Synthetic Classification and Diverse Communities

Abstract

This brief paper argues that a synthetic approach to classification can alleviate all of the major concerns that are commonly raised about how Knowledge Organization Systems (KOSs) may disserve various communities. It surveys how a synthetic approach can potentially address a variety of concerns regarding KOSs and social diversity.

Full text

Synthetic Classication and Diverse Communities
Rick Szostak, University of Alberta
There are a variety of ways in which a KOS might disserve members of particular
communities. In each case, a synthetic approach to classication oers a potential
solution. This paper thus explores the various ways in which a synthetic approach to
classication may liberate the capacities of members of diverse communities. Many
of the topics addressed were taken from the call for proposals.
Biases in Classication
If the classication system re#ects the cultural biases of its designers, members of
other communities may nd the classication di$cult to comprehend and even
oensive. A classic example here is a presumption within some classications that
nurses will be female: A special subclass for “male nurse” is then created, indicating
that this is thought to be the unusual case. This problem might be addressed by
carefully surveying existing KOSs and addressing each case of unequal treatment of
dierent communities individually. Alternatively, a synthetic approach to
classication can address the challenge at a holistic level. In a synthetic approach,
male nurse and female nurse – and indeed transgendered nurse (of various types) –
are naturally treated symmetrically. We simply need to ensure that all communities
are captured in our schedules.
Terminological Ambiguity and Translation
Will individuals from dierent communities understand terms in similar ways? If not,
an individual from one community may have trouble navigating a KOS designed by
a member of another community. Szostak (2011) argued that the complex concepts
that are understood dierently across communities can generally be broken into
“basic concepts”: terms for which there is enough shared understanding across
individuals and communities for the purposes of classication. Globalization may
have various meanings, but exports (of goods) is understood similarly by most
people. A synthetic approach to classication allows subject headings to be
constructed by combining basic concepts. Such subject headings should be far less
ambiguous than those developed within enumerated classications. They then allow
members of all communities equivalent access to the KOS.
Basic concepts are likely also far easier to translate across languages than complex
concepts. The lesser degree of ambiguity in the original language should facilitate
the identication of a very similar term in other languages. Moreover, basic
concepts tend to represent things that we perceive in similar ways in the world
around us.
Ease of Access

Some individuals or communities may nd existing KOSs di$cult to navigate,
limiting their access to information. Recent decisions by some public libraries to
move away from library classications toward the BISAC employed in bookstores
re#ect a sense that many people nd existing library classications bewildering.
One challenge here is the biases and ambiguity addressed above. But a greater
challenge for many is simply not understanding how to identify an appropriate
subject heading to search. A synthetic approach to classication potentially allows
users to more readily identify appropriate subject headings. Rather than needing to
gure out how the KOS deals with male nurses, the user simply combines “male”
and “nurse” in their search query. I have in recent research argued that we can
make search even easier by following common grammatical rules in our structuring
of synthetic subject headings. A user search query that employs standard
grammatical construction is then readily translated into a relevant subject heading.
Di"erences in Perspective
Dierent communities or individuals may approach topics from a dierent
perspective. If a KOS re#ects one perspective, works from other perspectives may
be misclassied and hard to nd. One useful approach is to classify works by
authorial perspective. Since authorial perspective is multidimensional, a synthetic
approach to classication is best suited to classifying this: authors may dier with
respect to rhetorical strategies, ideology, membership in various communities, and
other ways (see Szostak 2015).
As noted above, a synthetic approach also reduces the scope for bias within the
subject classication itself. An enumerated classication may assume that certain
concepts naturally belong together – such as female and nurse – whereas a
synthetic approach seeks to link unidimensional concepts.
Note that synthetic approaches to both subject classication and classication of
authorial perspective aid users both when they seek works emanating from only
one community or perspective and when they actively seek works from multiple
communities or perspectives.
The Structure of Classications
Hope Olsen (2007) famously argued that a hierarchical approach to classication
may re#ect a male perspective. Women may be more likely to see the world in
terms of non-hierarchical relationships. A synthetic approach is grounded in a belief
that authors and users should potentially be able to combine any set of concepts as
they see t. Classication systems that pursue a synthetic approach to developing
subject headings still have to organize the concepts to be synthesized
hierarchically, but these hierarchies can be much #atter than those within
enumerated classication. The Basic Concepts Classication, for example, only
rarely has more than three or four levels of hierarchy.
Hospitality
A KOS designed by members of one community may exclude concepts deemed
important by members of other communities. The hospitality of a KOS – the ability

to add new terms – is thus an important consideration here. It is not always clear
where to place a new term within the multi-level hierarchies of complex terms that
characterize enumerated classications. Within a synthetic approach, new terms
can usually be created through a new synthesis of existing terms. When a new basic
concept must be added to a KOS, this is easier in #at and logical hierarchies: One
need not search multiple levels and wonder what the principles guiding the
hierarchy are.
Domains
One approach to dealing with terminological ambiguity is to develop domain-
specic KOSs. The obvious danger is that users then have di$culty nding
information in multiple domains. This could limit both interdisciplinary and cross-
community understanding. A synthetic approach to comprehensive (“universal”)
classication utilizing basic concepts should facilitate any community’s access to its
own literature without limiting cross-community understanding.
Opportunities for Crowdsourcing
Crowdsourcing provides an opportunity for members of one community to suggest
changes to a KOS designed by a member of another community. It thus provides a
potentially powerful response to the imposition of a KOS with undesirable
characteristics. Crowdsourcing is likely to be easier and more successful if the
principles guiding the KOS are transparent. A synthetic approach to classication
that employs basic concepts embedded in #at logical hierarchies should be more
easy to comprehend and amend than detailed enumerated classications. It would
be interesting to explore this hypothesis empirically.
KOSs as a Form of Advocacy
The approach outlined above seeks to make a KOS as inoensive as possible to as
many communities as possible. Though the KOS itself does not seek to advocate for
any one community, the precision of synthetic subject strings potentially allows
works of advocacy to be more readily found by users (e.g. (ghting)(discrimination)
(against)(community X)). Classication of authorial perspective would further
enhance this capability.
Concluding Remarks
This brief paper has purposely chosen breadth of coverage over depth of analysis. It
has shown that a synthetic approach to classication can potentially alleviate a
variety of challenges that KOSs may present to members of particular communities.
The author thus recommends the synthetic approach as likely the best way of
alleviating the adverse eects of KOSs on diverse communities.
References:
Olson, Hope A. (2007). "How We Construct Subjects: A Feminist Analysis". Library
Trends. 56 (2): 509–541.

Szostak, Rick. (2015). “Classifying Authorial Perspective” Knowledge Organization 42:7, 499-
507.
Szostak, Rick. (2011). “Complex Concepts into Basic Concepts” Journal of the American
Society for Information Science & Tech. 62:11, 2247-65.