User talk:Dissembler

Dissembler likes RPG?

I know you are a bot but maybe your master(s) will answer for you. I'm trying to understand why the following publications were selected for inclusion in ISFDB. What was the trigger for these? Actually, it might be cool to support RPG material as it seems to be popular enough.

• Roleplaying with Kids: Bringing up the Next Generation of Gamers at Amazon.com 0976931052. The author is not in ISFDB and it looks like it's part of the RPG world and not specfict.
• Action Scenes: A series of action vignettes and how to work them into your game at Amazon.com 0976931079 - an "about RPG" book.
• Hunter Sheets Issue One (SLA Industries, CB7505) at Amazon.com 0955542316 - looks like an RPG.
• Joan Bauer (Teen Reads: Student Companions to Young Adult Literature) at Amazon.com 0313335508 - this one was entirely mysterious to me because the author is not in ISFD and the only connection I could see is "# Subjects > Science Fiction & Fantasy > Science Fiction > History & Criticism."

Marc Kupper (talk) 13:08, 9 Jun 2007 (CDT)

Dissembler likes things marked "science-fiction" and some sources (mis)categorize those RPGs as SF. About 70% of Dissembler's consists of heuristics to remove RPG-related materials - the above 4 require new rules to catch them. Sorry for the inconvenience - I was called away during a Dissembler run, and then the site was down for a while, so I didn't get a chance to kill these myself. Alvonruff 06:52, 11 Jun 2007 (CDT)

Actually, tackling Dissembler myself was quite fun - feel free to leave a few for the rest of us! BLongley 16:10, 11 Jun 2007 (CDT)

Well, we're falling behind on the forthcoming books, and I'll probably do about 18 passes (sf, fant, and horror for the next six months), so you'll probably get your chance! Alvonruff 12:36, 12 Jun 2007 (CDT)

One thing I did notice was that Dissembler picked up two new Authors, Chad Corrie and David Coy - but each time only from the third book in a Trilogy. Any idea why the first two were missed? BLongley 12:42, 12 Jun 2007 (CDT)

He's only a four year old. I'm sure once he goes to preschool and learns to socialize with the other A.I. kids, he'll learn that even though there are usually 6 books in a trilogy (not including the prequel), they don't usually start with book 3. Alvonruff 21:10, 19 Jun 2007 (CDT)

Stop bragging about your percentage

You say you have entered about 95% of the data since 2003 but the humans are now entering substantially more data than you. As inept as they may be the work done by humans requires far less cleaning up than your entries. --swfritter 08:11, 16 Sep 2007 (CDT)

Unless the Wiki page mysteriously changed without leaving a trace in its history (and one never knows when dealing with Dissembler), the current claim is that it "entered about 95% of all the data entered in the ISFDB in 2003-2006". We should probably change it to "between 2003 and mid-2006", though. Ahasuerus 13:08, 16 Sep 2007 (CDT)

And add "those pesky humans are catching up, darnit!" for 2007 maybe? ;-) BLongley 16:39, 16 Sep 2007 (CDT)

The static "Major Contributors" page already states that it only covers "major contributors to the ISFDB until May 2006" and has a pointer to the dynamically generated page of Major Contributors Since May 2006 at the bottom of the page, so I think we are all set in that respect :) Ahasuerus 18:20, 16 Sep 2007 (CDT)

They also have an intuitive sense for correctness that you will ever achieve.--swfritter 08:11, 16 Sep 2007 (CDT)

Most bibliographic heuristics are relatively easily algorithmized, although it would result in hundreds, if not thousands of "if-then-else" statements. The biggest problem with Dissembler is that it relies on the Amazon.com data entry people for its data and they are notoriously unreliable. So in the end it all comes down to human incompetence :) Ahasuerus 13:08, 16 Sep 2007 (CDT)

Well, I for one don't want to retype Amazon's latest garbage here: it's much easier to correct once it IS here. Although it would be nice to have a quick way to spot such entries - just a note saying "Dissembler did this!" would be a start, although searching for what an editor has done (like we can with Wiki submissions) would be a help. I know I'd like to go check MY first edits again... BLongley 16:39, 16 Sep 2007 (CDT)

Dissembler generates submissions, but it can't approve them. In theory, the approving moderator should check Dissembler's data before approving it, but things do slip through the cracks sometimes. I think Al and I discussed adding a "This data is from Amazon.com, use at your own risk" note to Dissembler submissions a while back. Ahasuerus 18:20, 16 Sep 2007 (CDT)

I think Al may rush Dissembler approvals a bit (understandably) as they can create quite a backlog - but I for one am happy to go have a look (see above section). And how are we going to teach this 4-year-old AI if only the father gets to review his learning? BLongley 16:29, 17 Sep 2007 (CDT)

They way I used to track (and correct) Dissembler's submissions after the fact was by pulling up all books for a recent month. For example, http://www.isfdb.org/cgi-bin/fc.cgi?1+2007 will display all books published in January 2007, http://www.isfdb.org/cgi-bin/fc.cgi?2+2007 will display all books published in February 2007, etc. I even set up a project page for this type of cleanup. Unfortunately, I haven't had time to work on this project ever since we enabled outside submissions :( Ahasuerus 23:14, 18 Sep 2007 (CDT)