UTC)
My understanding was that if there was no book publication, there was no need to create a placeholder record for the work as a whole, and the existing serial variants were fine. I marked all such as OK. The "instructions" say "In some cases, these Serials were published pseudonymously and have not appeared in book form. These Serial records are currently VT'd to the Serial's author's (or authors') canonical name(s) and should be kept 'as is'." I followed that. I marked "Minotaur in a Mushroom Maze" OK, as as I understand things, it IS ok. There is no book publication under the canonical author, so a serial is used to place each installment under the canonical author. It appears that the various installments were published with different author credits (unless #2 was incorrectly entered here). The 2nd part is entered in Analog June 76 as by "Richard Carrigan" (without the A.) The other parts are entered as by "Richard A. Carrigan". Unless this is a data entry error, or I have misunderstood the purpose of the "Serial Clenaup" I see no reason for any correction in this case. -DES Talk 14:23, 2 September 2009 (UTC) (The odd thing is that I once owned a copy of this issue -- I remember the cover and the lead story -- But I don't think I still have it. -DES Talk 14:28, 2 September 2009 (UTC))
That's right, the current plan is not to create placeholder Novel Titles if a serial hasn't appeared in book form. There are arguments for and against placeholder Novel Titles for Serials, especially in complicated circumstances involving pseudonyms, but I figured that we had enough on our plate as it was to open yet another can of worms at this time :) In addition, determining whether a Serial has been reprinted as a book under a different title is not always straightforward (I set up a couple last night after reviewing the Archive page), but that's always been the case. Ahasuerus 19:56, 2 September 2009 (UTC)

Missing serial variants

I was looking at Poul Anderson's page and noticed the serial "A Plague of Masters" which already had a variant under Earthman, Come Home! but was not a variant itself of that parent record. It wasn't on any of the lists generated for this project. Seems like it should have been. Or will the software change find these and make variants of them to a parent record? MHHutchins 04:40, 2 September 2009 (UTC)

In this case we have two Serial records which are set up as Variant Titles of a non-VT Novel record, a combination that our Serial cleanup scripts didn't search for. The scripts were interested in (inquisitive little critters that they are):
  • Serials VT'd to other Serials, and
  • Serials VT'd to Novels or Shortfiction records which were, in their turn, VT's of other records
Assuming that these Anderson records haven't changed over the last month, they were already OK when we started. Ahasuerus 19:42, 2 September 2009 (UTC)
So the software change would have had no effect on this particular serial, because its title ("A Plague of Masters") didn't match the parent record (Earthman, Come Home!) and it only matched a variant (the novel appeared as "A Plague of Masters" in an omnibus.) Right? If that's correct then these would have had to have been matched manually regardless of the software, just like some serials with non-matching names that have never been made variants to book titles, like another one I found while doing these serial fixes (Simak's serial "Time Quarry" to his novel Time and Again). MHHutchins 19:55, 2 September 2009 (UTC)
The software change tentatively planned for this weekend will do two things:
  1. Find all Serial Titles that are:
    • not VTs, and
    • match one (and only one) Novel or Shortfiction Title by the same author(s)
  2. and create a VT relationship between these Serial records and the matching Novel/Shortfition record
  3. Change the behavior of the Summary Bibliography and Title Page logic so that it no longer looks for matching Serial records (aka "lexical match")
In the case outlined above, the two Serial records are already VT'd to the canonical Novel Title, so the software change will not affect them. Assuming they were already VT'd that way when we started, of course. Ahasuerus 20:11, 2 September 2009 (UTC)

serials in non-genre magazines

The "Malformed" section currently has a list of "The following need to have pub records created for serials in non-genre magazines." I think there may be a problem of data in doing that.

For example i just checked out In Polar Seas. The title note says "First published as a 19-part serial in The Nelson Lee Library, October 16, 1915 - February 19, 1916." Galactic Central has a listing for the Nelson Lee Library, and OCLC has 5 records for it (223170652, 221562899, 220351018, 317998933, & 317998931). These agree that it was published in the period given, with weekly frequency. But neither gives specific issue dates or contents. By [Help:Entering non-genre magazines#Required data|our current standards]] I don't think that is enough to create db entries for these issues. ("If the date is not reliably known with sufficient precision to identify a specific issue, do not create an entry for the magazine."). While I could find a calender for 1915-16, and assume that the day of the week was always the same as one of the few specified dates of issue, and further assume that the serial appeared in every issue in sequence without fail, I am not willing to make an entry with that high a ratio of assumption to data.

As I was about to sign the above, i did some further checking. I found this list of issues and this list of issues with contents. I will be entering these after all. However, my general point above stands, with the caveat that one shouldn't give up to quickly on finding the required data. -DES Talk 16:22, 2 September 2009 (UTC)
True, there have been cases when we had to correct our data because we later discovered that there was a skipped/combined issue or the magazine changed frequency (or even its title) half way through. When we know the start and the end dates, the safe thing to do would be to create the "(Part 1 of 19)" and the "(Part 19 of 19)" pubs and leave a note at the Title level. If the number of "in-between" issues is low, say 1-2, and the missing dates can be guessed with a very high probability of success, typically when dealing with major magazines/newspapers (like Time) which appear(ed) very consistently, it's probably safe to guess as long as we record the guess in Notes, but guessing the dates and titles of 17 issues is probably too risky. Ahasuerus 19:30, 2 September 2009 (UTC)