Modeling, Explanation, and Ontology in the Cultural Sciences

Humanities computing must make conceptual modeling
one of its defining activities. This is necessary not only
to provide more effective computational support for scholarship
in the humanities, but also in order for humanities computing
itself to become something more than a “bag of tricks”. This
paper describes promising developments already underway in
several areas within humanities computing, generalizes from
past, if partial, successes (text encoding), draws parallels with
recent work in bioinformatics, and argues that the logical
identity of conceptual modeling and theory-construction makes
conceptual modeling simply a computationally oriented variant
of scientific explanation, and, more broadly, interpretation and
understanding in general. As an example of how model
development can provide fundamental insights into the cultural
world a case is described where a set of ontology evaluation
criteria developed within computer science is applied to an
influential framework for cultural entities and the result that a
radical reconfiguration of that framework is suggested.
Acknowledgements: There is undoubtedly much here that is
influenced by seminal papers on these issues by Willard
McCarty and John Unsworth; I acknowledge a debt to them
without of course any suggestion that they see things as
described here. Introduction
The nature of humanities computing has long been
problematic in the digital humanities community. Part of
the reason for this anxiety is that humanities computing is
struggling to move out of a pre-scientific phase. “Science” here
is of course intended to be taken in the broadest possible sense
— the point is not to separate disciplines such as physics on
the one hand from, say, literary studies on the other, but rather
to distinguish practices that have primarily non-epistemic
non-theoretical objectives from those that have understanding,
explanation, and knowledge as their principal immediate goals.
Nor is there any implication of a clear boundary between
scientific practices and non-scientific practices, or that there
isn’t a complex and essential mutual involvement — the claim
is only that some rough distinction of this sort is possible and
useful.
If humanities computing is to mature as a coherent field of
intellectual inquiry it will need to make a transition from being
an ad hoc set of useful techniques, to more systematic general
methods and theories that make direct contact with the specific
methods and theories of the disciplines it supports. This theme
is not new, but our progress has been modest so far and some
further discussion and elaboration is in order.
First, there are new reasons to be optimistic. For one thing this
progress is already occurring in other informatics
fields—bioinformatics as an example—and the specifics of
these changes provide an indication of how things might go in
humanities computing as well. Moreover the growing
significance of conceptual models and ontologies that is already
in evidence in humanities computing, and the nature of the
discourse around them, is a sign that this transition is perhaps
already underway. Finally one of the most theoretically
productive areas within the humanities computing community,
namely text encoding, arguably has been as successful as it has
precisely to the extent that it has embodied the features
recommended here — and owes its limitations and
disappointments to the extent to which it has not. What all these
things have in common is the foregrounding of conceptual
modeling, explicitly or implicitly.
I use the term “conceptual modeling” broadly, meaning any
formally defined abstract representation of a domain of interest.
Typical examples of modeling languages are the ER and UML
diagrams used in business applications, logic-based knowledge
representation formalisms such as description logics or the
frame languages (such as the KL-ONE family) common in
artificial intelligence, and the various formal ontologies now
used in computational biology and elsewhere
Science in the broad sense is the development of understanding.
Scientific understanding comes typically in the form of theories,
and theories are at least in part the systematic identification of
objects and relationships that are the salient features of some
domain. Informatic disciplines must put these theories,
formalized as conceptual models that support computational
processing, at the center of their identity. And for informatic
disciplines to be sciences themselves they must not only exploit
such conceptual models, but participate in their development
and exploration. This is true for informatics in general, and for
humanities computing in particular. Modeling in Bioinformatics
Other areas that began as craft-like bag-of-tricks
approaches to applying computing to a scientific field
have now begun to evolve a more coherent body of method,
theory, and metatheory. Perhaps the most dramatic case is
bioinformatics. Although initially a haphazard and opportunistic
application of computing techniques, there are now a
considerable number of families of well-theorized methods
tightly tied to biological theory. Of particular interest here is
the role of so-called ontologies in contemporary bioinformatics
such as the Gene Ontology (Ashburner et al., 2000), the
Foundational Model of Anatomy (Rossee & Mejino, 2003),
and other ontologies in genomics, molecular function,
neuroscience, biodiversity, and other areas of the life sciences
as well. An enormous amount of effort is going into the
development and use of these formal models.
In some cases the influence on biological science per se has
still been modest, but generally the results are extremely
promising and in a few cases the influence has been stunning.
The Gene Ontology in particular, has been enormously
successful and provides the overall framework for contemporary
genomics. Of special significance for us is the extent to which
the fundamental explanatory entities and properties of biological
theories are explicitly identified as the core constituents of these
conceptual models, blending the work of first order science
with its bioinformatics support.
Modeling in the Cultural Sciences
Within the humanities computing community text
encoding systems such as the TEI can be seen as a step
in the same direction as the new ontology-based models in
computational biology. Behind the methods of the TEI is at
least arguably an informal positing of text models, components,
and relationships. However unlike the bioinformatic ontologies
the models being indicated by text encoding techniques are
only indirectly and implicitly identified. This has been pointed
out by a number of critics (Raymond & Tompa, 1992;
Raymond, Tompa, & Wood, 1996, 1998, 2001; Buzzetti, 2002),
and projects to develop remedies are underway
(Sperberg-McQueen et al., 2002; Renear et al., 2003).
Another promising sign are the ontologies being developed in
museum studies CIDOC/CRM (Crofts et al., 2003) and library
science (IFLA 1998). For the most part these are being put
forward as frameworks for designing digital information
management systems or shaping the development of standards,
policies, and procedures, but to the extent that they succeed at
efficient, functional, and interoperable systems they may be
considered confirmed as substantive theoretical proposals for
how to understand the nature of the cultural world. And where
these models and ontologies appear to be applications of
existing explanatory theories of cultural objects and
relationships that is more evidence that they are achievements
which advance our understanding of that world.
An Example: Revising FRBR
What follows is an example intended to support two
claims. First that the evolution of conceptual models
intended to support information systems design can in fact be
an activity of first order science (in the broad sense), and second
that precisely the same techniques designed for general
evaluation of ontologies in natural science can be effectively
employed in evaluating cultural ontologies.
Several models for cultural objects identify such things as
works, texts, editions, and individual items as types of things,
fundamental kinds. Not only are various sorts of conceptual
arguments advanced in favor of such an analysis, but so is the
relative success of such frameworks in guiding the design of
effective information systems that are effective, efficient, and
interoperable. That is, the conceptual and practical success of
these systems can be taken as indicating that they correctly
represent how things are. A well-known model of this sort is
IFLA’s Functional Requirements for Bibliographic Records
(IFLA 1998), a framework designed primarily to support library
cataloguing, but which clearly has wider application — FRBR
in fact describes itself as a “conceptual model of the
bibliographic universe”.
A puzzle recently emerged about an assignment of one
particular cultural object (XML documents, as defined in the
W3C XML standard) to the appropriate FRBR entity type.
Some considerations argued for an assignment to the FRBR
expression (roughly: text) entity type, and other competing
considerations argued for assignment to the FRBR manifestation
(roughly: edition, or format) entity type. The proposed
resolution was an awkward and unsatisfying exception to the
FRBR framework which preserved the ontology of works,
expressions, editions, at the cost of complexities elsewhere.
(Renear et al 2002, Renear 2005).
Two years later however it was noted that when a proposed
criterion for ontology evaluation which had been developed
for scientific and general purpose ontologies (Guarino & Welty,
2000; Guarino & Welty, 2002) was applied to the FRBR
framework the result was an anomaly that suggested refactoring
the four Group 1 entity types into new entities and
corresponding roles. (Renear, 2006). Under the resulting revised
model the manifestation and expression entity types are not
treated as true entity types (fundamental kinds), but rather roles
(relationships) that entities of other types have in particular circumstances—and in fact this revision was extended to all
four FRBR entities (Renear & Dubin, forthcoming). We realized
that this new way of conceptualizing cultural objects appeared
to be an application of a more general approach developed by
John Searle (1995) as well as consistent with some work in
aesthetics (Levinson 1980). In short, all four FRBR “entities”
can be reinterpreted not as entities but as roles that other
non-cultural things have under specific social contexts of
“collective intentionality”.
That this is not just a practical adjustment in a conceptual model
for humanities computing, but a substantive claim in the science
of cultural objects in general, can be established by comparing
it to familiar claims about cultural objects by scholars such as
Ingarden, Wellek, Richards, Fish, Goodman, Tanselle,
Schillingsburg and others. In this comparison contradiction will
serve as well as convergence to make the point that these
assertions fall within the domain of humanist inquiry, and not
merely within some auxiliary practice.
The Relative Neutrality of these
Recommendations
It may be thought that putting conceptual modeling at the
heart of humanities computing and seeing the development
of models as first order scholarship in the humanities requires
accepting some archaic and dubious philosophical view, a
positivistic scientism perhaps, or even philosophical realism.
But this is not so. It is possible to engage in ontology without
taking a philosophical meta-ontological position. Many different
philosophical positions (including constructivism and
relativism) are all perfectly consistent with taking conceptual
modeling as the natural activity of humanities computing, and
at least part of humanities scholarship more generally. Of course
the results of modeling will not be neutral vis-à-vis other first
order theories, but the activity of modeling need not be
exceptionally controversial. Again the experience of
bioinformatics, where researchers with different philosophical
views (or no philosophical views at all) nevertheless often agree,
or more importantly disagree, on specific issues without
rejecting the overall approach, is significant.
This is not to say that the position is entirely neutral. It may
well conflict with some general accounts of humanistic inquiry
which hold that such inquiry is radically different than the
natural sciences, especially if those accounts are understood as
applying in general or without exception to the entire range of
humanistic scholarship — although I am not yet convinced that
it does in fact conflict with the traditions of hermeneutics and
verstehen, despite the differences in tone and direction. In any
case that the recommendation is not entirely uncontroversial
cannot be taken as a decisive mark against it at the outset.
Whether it is sound or not will depend on how it fares against
its competitors in improving our knowledge and understanding
of the cultural world.
Concluding Remarks
Ihave argued that humanities computing must make
conceptual modeling its central defining activity if it is to
fully realize its promise to substantially advance humanities
scholarship, and that such modeling is an intrinsic part of first
order humanities inquiry, and not merely an auxiliary activity.
Bibliography
Ashburner, Michael, et al. "Gene Ontology: Tool for the
Unification of Biology." Nature Genetics 25 (2000): 25-29.
Buzzetti, Dino. "Digital Representation and the Text Model."
New Literary History 33.1 (2002): 61-88.
Cover, Robin. "XML and Semantic Transparency." 1998. <h
ttp://xml.coverpages.org/xmlAndSemantics.
html>
Cover, Robin. "Conceptual Modeling and Markup Languages."
2001. <http://xml.coverpages.org/conceptual
Modeling.html>
Crofts, Nick, I. Dionissiadou, Martin Doerr, and Matthew Stiff.
Definition of the CIDOC Object-Oriented Conceptual Reference
Model, ISO/TC46/SC4/WG9/N2, International Organization
for Standardization. 2003.
Guarino, Nicola, and Christopher Welty. "Evaluating
Ontological Decisions with OntoClean." Communications of
the ACM 45.2 (2002): 61-65.
Guarino, Nicola, and Christopher A. Welty. "A Formal
Ontology of Properties." Proceedings of the 12th European
Workshop on Knowledge Acquisition, Modeling and
Management Lecture Notes In Computer Science 1937 (2000):
97-112.
International Federation of Library Associations. Ed.
Marie-France Plassard. Functional Requirements for
Bibliographic Records: Final Report. UBCIM
Publications-New Series. München: K.G.Saur, 1998.
Levinson, Jerrold. "What a Musical Work Is." The Journal of
Philosophy 77.1 (1980): 5-28.
McCarty, Willard. "Humanities Computing: Essential Problems,
Experimental Practice." Literary and Linguistic Computing
17.1 (2002): 103-125. doi:10.1093/llc/17.1.103 McCarty, Willard. "Modeling: A Study in Words and
Meanings." A Companion to the Digital Humanities. Ed. Susan
Schreibman, Ray Siemens and John Unsworth. Oxford:
Blackwell Publishing Ltd, 2004. 254-70.
Raymond, Darrell R., Frank Wm. Tompa, and Derick Wood.
Markup Reconsidered. Department of Computer Science,
Technical Report No. 356. The University of Western Ontario,
1993. Presented at the First International Workshop on the
Principles of Document Processing, Washinton DC, October
21-23 1992; an earlier version was circulated privately as
"Markup Considered Harmful" in the late 1980s.
Raymond, Darrell R., Frank Wm. Tompa, and Derick Wood.
"From Data Representation to Data Model: Meta-Semantic
Issues in the Evolution of SGML." Computer Standards &
Interfaces 18.1 (January 1996): 25-36.
Renear, Allen H. "Is An XML document a FRBR Manifestation
or a FRBR Expression? — Both, Because FRBR Entities are
not Types, but Roles." Proceedings of Extreme Markup
Languages, Montréal, Québec, August 2006. 2006. <http:/
/www.idealliance.org/papers/extreme/Proce
edings/html/2006/Renear01/EML2006Renear01
.html>
Renear, Allen H., David Dubin, C. M. Sperberg-McQueen, and
Claus Huitfeldt. "Towards a Semantics for XML Markup."
Proceedings of the 2002 ACM Symposium on Document
Engineering. McClean, Virginia, November 2002. Association
for Computing Machinery, 2002.
Renear, Allen H., and David Dubin. "The Four FRBR Group
1 Entities are Roles, Not Types." Forthcoming.
Renear, Allen H. "Text from Several Different Perspectives,
the Role of Context in Markup Semantics." Atti della conferenza
internazionale CLiP 2003, Computer Literacy and Philology,
Firenze, 4–5 December 2003. Ed. C. Nicolas and M. Moneglia.
Florence: University of Florence Press, 2005. < http://p
eople.lis.uiuc.edu/~renear/clip2005.pdf>
Renear, Allen H., Christopher Phillippe, Pat Lawton, and David
Dubin. "An XML Document Corresponds to which FRBR
Group 1 Entity?" Proceedings of Extreme Markup Languages,
Montréal, Québec, August 2003. Ed. B. T. Usdin. 2003.
Rosse, Cornelius, and José L. V. Mejino, Jr. " A Reference
Ontology for Biomedical Informatics: the Foundational Model
of Anatomy." Journal of Biomedical Informatics 36.6 (2003):
478-500.
Searle, John R. The Construction of Social Reality. New York:
The Free Press, 1995.
Sperberg-McQueen, C. M., David Dubin, Claus Huitfeldt, and
Allen H. Renear. "Drawing Inferences on the Basis of Markup."
Proceedings of Extreme Markup Languages 2002, Montréal,
Québec, August 2002. Ed. B. T. Usdin and S. R. Newcomb.
2002.
Sperberg-McQueen, C. M., Claus Huitfeldt, and Allen H.
Renear. "Meaning and Interpretation of Markup." Proceedings
of Extreme Markup Languages, Montréal, Québec, August
2000. Graphic Communications Association, 2000.
Unsworth, John. "Scholarly Primitives: What Methods do
Humanities Researchers Have in Common, and How Might
Our Tools Reflect This?" Paper presented at Symposium on
Humanities Computing: Formal Methods, Experimental
Practice. King's College, London, May 13, 2000. 2000. <htt
p://www.iath.virginia.edu/~jmu2m/Kings.5-
00/primitives.html>
Unsworth, John. "What is Humanities Computing and What is
Not?" Distinguished Speakers Series, Maryland Institute for
Technology in the Humanities, University of Maryland, College
Park MD, October 5, 2000. 2000. <http://www.iath.v
irginia.edu/~jmu2m/mith.00.html>
Unsworth, John. "New Methods for Humanities Research."
The Lyman Award Lecture, National Humanities Center,
November 11, 2005. 2005. <http://www3.isrl.uiuc.
edu/~unsworth/lyman.htm>
Unsworth, John. "Knowledge Representation in Humanities
Computing." n.d.. <http://www.iath.virginia.edu
/~jmu2m/KR/>