BIBFRAME Through The Looking Glass

A strong indication that something may be  amiss in BIBFRAME is that the picture drawn by its  model of the bibliographic universe  does not include books, and  as Oxford logician Charles Lutwidge Dodgson observed, “What use is a picture without books?”.

What, If Anything, Is a bf:Holding?

Rather than modelling a specific copy of a book, a BIBFRAME Holding models someone saying things that appear to be about a specific copy of a book. Surprisingly, there is nothing in BIBFRAME’s ontology that corresponds to any copy of a book that the statement might be about.

We cannot ask how many copies of books we know about (unless we are satisfied with the trivial answer “none”). We can only count how many times we have been told about copies of books.   If we have a BIBFRAME Holding from OCLC saying that UNC has a copy of a certain book, and another from UNC saying the same thing, we have two different Holdings.

From this behavior, it seems that a BIBFRAME Holding corresponds to a MARC Holdings Records, rather than a physical book, or FRBR Item.

Do BIBFRAME Annotations  Support Provenance for Holding?

One of the reasons given for  BIBFRAME Annotations is that they carry information about who asserted a statement (via the bf:annotationAssertedBy property).

Despite the  conflation  of Epistemic  and Ontological concerns that we see in bf:Holding, we cannot rely on  the annotationAssertedBy property to support provenance.  If we have one data source that gives one value for  bf:annotationAssertedBy, and a different data source that gives another, conflicting value, we  have no principled way to choose between them, unless we have other provenance information to support our decision.

The problem is readily apparent in the example used for the Holdings Annotation.

<http://xyz.org/bibframeTestCases/Annotations/Annotation/Holding>
    a bf:Holding ;
    ...
    bf:annotationAssertedBy <http://id.loc.gov/vocabulary/organizations/dlc> ;
    bf:dateOfAssertion "20130807" ;
....

The URI for the Holdings Annotation is in the namespace “http://xyz.org/”.   The host from which the document was   retrieved is in the bibframe.org domain.

There is a statement claiming that the Holdings Annotation was asserted by http://id.loc.gov/vocabulary/organizations/dlc .

There is no a priori reason to prefer this assertion  over a claim, in an equally unrelated graph, that

<http://xyz.org/bibframeTestCases/Annotations/Annotation/Holding>
     bf:annotationAssertedBy <http://id.loc.gov/authorities/subjects/sh89004503> ;

The value of a provenance mechanism that cannot be used without an additional  provenance mechanism is questionable.

Are Annotations Consistently Used For Opinions?

Section 2.1 of the Annotations document states that “Annotations are asserted in order to: [...] express opinions about a resource”.

However, this principle does not seem to be widely applied.  To take the most obvious example, “subject” is treated as a property, despite  Patrick Wilson’s generally accepted arguments that “the subject of a work” is indeterminate.

It is hard to understand the theoretical motivation for  treating subjects as objective, yet regarding cover images regarded as subjective.

Is Provenance Information Needed For All Assertions?

Since  information about a resource is only obtainable empirically, filtered through perception and mind, and  recorded empirically through misperception and typos, without recorded provenance information  we have  weak justification for any bibliographic knowledge we may claim.

The more independent sources for a statement about a sensible property, the more strongly we may be justified in  believing it.

The more trust we have in the expertize of a source with regards to a particular topic, the more we may trust their opinions that a work is or is not about a subject within that topic; however we might simultaneously accept as equally valid the opinion of a specialist in a different field that the work is about a subject in that  field.

If the goal of BIBFRAME is to support the fusion  of  descriptions from multiple different sources, then  provenance  must be supported for all assertions.  If the goal of BIBFRAME is to provide another syntax for storing monolithic records, then the overhead of decomposing these records into individual assertions would seem to require additional justification.

Comments No Comments »

In these prolegomena I will be examining how a lack of shared understanding of these issues may have lead to certain choices that have resulted in unintended consequences in the current proposed BIBFRAME. I will show that these choices and consequences are not intrinsic to all linked data or semantic approaches to bibliographic and other metadata, and that alternatives approaches may have advantages.

I will attempt to give general explanations of some of the terms and  concepts that may not be obvious to those outside the communities in which they involved.   By attempting to bridge the gaps in understanding separating the different communities needed to make any future metadata framework, I hope to make the case that such a framework is both possible and desirable, and to allow more people to make positive contributions to the creation and deployment of such a framework.

In this introduction I give an outline of the remaining prolegomena, and look at the types of knowledge and skills that are needed to build a future Bibliographic Framework .

An outline of the prolegomena

Part II.  What is a Holding that cannot be held?

In the second prolegomenon, I use the Holdings Annotation as presented in the current incarnation of the bibframe proposal as an example some of the problematic features of the current BIBFRAME model.

Part III.  Only Connect: Why a Framework Based on Linked Data and Semantic Technologies Is Worth Pursuing.

In the third prolegomenon I discuss the how using Linked Data and Semantic Technologies can improve the quality and accuracy of existing cataloging practice, and look at some new architectures and applications that a  framework designed for networked use can enable.

Part IV.  IMEW: The Opposite of FRBR is FRBR.

I examine some properties of  FRBR-like data models.  Taking a bottom-up approach, starting with  Items, I relate the main entities in the FRBR WEMI model to concepts found in common  version  control systems used in software development.  I discuss some issues involved in inferring  properties of an Item based on it’s corresponding  Manifestation, Expression, and Work,  raise some questions as to possible identity criteria, and look at some other possible entities that may be implicit in the model .

Part V. On What The…? Ontology and Linked Data for BIBFRAME

Part IV briefly covers some concepts in linked data and ontologies, makes the point that owl and RDF are not the only targets, and that it is more important to accurately capture the semantics of what is being modeled than it is to think about what something might look like in XML. I argue for the use of a more expressive form of modeling and for the use of controlled natural language to specify the model. CNL makes it much easier for people to contribute, and makes it much easier to review.

Part VI:  Implementation: A Simple Matter Of Programming

Part V looks at some of the technical solutions available, some areas of computational complexity, and areas where work is needed. In particular, I describe a possible approach to applying Provenance information to Named Graphs that is compatible with the forthcoming RDF/1.1 recommendations.

Who Knows?  The Location of Knowledge For A New Bibliographic Framework.

The development of the BIBFRAME model of the Bibliographic Universe, and the representation of that model as linked data, using some combination of semantic technologies  has been the occasion of much discussion, and much confusion. The cause  of much of this confusion may perhaps be found in the interwoven, interdisciplinary sources of the basic concepts needed to design and implement a new framework for bibliographic description.

Many words that are precise terms of art in a discipline are used in BIBFRAME to carry very different meanings.  Sometimes this is because the terms has a different meaning in  a different discipline.  In other cases the use of a term appears to have arisen within BIBFRAME.

The blind men and the metaphor

We find ourselves in the familiar situation of  blind men and elephants- arguing fruitlessly amongst ourselves as  dread beasts thunder towards us from  directions unknown  shaking the very earth with each stride. Ontologists and Linked Data specialists; catalogers and cataloging theorists; programmers and data modellers- each hold different pieces of the puzzle, but until the lid of the box is scraped off the  elephant’s foot, we can’t tell how  all the pieces fit together, or which pieces might be missing.

What The Cataloger Knows

As cataloging has evolved over the course of the past few centuries it’s practice has accrued  many rules and conventions, and has been the subject of Great and lesser debates.  As with any evolved system, some of these rules have evolved in response to conditions that no longer reply.

Some practices may be overcomplicated  and purely vestigial – for example,  rules dealing with the works attributed to Spirit Authors.

Other practices may be oversimplified due to the limitations of available technology  to economically produce or make use of more specific technologies. Punctuation that could be generated automatically may be recorded in data elements; information may be recorded only as plain text, whence only natural or artificial intelligence approaches can recover it.

The cataloger knows that all records are wrong.  The cataloger knows that some records are useful.

 Unless we understand basic principles of cataloging, we cannot consider “is this rule necessary?”; we cannot even reach the question   “is this even a rule?”

What The Network Expert Knows

Without a general understanding of how the Internet and the World Wide Web work, we cannot talk about how to design a system that will work at the scale.  We should  understand how replication and caching can be used to provide reliability and performance.

We should understand the dangers of building systems around single, authoritative points of failure; we should appreciate the freedoms that real-time, distributed  designs can allow.

We should understand the security requirements that a framework requires; we should understand what frameworks security technologies enable.

The Bibliographic Network must work in harmony with the ‘Net. 

What The Linked Data and Semantics Expert Knows

Without an understanding of  linked data, ontologies, and semantic technologies in general, it is can be hard to see the benefits that using these technologies can provide.

If we miss the distinctions and purposes of between  different types of languages and systems, it is easy to confuse an ontology language like OWL with a markup description language like XML Schema.

If we confuse the capabilities and  limitations of RDF  and OWL  with  the capabilities and limitations of  all ontology languages and technologies we may lose the opportunity  to capture the knowledge that the Subject Matter Experts are trying to convey.   It is easy to generate  a simplified model in a less expressive language from a more precise model. It is easy to generate simplified models in multiple less expressive languages.

Without knowing what kinds of knowledge can be captured, how that knowledge can be used, and what the costs of using that knowledge in different ways might be,  we cannot decide what level of detail is necessary.

Knowledge that is captured need not be used;  knowledge  not captured  cannot be used.    

What The Philosopher “Knows”

Without a grasp of the outline of some basic concepts from Metaphysics, it is easy to get caught up in trying to find the “one true answer” to questions which have many equally valid solutions. Conversely, one may not be aware of possible solutions that may be well particularly well to suited to a particular problem, or of how some sets of choices may be incompatible.

It is easy to accidentally conflate questions of  Ontology (what there is), and Epistemology (what we know, and how we know it).

⚠ Important Safety Information ⚠

  • Questions that have resisted a definitive answer for thousands of years are seldom resolvable even with video teleconferencing and Powerpoint.
  • Metaphysics is an intrinsically dangerous tool.
  • Always follow safe handling procedures. Never allow the muzzle to cover anything that doesn’t exist.
  • When in doubt, leave the room and  consult a doctor of philosophy.
  • When not in doubt, leave the room, and consult two doctors of philosophy.

 

Comments No Comments »

Jeff Beall reposts some of the discussion on the  MODS mailing list on the recently published draft of MADS/RDF.  Beall wonders if the proposal, and the initial comments indicate that the entire semantic web effort is doomed. This may be an over-reaction.

I spent some time working on a formal analysis of the semantics of  pre-combined[1]/subdivided subject headings, and how they could be added in a downwards compatible way to SKOS. I had hoped to discuss the topic at the DC VOCAMP held at The Library  immediately after ISWC 2009, but there was no-one at the camp who was interested in library vocabularies, I let the matter drop[2].

Since support for pre-combined headings are one of the  major design goals of  MADS/RDF, I have been looking quite closely at the initial  draft.  I am not yet ready to publish detailed results, but I wanted to take this chance to make a few quick notes on the initial draft.

Minor Issue

There is a quick change that could make it easier for others to evaluate the proposed ontology.  Currently, the URI/IRI of the ontology is “http://id.loc.gov/ontologies/mads/2010/11″.  At the moment does not resolve as a URL.  Also, the name is slightly different from current usual practice.  Ontology URIs are typically of the form …/<Year>/<Month>/<Name>; the MADS IRI is in the form …/<Name>/<Year>/<Month>. This encourages some common tools, such as Protege, to abbreviate the ontology name as “11″.

Major issues

  1. The SKOS mappings are imported as part of the reference ontology; however due to problems in naming, this may not show up correctly.  It might be better to include the axioms in the mapping file directly in the main ontology.
  2. The SKOS mappings are incomplete; it is possible that concentrating  on narrowing the differences between SKOS and Standard(NISO Z39.19)/Library KOS semantics would give an end result that should be more useful for use by The Library.
  3. The current draft was created without support from an ontologist or owl expert. There are several problems in the modeling that result from this lack of resourcing.  For example, there are several places where a constraint or restriction is given in a comment that could readily have been represented in OWL, and many properties are not fully or properly specified.  There is a big difference between the ontology design and schema design, and without support resources, it is easy to fall in to some common anti-patterns.
  4. The current draft was created without ready access to Subject Matter Experts (SMEs) in the Cataloging COI.  As a result, the ontology may not adequately represent the problem domain, and may make it harder to extend so as  to support LC rules and policy on subject heading formation. When modeling any complex area, regular interaction with relevant SMEs is essential throughout the process, both for knowledge elicitation and model validation.  Early, regular review of the ongoing work  by MODS/MADS Editorial Committee may be helpful.
  5. The current draft does not define the semantics of pre-combined headings; there are different possible interpretations,  but none are discussed, and some inferences appear to be common across all readings.
  6. Given known  data errors in the syndetic structure of LCSH (see e.g. LCSH is to Thesaurus as Doorbell is to Mammal), it may better to devote resources to correcting these issues before deepening the semantics of exposed Linked Data.  This would make the LC vocabularies more reuseable in the web of data, as recommended in the Deanna Marcum’s  ”Response to: On the record: Report of the Library of Congress Working Group On The Future of Bibliographic Control“.



[1] Following Soergel, I prefer this term to “Pre-coordinated”

[2] I also worked on relating the to the implied ontologies inferable from Knowledge Organization Systems that used sub-types of BT relationship (BTG, etc) that are part of the international standards, and which were originally included in SKOS, but which were moved to an extension , then finally dropped and relegated to examples in the published recommendation.

Comments No Comments »

Now that registration is open, you may be thinking about attending this year’s ADHO Digital Humanities Conference, DH2010, which will be

hosted at King’s College London by the Centre for Computing in the Humanities and the Centre for e-Research, with the support of the School of Arts and Humanities, Information Services and Systems, and the Principal, Professor Rick Trainor.

Before you decide to register, you might wish to consider whether King’s can really afford to be hosting an event of this size, especially  in the light of the savage cuts  that King’s College has felt forced to make this year . Faculty cuts include Shalom Lappin! It’s like MIT laying of TimBL.

You might feel it better for King’s  for you to focus your attention and submissions on other venues, and to bypass DH2010.

If you do, be sure to drop an email to  Professor Trainor explaining your decision.

Comments No Comments »

Translation party uses the Google Translate API to  turn Machine Assisted Translation abuse into a one-click spectator sport.

They automagically translate an phrase from English to Japanese and back again until they reach an equilibrium, or decide that they’ve tried enough (sometimes they give up prematurely, but you can only beat a dead horse for so long (or as they would say “Only between the wild-goose chase“.

Sometimes, however,   Translaton party discovers makes some remarkable discoveries.

[table id=2 /]

Comments No Comments »

Damned Yankees

allRecords/ArchivalDescriptions/48/Item_95313.xml:

<display-name>New York World's Fair (1964-1965)</display-name>

Good ol’ Boys

allRecords/ArchivalDescriptions/68/FileUnit_763881.xml:

<title>Cleveland-Harrison (1885-1893): South Carolina – Worlds Fair</title>

Comments No Comments »

Whilst checking up on some of the character set fixes, I’ve noticed that there seem to be some inconsistencies in the use of right quotes, apostrophes, and accute accents, which may lead to problems down the road when doing searches, or performing name authority work.

Here are some examples that may make things clearer.

Use of a right single quote as an accute accent.

In  allRecords/ArchivalDescriptions/68/FileUnit_744305.xml we find the following title:

<title>Navarro, Jose’ Antonio</title>

There are several ways to represent é in UNICODE; composed, decompsed, or as an e followed by a non-combining accent  (U+00B4); none of these approaches will match this title string.

One correct form is:

<title>Navarro, José Antonio</title>

(see e.g. http://www.tamu.edu/ccbn/dewitt/Navarro.htm)

Some accute accents are encoded correctly

For example, in allRecords/ArchivalDescriptions/1/Item_197118.xml we find:

<display-name>Québec (Québec province, Canada, North and Central America) inhabited place</display-name>

Vive Jay Sherman! Vive Québec

Comments No Comments »

In a case most similar to the landmark US case of Feist vs. Rural Telecommunications,  the Federal Court of Australia  has ruled in the case of Telstra Corporation Limited v Phone Directories Company Pty Ltd that mere sweat-of-the-brow does not reach the minimum level  of creativity required to create copyright in compilations of facts. As with Feist in the United States, OCLC’s copyright claims on the Worldcat database  would not seem to survive  in Australia under this decision.

Commercial value not relevant in deciding whether copyright exists

(20)(8) … Copyright is not given to reward work distinct from the production of a particular form of expression: IceTV [2009] HCA 14; 254 ALR 386 at [28] and [31]. Accordingly, it is “unhelpful to refer to the ‘commercial value’ of the information, because that directs attention to the information itself rather than to the particular form of expression”: IceTV [2009] HCA 14; 254 ALR 386 at [31] and [166].

Application of rules, no matter how skillfully is not enough to obtain copyright

Section (C) (6) seems especially applicable to some mistaken claims of copyright in bibliographic records;  as in Batlin v. Snyder, <<the requirement of originality [cannot]  be satisfied simply by the demonstration of “physical skill” or “special training”>>

(162) As the evidence shows, the person or persons who utilise the Rules and who, therefore, are submitted by the Applicants to be authors of the Works, do not exercise either “independent intellectual effort” or “sufficient effort of a literary nature” to be considered an author within the meaning of the Copyright Act.

(163) First, it is the Rules that prescribe the particular form of expression of the Works, not any individual person alone or in collaboration with others. The Rules control the content of each Work. The various computer systems generate and control the choice of the content on the basis of the Rules and are responsible for ensuring the vast majority of their valid application. In some circumstances the Rules expressly prohibit certain content.

(164) Secondly, even where there is any level of human “discretion”, it must be exercised in accordance with the Rules. There is no independent effort, let alone such effort being intellectual. There is no effort, let alone sufficient effort, of a literary nature. I reject the Applicants’ contention that there is judgment or discretion used in selecting the material for inclusion. The Rules prescribe, presume and prohibit the actions of the contributors. What choice there is, is the choice given by the Rules, not by any person or persons.

(165)Thirdly, even if the judgment or discretion of the kind asserted by the Applicants was “independent intellectual effort”, it is not relevant intellectual effort. I reject the Applicants’ contention that the relevant intellectual effort involved is in understanding and applying the Rules. The “independent intellectual effort” required must be directed to the creation of the Work: IceTV [2009] HCA 14; 254 ALR 386 at [33]. Such skill and judgment, on the facts, is not directed to the creation of each Work but to the application of the Rules. Moreover, the “work” undertaken by many of the Sensis workers, is work directed (at least in part) to the maintenance and operation of the Genesis Computer System and associated computer systems (including the Genesis database) which is accessed or used for a multitude of applications.

(166) Finally, even if it mattered, there is insufficient evidence about who created the Rules themselves. The Rules alone cannot be said to be an author for the purposes of the Copyright Act.

Comments No Comments »

I’m happy to announce the availability of the new  Tyler Hansbrough Special Edition of the NARA Archival Research Catalog. This special edition has fixes for the  character set encoding problems noted below. These problems were being caused by problems in the  UTF-8 encoding of   various single and double quotes, and en- and em- dashes, resulting in unparsable XML documents.

Downloading the modified data sets

The  Tyler Hansbrough special edition can be downloaded from the Fred 2.0 repository on iBiblio.

Affected characters

The following characters were affected – the counts given are for the number of times each character occurred, and not the number of documents each character occurred in.

[table id=1 /]

Affected documents

533 documents were affected.

A detailed list of which documents were affected, and which characters were replaced is available here.
Read the rest of this entry »

Comments No Comments »

There are a number of problems in the ARC data files that are caused by truncation of 16 bit unicode punctuation to 8-bits. There may be other undetected truncation errors where the truncated value is a legitimate UTF character, but this seems unlikely.

I have written code to work around this problem; it’s a fairly simple fix.

There are also about a dozen records that are ill-formed- the contain tags generated by an Oracle xml sql engine.

Scary thought for the day:
Imagine if Oracle made an ILS

Comments No Comments »