Re-Model the Model

There’s a scene in the Adventure Time episode Vault of Bones (season 5, episode 12) that really resonated with me. Finn and Flame Princess are exploring an underground dungeon and early on they encounter a locked door. Much later on, they find a key.

The following exchange occurs:


Flame Princess saying "We have to go back"
Flame Princess: We have to go back?


Finn responding, "We don't have to, we *get* to!"
Finn: We don’t have to, we get to!


Now while the episode serves an important lesson for Finn (and the audience) about the importance of listening to your partner about their likes/dislikes and not trying to control every aspect of a shared activity — taken out of that context, this exchange speaks to me in a different way.

Because I get it.

As any scientist will tell you, nothing is more exciting that finding data that doesn’t fit the your expectations or model you’ve been working from. It means that you might have an opportunity to revise the model and may even gain a better understanding of how that which you’re modeling works!

Now I’m no scientist (library science degree notwithstanding), but I like to think that I have some of that science mojo in me, that desire (and belief) that given sufficient experimentation and brains, we can learn anything about anything.

I was working on an indexing project for my final semester at Simmons and about half-way through I thought of something that would radically change the entire project. So I started over from scratch, and literally quoted the above exchange myself. We get to go back, and learn something new, find something cool, and go farther than we could if we didn’t go back.


So what does all this jazz have to do with anything? The Working Group on Aggregates’ discussion paper came out!

This now gives me an opportunity to revisit my conceptual model of how comics fit into the FRBR structure, and revisit I did! It means I get to go back a little bit, but in the end it’s actually going to save me some time.

I’ve decided to use Option C from the working group’s discussion paper, and I’ll briefly explain why:

Option A: a comprehensive description of the Aggregation Work and Expression, and the aggregating Manifestation(s).

What this would mean for a comic is a detailed description of the issue (the aggregating manifestation) but not the stories contained within. That doesn’t work for me, as I specifically want to describe the stories and give access to them.

Option B: an analytical description of one or more of the distinct Works and Expressions, and the aggregating Manifestation(s):

At first blush this seemed pretty good — I could describe the stories with as much detail as I wanted, and then describe the issue in detail. Unfortunately, because this option leaves off the aggregate (the issue) at the work and expression level, I wouldn’t be able to provide access to the editor(s) of an issue which are associated with the expression level. No good.

Option C: A hierarchical description of the Aggregation Work and Expression and the aggregating Manifestation(s), and one or more of the distinct Works and Expressions, incorporated in the aggregating Expression.

At last we’ve arrived! Option C allows me to describe the issues and  the stories at the work and expression levels.

So what’s the difference? How am I revising things and getting to go back?Previously, I was describing individual stories to the manifestation level and then using a contained-in/container-of relationship to the manifestation of the issue. This meant duplicating a lot of content because the publishers, editors, etc. were the same for that manifestation of the story as the manifestation of the issue.

I won’t be doing that anymore! Exciting times. I look forward to revising and rethinking my modeling more as the paper gets discussed and evaluated by CC:DA and friends.


Addendum to Fictitious Characters

Somehow that post really got away from me got a little bloaty. There was a significant reason I’ve been thinking about fictitious characters so much lately, and of course it’s my RIMMF-comics project. But I completely forgot to talk about how it related. This is my attempt to rectify that.


As part of this project I’ve been *gasp* reading the literature. I know, what a total professional, right? In William Fee’s “Where Is the Justice… League?: Graphic Novel Cataloging and Classification”, he mentions how Library of Pennsylvania adds local names for the characters appearing in comics and I got real excited.

Probably the most useful for a comic book cataloger is the $q, or fuller form of the name.

A 600 field allows a patron to perform a more targeted search for a particular version of a character. For example, 600 1 7 $a Blue Beetle $c III $q (Kord, Ted) $g (Fictitious character) would be the third, or Charlton/DC, version of the character (following Dan Garret and Dan Garett). For searching and display, this is much more targeted and more likely to satisfy more patrons than the following example. 650 1 0 $a Blue Beetle (Fictitious character)

600 0 7 $a Deathbird $q (Cal’syee Neramani) $g (Fictitious character) $2 local
600 0 7 $a Gladiator $q (Kallark) $g (Fictitious character) $2 local

While I quibble with some of the coding choices on display here (see below* for all my quibbles)

don’t quibble with the idea! As I stated before in both my post on fic characters and in my post on comics-characters-differentiating specifically — I think it’s great to separate out the different identities of the people behind the mask. So it’s really exciting to see someone putting this into practice (and if you browse the catalog of the State Library of Pennsylvania, you’ll see a LOT more!




  1. This is a pretty off-brand use of subfield q. It’s intended to hold “fuller form of name” true, but what is being stored there is clearly a variant name (or an alternate identity, but that’s a different discussion).  The old instructions in AACR2 (22.18), and the newer instructions in RDA (9.5) are very clear on what a “fuller form of name” is — the spelling out of an initial, abbreviation, or otherwise shortened piece of the preferred name.
  2. If using a 600 field, first indicator would mean that the first piece of data is a surname. “Blue” is not Blue Beetle’s surname.
  3. Ted Kord is the second Blue Beetle, not the third: Dan Garrett, Ted Kord, and Jaime Reyes.
  4. 650 with a subfield 1 is valid….but very very unusual, I assume that’s a copy-paste error.
  5. Fictitious character (yes! good to display this data prominently!) should be in subfield c, not g.


As always — check out the full approved list

It’s Just New LCSH, You Should Really Just Relax

A couple notes before we begin:

The following subdivisions will be added to lists of free-floating subdivisions:

  1. H 1095, Free-Floating Subdivisions
    $x Case studies
  2. H 1100, Classes of Persons
    $x Case studies
  3. H 1103, Ethnic groups
    $x Case studies
  4. H 1105, Corporate bodies
    $x Case studies

African American zoologists

Picture of Roger Arliner Young, first African American woman zoologist
Roger Arliner Young received her PhD in zoology in 1940. Movie please.

Anti-communist movements in motion pictures

still from the film Dark Knight Rises
“Take *that* Occupy Movement” — Christopher Nolan

Archery ranges

Geena Davis notching an arrow, preparing to shoot it.
Be a fletch, be very a fletch.

Artificial limbs in art


Misty Knight, posing real cool
This is Misty, how may I kick your ass?


a NPC from the game Everquest
……not this thing, the Chinese mythological creature


A cartoon image of me.
Haven’t used this in a while!

Chinese American figure skaters

Figure skater, Tiffany Chin
Bet you thought I’d use Kwan. Think again.

 Cooking (Couscous)


bell peppers stuffed with couscous
brb gonna get some Near East

Excellence in the Qurʼan

انظُرْ كَيْفَ فَضَّلْنَا بَعْضَهُمْ عَلَىٰ بَعْضٍ ۚ وَلَلْآخِرَةُ أَكْبَرُ دَرَجَاتٍ وَأَكْبَرُ تَفْضِيلًا

(if you use Yusuf Ali’s English translation, this is hilarious)

Fans (Persons) in mass media

Still from Swimfan (2002)
Truly, the Misery of my generation.

Figure skating jumps

Surya Bonaly - flipping
The one. The only.
Surya Bonaly


Human-Neanderthal encounters

Still from Encino Man (1992)
Leave the “history + California” stuff to Bill and Ted.

Obscenity (Aesthetics)

Picture of piss christ
I hope the metadata folks got some good medium terms out of this.

South Asians on television

Still from "Master of None" (2016)
Does Netflix count as tv?

Spiralizers (Utensils)

cover of book called "8 life-changing ways to use a spiralizer"

The Future of Access Points

Authorized access point — Here’s where we REALLY get to the fun stuff! You’ll note that in current RDA/MARC authority records there is no separation between the preferred title and the authorized access point.

I said this in an earlier post — and I thought it was worth expanding on. This is that expansion.


As I continue to RIMMF my way along through comicbook-land, and as I continue to create name authority records in the Hebraica NACO funnel, I am also continuing to make access points.

Authorized access points (AAP) and variant access points (VAP) serve important roles in our bibliographic retrieval systems, so we should talk about them.

What are we even doing again?

It’s crucial that we (as catalogers and also the people who design our systems) think about what the heck we’re doing, and why we’re doing it. Why do we have AAPs?

A brief answer:

We have AAPs so that we can uniquely identify Agents (people/families/corporate bodies) Resources (works/expressions/manifestations/items) and Subjects. Furthermore we have them to serve as human-readable (and understandable) names for those same Agents, Resources, and Subjects.

Notice that that’s two entirely unrelated things that our current AAPs are supposed to be doing. Uniqueness, and human understandability are different!

Of course, it’s because card catalog.

(I’m trying to turn that into a stock phrase to represent why we do so much of what we do)

The Library of Congress has always recognized that we’re attempting to serve two masters and that using the strings as the unique identifier wasn’t ideal. That’s why every name authority record (NAR) has a unique ID in the 010 field in addition to the string found in the 1XX field.

So how will we decouple these in the future?

The AAP of Tomorrow

Some of this is based on what I heard Gordon Dunshire say at the RIMMF-a-Thon at ALA Annual. Some it is my own pie-in-the-sky guessing.

  1. The AAP of tomorrow will not be carried in the same field as Preferred name of the person, or Preferred title of the work. These are separate elements from the AAP and will be treated as such.
  2. While current instructions for RDA do provide guidance (some tighter than others) on creation of AAPs, it won’t always do that. It will provide guidance on the construction of each potential piece of an AAP, the preferred name/title, significant dates, form/content types, etc. But the construction of the AAP itself will be left to the application profile of your local system.

Let me say #2 again in more/different words because it is so crucial.

RDA (or whatever content standard we’ll be using) will help you identify and record elements in authority records

  • Names
  • Dates
  • Forms
  • Everything

But the unique identifier that you will not have to look at or even care about will be system generated. Since it will be unique, the AAPs will not have to be, and we will use our best judgement (helped along by well-vetted research as to what patrons find useful) to make the authorized access point for display.

Whatever link is generated for patrons to click to collate all the resources associated with that person, it will be based on that unchanging unique identifier, whether the text they click on says

  • Jackson, Michael, 1958-2009
  • Michael Jackson (Singer)
  • Michael Jackson, and there’s a picture of him next to the name

Should be left to what patrons find useful. The requirement “add X element to the AAP if necessary to distinguish” will be irrelevant because no element will be necessary to distinguish – -the unique identifier will do that.

What we’re going to do as future catalogers will be to record as much data as we can about agents and resource and provide identifiers for those agents/resources from other systems if they exist.


The Implication

I can see that some of you have already realized the implications. But I’m going to spell it out anyway.

Variant access points won’t exist.

Okay okay, maybe that’s taking it too far. But return to the questions above — why do we have VAPs? What purpose are they intended to serve?

VAPs provide alternative entry into the catalog for when an agent or resource is known by another name.

Well again, as above — RDA (or whichever content standard) will provide guidance on how to identify and record variant names and titles.

That’s it. That’s all you have to do!

Because there no longer will be the notion of a single authorized access point, thou shalt not have any other access points before me — there aren’t tiers of them either.

Systems (either local or shared) will generate access points for display and can generate hundreds or thousands hidden access points which can be matched on in case a patron searches differently than you expect. They are all equal to one another — and none have to be created individually by you, the cataloger.


Pretty cool, right?

Fictitious Characters (2016 : Ganin)

(Full disclosure, a lot of these thoughts rose out of my current Comic Book Project, but I’ve attempted to make it relevant to those who don’t give a fig about comics)


Ever since 2013, fictitious characters have been making their way from the Subject Authority File into the Name Authority File. This has lead to a lot of arguing amongst catalogers.

These arguments tend to take 2 forms:

  1. Philosophical
  2. Practical

The first argument centers around the fact that fictional characters aren’t actually capable of creating intellectual works. If you’ve read the FRBR-LRM document, you’ll notice that it excludes fictional characters (and non human animals) from the Agent class on those grounds. The document acknowledges that no matter what is printed on a resource, no matter what ‘legal fiction’ a publisher is purporting — some real human person (or persons) created the resource in hand, and that needs to be reflected in our metadata for said resource.

I’m not necessarily disputing this particular point. I don’t particularly care in which file we record fictional characters (and as we move to linked-data world, the distinction may become less important or even vanish) — but I care a great deal about the second point of contention: the practical.

Regarding the practical, there are three specific things that I want to talk about:

  1. Recording “Fictitious character” in the AAP
  2. Recording the other chapter 9 attributes for fictitious people
  3. Differentiating realizations of fictional characters

Fictitiousness in the AAP

For the first, if you’ve spent any time on AUTOCAT (amongst other list servs) over the last few years, you’ve seen this debate. The RDA rules are very clear that you do not have to do record it in the AAP, many wish you would…

(bolding mine):

9.6 Other Designation Associated with the Person

Other designation associated with the person is a core element for a Christian saint, a spirit, a person named in a sacred scripture or an apocryphal book, a fictitious or legendary person, or a real non-human entity.    Recording Other Designations Associated with Persons
Record other designations associated with the person by applying these instructions, as applicable:
saints (see
spirits (see
persons named in sacred scriptures or apocryphal books (see
     fictitious and legendary persons (see
non-human entities (see
other designation (see
Record other designations associated with the person as separate elements, as parts of access points, or as both. For additional instructions on recording a designation as part of the authorized access point, see and Fictitious and Legendary Persons

For a fictitious or legendary person, record Fictitious character, Legendary character, or another appropriate designation.

So as you can see, if the person is fictional it is required that we record that information somewhere (as 9.6 states it is a core element). This is undisputed. The arguments break out about where, and unfortunately (in this cataloger’s opinion) RDA is too lenient.

As the rule at states, there are three options for recording the fictional aspect:

  1. Separate element (and in MARC this would be a 368 subfield c)
  2. Part of an access point (in MARC a 100 subfield c)
  3. Both

It is here that RDA and I part ways. Because it is not required that you add fictitious status in an AAP, some people don’t — the famously contentious example being Captain James T. Kirk.

Kirk, James T., 2233-2371

The reason this bugs me so much, and the reason I think it should always be recorded in the AAP when applicable, is that the point of an AAP is to help the patrons (and the catalogers) differentiate between the ludicrously massive number of names and identities.

“Is this the John Paul Jones who played bass in Led Zeppelin, or the Navy guy?”

“Is this Michael Myers who murders people on Halloween, or the Canadian actor?”

I imagine it like we’re playing a game of twenty questions with a patron, and each question we ask ought to narrow down the pool of potential matches until we’ve identified their person.

“What is the name of the person?”

“What are some significant dates associated with the person?”

“Is the person fictional or real?”

How could you ever think to ask the second question before the third?! Imagine the set of all people: living, dead, and fictional. Clarifying dates can slightly help narrow a choice, but asking a boolean like “fictional” immediately cuts either 107,602,707,791 people (according to Google’s knowledge graph, that’s the number of people ever existing) or…however many fictional people ever existed which I feel like is….more than that.

So anyway — that’s my argument for why RDA should be revised to require fictitiousness to be recorded in the AAP.

Recording Other Chapter 9 Attributes + Differentiating

These last two of the practical points are so tightly related, that I don’t think I can separately discuss them. Let’s dive in.

There are two sets of instructions in the Subject Headings Manual which make it clear that LC’s policy is for a single name heading to stand in for all iterations of that character.

From H 1610 Fictitious Characters Section 5. Assignment of headings (emphasis mine)

For individual plays or poems assign a subject heading only if the character has been borrowed by the author from another author or source and used in the creation of a new work. For subject cataloging purposes, the borrowed character’s identity is considered the same as that of the character created by the original author. Do not create a separate name heading for the borrowed character. Instead, assign the same heading that would have been assigned to the original work.

We also see in H 1790 Literature: Fiction Special provisions. Section 4. Character(s) (emphasis mine)

Note: Fictitious characters may be borrowed by an author from another author, or from another source, and used in the creation of a new work. Assign the same heading to works by the original creator of the character and to works in which the character has been borrowed.

So this is pretty clear, right? A single heading/record for a fictitious character is meant to stand in for all incarnations of that character.

Well if that’s the case, then we need to stop recording some of the other chapter 9 attributes in those records. Take 9.3 “Date associated with a person”


  • William Shatner portrayed Captain Kirk in the original Star Trek tv series/films.
  • This Captain Kirk was born on March 22nd, 2233 in Iowa.
  • Chris Pine is currently portraying Captain Kirk in a series of movies and this Captain Kirk was born January 4th, 2371 in space.


Captain Kirk’s authority record in the NAF contains the following:

  • 046 $f 2233-03-22 $g 2371 $2 edtf
  • 370 $a Riverside (Iowa)

So I ask you this — if the authority record is really supposed to represent all versions of the character, then why are we fixing it with respect to birthplace and dates. Those are malleable attributes which may change from iteration to iteration.

Two more examples:

  • You’re going to find the heading “Holmes, Sherlock” on DVDs of and books about Moffat’s Sherlock. But that heading has an associated 046 of 1854. (Moffat’s Sherlock is obviously
  • ‘Spider-Man (Fictitious character)’ is applied to movies of Raimi’s or Webb’s Spider-Man. Yet the associated authority record has an 046 of 1962-08. (Never mind the fact that that’s not even the birthdate for Ditko/Lee’s Spider-Man, it’s the first issuance of the character in a comic!)

I can hear you now:

“Netanel, maybe the authority records should be recorded with data from the ‘original iteration’ of the character”

To which I say —

MUAHAHAHA, you fell right into my trap!


Catherine and I were talking about fictional characters in RDA (as we do) and she said something which blew my mind:

are fictional characters, in a sense, “works” that can have different expressions?


I hereby submit a “Work” record and two “Expressions” of Captain Kirk (most important bits highlighted)
100 1 $a Kirk, James T., $c (Fictitious character)
368 $d Captain
368 $c Fictitious characters $2 lcsh
372 $a Space flight $2 lcsh
374 $a Space travelers $a Astronauts $2 lcsh
374 $a Space ship captain
378 $q James Tiberius
500 1 $i Derivative (person): $a Kirk, James T., $c (Fictitious character : Roddenberry)
500 1 $i Derivative (person): $a Kirk, James T., $c (Fictitious character : Abrams)

046 $f 2233-03-22 $g 2371 $2 edtf
100 1 $a Kirk, James T., $c (Fictitious character : Roddenberry)
368 $d Captain
368 $c Fictitious characters $2 lcsh
368 $c Roddenberry
370 $a Riverside (Iowa)
372 $a Space flight $2 lcsh
374 $a Space travelers $a Astronauts $2 lcsh
374 $a Space ship captain
375 $a male
378 $q James Tiberius
376 $a Representative expression
500 1 $i Based on (person): Kirk, James T., $c (Fictitious character)
500 1 $i Created by: Roddenberry, Gene

046 $f 2233-01-04 $2 edtf
100 1 $a Kirk, James T., $c (Fictitious character : Abrams)
368 $d Captain
368 $c Fictitious characters $2 lcsh
368 $c Abrams
370 $a Medical shuttle 37 (Imaginary shuttle)
372 $a Space flight $2 lcsh
374 $a Space travelers $a Astronauts $2 lcsh
374 $a Space ship captain
375 $a male
378 $q James Tiberius
500 1 $i Based on (person): $a Kirk, James T., $c (Fictitious character)
500 1 $i Adapted by: Abrams, J. J. (Jeffrey Jacob), 1966-

I’m not saying this is perfect. Far from it, it’s a glimmer of an idea.

What I’m proposing though is that a “Work” record for a fictional character would maintain only the barest of bones, the ones that will carry over for every iteration, the ones which, if they changed, would necessitate the creation of a new work akin to the existing WEMI model.

I also added a made-up-field for the representative expression (a concept not fully codified but introduced in the FRBR-LRM document) because I do want to indicate that the Roddenberry Kirk is the first Kirk. (same reasoning for ‘created by’ versus ‘adapted by’)


What’s the freakin’ point?

I know, right? This is an awful lot of digital ink spilt for an idea (or a rumination) which will never be implemented and isn’t even necessarily well thought-out.


Fictional characters which have been transformed through media and time are studied by scholars, and the laypeeps alike. Why not help them find either the specific one they’re looking for, or a broader book about several of them?

I just don’t think that the following two resources really are equivalently about the same person and deserve the same heading of “Holmes, Sherlock”

  • The Lure and Intrigue of Sherlock Holmes : from Doyle to Rathbone, from Moffat to Miller
  • Sherlocked up with Watson : 15 highly erotic BDSM tales about Stephen Moffat’s Sherlock and Watson