Theoretical foundation of Knowledge Organization: Positivism versus pragmatism. Invited speech Sunda

1
Theoretical foundation of Knowledge Organization
Positivism versus pragmatism. Invited
speech Sunday Oct 28, 2007 VIII ENANCIB in
Salvador, Bahia, Brazil. Birger Hjørland
2
Introduction

• LIS discourses (Perspectives modern, Western)
• on users ("user oriented" and "cognitive
  approaches"),
• on technology ("systems oriented approaches"),
• on the library as institution ("the institutional
  approach") or
• on management perspectives (e.g., "information
  management")
• while a bibliographic perspective focusing on
  documents and information resources, their
  description, organization, mediation and use is
  almost absent, although this is what the field is
  all about.

3
The Bibliographic Paradigm

• The term bibliographic paradigm' has only been
  used negatively
• As a contrast to something better.
• As a part of "the systems-oriented perspective"
  (or "physical paradigm") in LIS
• As opposed to user-oriented paradigms.
• Why is this the case?

4
The Bibliographic Paradigm

• Traditionally library and information services
  have focused on sources and technology and in
  doing so have developed sophisticated systems for
  collecting, organizing and retrieving sources and
  have applied information technology to provide
  extensive access to vast sources of information.
  Kuhlthau, 1993.
• User education has, therefore, concentrated on
  manipulative skills.
• This bibliographic paradigm has underplayed the
  cognitive aspects of the information process that
  highlight understanding and meaning. (Henri
  Hay, 1994). .

5
The Bibliographic Paradigm

• Librarianship, documentation and information
  science are about documents, their selecting,
  organizing and retrieval in relation to
  satisfying the needs of the users.
• Perhaps, what is meant by Henri Hay is that it
  is not enough to regard documents and information
  resources isolated from the needs of the users.
• If, as Henri Hay wrote above, librarians are
  too isolated from a school's mainstream
  curriculum, the most important sources to teach
  may be unknown to the librarian, why they may
  fall back on certain sacred documents
  considered universally relevant.

6
The Bibliographic Paradigm

• It would be very wrong, I believe, if
  LIS-professionals give up the focus on
  documents and information resources because of
  the difficulties mentioned above.
• To focus on user studies or cognitive
  studies is not an alternative. However, it seems
  like this is, nevertheless, what often happens.
• According to Nahl (1996, 2003) a user-centered
  revolution entered our field around 1970. The
  term user-centered is however very ambiguous
  when used about approaches to LIS.
• Empirical studies of users can not in any way
  substitute for empirical studies of documents.

7
The Bibliographic Paradigm

• On the CoLIS6 conference Sanna Talja Jenna
  Hartel (2007) made some very interesting points
  about the so-called user oriented paradigm. They
  seriously challenged the traditional dualism
  between user-oriented and systems oriented
  approaches in LIS
• My own view is that whether LIS investigates
  "users" or "systems" the dominating understanding
  has been "positivist", ahistorical and
  decontextualized. Users have been considered as
  biological beings more often that as cultural and
  specialized beings.
• Hermeneutical, historical, sociological and
  critical perspectives may potentially be much
  more fruitful.

8
Positivism and pragmatism

• In my opinion the tendency within LIS has always
  been dominated by a view of knowledge, which may
  be labeled positivist. This is a difficult
  concept, but I understand it as a tendency (or an
  ideal) to rely on observations and logic only,
  disregarding context, values, interests,
  historical development and socio-cultural issues.
  
• Positivism as I understand it is also a
  tendency to neglect the reading of relevant
  literature. Background knowledge is a vague
  concept and the ideal of being well read is
  considered less important than to do empirical
  studies. Of course nobody will admit this, but it
  is often visible in actual behavior. (Such an
  ideal is of course a double paradox in a field
  devoted to libraries and information).

9
Positivism and pragmatism

• The alternative view may be termed pragmatic or
  critical. It also relies on observations and
  logic, but do not consider observations as given,
  but as a acts done by persons in given
  socio-cultural contexts. Observations are
  theory-laden and not just objective, mechanical
  recordings.
• Thus observations and logic must be made on the
  basis of contextual knowledge as well as based on
  values and on goals.
• As researchers it is important that we consider
  our own points of view in the context of other
  points of view. The information we study is
  also more or less a merge of different points of
  view.

10
Positivism and pragmatism

• The proper understanding of this view of
  knowledge is in my opinion extremely important,
  but not something you learn just because I
  briefly mention it.
• LIS professionals have to read texts in the
  theory of science by authors such as Popper,
  Kuhn, Habermas, Gadamer, feminist scholars and
  others, develop a consistent view and apply this
  view as your glasses no matter what intellectual
  activity you are involved with.
• Thus in my opinion this should be mandatory stuff
  in all LIS educational programs.

11
Positivism and pragmatism

• There is a natural connection between a
  bibliographical paradigm in LIS focusing on
  documents and literatures on the one side, and on
  the other side more hermeneutic, sociological and
  critical views on knowledge. This is indicated
  in the following quotes by Paling and McKenzie
• "Bibliography provides a compelling vantage from
  which to study the interconnection of
  classification, rhetoric, and the making of
  knowledge. Bibliography, and the related
  activities of classification and retrieval, bears
  a direct relationship to textual studies and
  rhetoric. . . ..

12
Positivism and pragmatism

• A striking similarity to problems raised in
  rhetoric and which spring from common concerns
  and intellectual sources is demonstrated around
  Gadamer's notion of intellectual horizon.
  Classification takes place within a horizon of
  material conditions and social constraints that
  are best viewed through a hermeneutic or
  deconstructive lens, termed the "classificatory
  horizon." (Paling, 2004).
• ". . . bibliography is the discipline that
  studies texts as recorded forms, and the
  processes of their transmission, including their
  production and reception . . . I define 'text' to
  include verbal, visual, oral, and numeric data .
  . ." . (McKenzie, 1999, 12).

13
Positivism and pragmatism

• Thus, what is needed in LIS is an approach which
  considers documents and information as well
  as systems, institutions and users as more or
  less structural coupled by social, cultural and
  subcultural and domain specific processes.
• This should also be an approach, which considers
  the functions of knowledge, information and
  documents, and which considers the interests,
  goals and values of all actors, including, of
  course the functions and values of both public
  libraries and scientific communication systems.
  

14
Knowledge Organization (KO)

• Narrow sense (KOP)
• knowledge organizing processes
• classification
• indexing
• labeling
• summarizing etc.
• Broad sense (KO)
• knowledge organization
• about different disciplines and discourses,
  languages, conceptualizations.
• KO theories in the narrow field are deeply
  dependent on influences from the broader field.

• knowledge organizing systems (KOS)
• Classification systems,
• Thesauri
• Ontologies
• Any form of controlled vocabulary
• related to information retrieval.

15
Knowledge Organization (KO)

• Recently I made some observations and comments on
  the UDC-classification system
• Hjørland (2007). "Arguments for 'the
  bibliographical paradigm'. Some thoughts inspired
  by the new English edition of the UDC"
  Information Research, 12(4) paper colis06.
  http//informationr.net/ir/12-4/colis/colis06.html
  

16
UDC History and status

• UDC may be considered a sacred document within
  librarianship, and related to "the bibliographic
  paradigm".
• First edition of UDC was published more than 100
  years ago (1905-1907).
• Based on the Dewey Decimal Classification, but it
  was expanded in order to serve as the organizing
  system for a planned world bibliography of all
  documents (including articles).

17
UDC History and status

• UDC has played an important role as a
  classification system in research libraries in
  many countries around the world. It is still very
  much used. Some nations use the system in their
  national bibliographies.
• Why are systems like the UDC not forming part of
  a strong bibliographic paradigm within LIS?

18
UDC in crisis

• A first kind of crisis
• Cranfield studies of the late 1950's, which found
  that UDC (along with other similar systems based
  on "human indexing") did not contribute to
  improve information retrieval in electronic
  databases.
• These studies are very important in the
  tradition labelled "information science" (with
  "information retrieval" as an important
  sub-discipline).
• This conclusion from the Cranfield studies may
  be only a statistical generalization that
  neglects some kinds of questions for which
  systems like UDC might be superior.

19
UDC in crisis

• It is said that the people responsible for
  UDC-classification in the Cranfield experiments
  felt that the experimental questions disqualified
  tasks for which the UDC might be superior.
  Alternative experiments were never conducted.
• Since then classification researchers have been
  rather invisible in, for example, bibliometric
  maps of LIS.
• We may thus say that the UDC, like classification
  in general, has never been important in the part
  or tradition of our field derived from
  "information science" (which has mostly been
  interested in "free text" retrieval as opposed to
  any form of "controlled vocabulary)."

20
UDC in crisis

• In spite of this first crisis, the practical use
  of the system has so far not declined !
• Another crisis in the 1980s related to the
  maintenance and further development of the
  system.
• Connected to a more general uncertainty in the
  library communities concerning the future role of
  knowledge organizing systems (KOS) such as the
  UDC.

21
UDC in crisis

• The middle of the 1980s was the heyday of
  artificial intelligence. Concepts such as
  "intelligent agents" for individualized
  information retrieval were often thought to make
  traditional KOS superfluous.
• Investment in the maintenance and development of
  KOS may have suffered without proper basis in
  research
• The mere suspicion that KOS systems were obsolete
  was strongly demotivating for further investment
  of time, energy and intellectual efforts in
  construction.

22
UDC in crisis

• Research itself may have suffered because many
  students and researchers within LIS did not
  engage themselves in specific contributions to
  the improvement of such KOS.
• Instead they engaged themselves in other kinds of
  studies, some of which may be productive in the
  development of alternative kinds of KOS. Other
  kinds of studies simply seem to have lost their
  relation and relevance to LIS.
• It has been somewhat depressing to follow how the
  concrete interests and contributions to
  classification of subject literatures have
  declined within LIS.

23
UDC in crisis

• In my CoLIS6 paper I mentioned missing concepts
  in the index of UDC-2005 and regard them as a
  symptom of a crisis within LIS. I shall not
  repeat these examples here.
• UDC represented once a dream of a cumulative
  project to map knowledge, providing a better
  overview of knowledge and possibility to identify
  just the knowledge we need for a particular
  purpose.
• It may have been based on some naïve premises
  (to be presented below), but still having an
  important kernel worth working for.

24
Theoretical progress in Knowledge Organization

• It is somewhat ironic that the most used tool for
  KO in libraries today is the DDC first published
  1876. More than hundred years of research and the
  development of other kinds of KOS has not
  resulted in making DDC obsolete. For example, the
  BC2 is generally considered theoretically more
  advanced, but is difficult to use in practice.
• Most of the English-language books are
  pre-classified with the DDC by the Library of
  Congress.
• It is, however, thought provoking that
  classification systems developed later and
  generally thought more advanced are not able to
  compete efficiently.

25
Theoretical progress in Knowledge Organization

• Systems like UDC, DDC and Bliss may all be
  criticized for their universalist assumptions
• While unitary documentary languages ensure a
  maximum of mutual understanding . . ., they do
  so by legitimizing a particular ideological and
  sociopolitical worldview, and by silencing other
  meanings, voices, and ways of knowing . . ..
  Unitary documentary languages embody a belief in
  the existence of a unified body of knowledge.
  They express a belief in the possibility to
  capture reality isomorphically in information,
  and presuppose a neutral ground from which to
  judge the truth-value of different theories.
• (Tuominen Talja Savolainen, 2003).

26
Theoretical progress in Knowledge Organization

• Any controlled vocabulary represents a
  prescriptive or normative KOS.
• Positivist view
• normative vocabularies represent neutral,
  objective solutions, that simply provide more
  efficient information systems.
• Pragmatic view
• any controlled vocabulary tends to favour some
  kinds of queries, while relatively making other
  kinds of queries more difficult to answer.

27
Theoretical progress in Knowledge Organization

• This understanding also applies to systems such
  as Google and systems based on relevance
  feedback!
• A search engine will always favor some kinds of
  questions and answers and relatively disfavor
  some kinds of interests and values. A core
  concern in, for example, Danish public libraries,
  should thus be From our perspective, which kinds
  of questions and answers should have priority? A
  search engine or any kind of KOS should
  support the goals, values and interests of the
  system for which they are designed. This is not
  just a question of technical efficiency.

28
Theoretical progress in Knowledge Organization

• Consider the use of statistical methods of
  measuring similarity between papers in order to
  retrieve information. There are many similarity
  measures and they do not provide the same
  results (cf. Schneider Borlund, 2007ab). How
  should a similarity measure be tested and
  selected?
• Objective, empirical data are not enough. We may
  apply such methods, interpret the outcome, modify
  the measures iteratively and in the end find some
  patterns that seem to be robust. In this process,
  however, we are using knowledge of what is
  important and meaningful

29
Theoretical progress in Knowledge Organization

• Let us consider a simple example of what a
  controlled vocabulary typically does
• Mills Ball (2007) mention that the concept of
  arts is ambiguous, it is used both about visual
  arts (or just paintings), or it used broadly
  about visual art, music, literature.
• A book titled French art could be just about
  paintings, or it could include music and
  literature as well. In order to make it possible
  to search separately for both kinds of books, the
  BC2 have different classes for each of these
  meanings of the term arts.

30
Theoretical progress in Knowledge Organization

• Apparently this is simple a neutral logical
  improvement and this example tends to justify
  what we termed the positivist view That a
  controlled vocabulary simply improves
  retrievability in a neutral way.
• The pragmatic view has to demonstrate that such
  kinds of logical improvements are not always
  desirable, that some queries benefit from them,
  but other kinds of queries may suffer, and that
  it is necessary in the design of controlled
  vocabularies to consider what kind of queries the
  systems should give priority to.

31
Theoretical progress in Knowledge Organization

• One may argue that the meaning of the word art
  is connected to theoretical views of art, which
  also implies cues on how to retrieve the
  literature that is relevant from a certain
  theoretical perspective Semantic relations are
  theory-dependent. In the words of Fast Leise
  Steckel (2002)
• A controlled vocabulary is a way to insert an
  interpretive layer of semantics between the term
  entered by the user and the underlying database
  to better represent the original intention of the
  terms of the user

32
Theoretical progress in Knowledge Organization

• The question then is From what perspective, with
  what kinds of justification, do LIS-professionals
  provide such an interpretative layer?
• One part of the answer might be that different
  groups use the word art in different ways. When
  literatures produced by those groups are merged,
  the words become homonymous.
• The information specialist, with an overview of
  these mixed meanings is in a position to make
  them univocal.

33
Theoretical progress in Knowledge Organization

• Another part of the answer might be that the
  pragmatic understanding seeks the meaning of
  words not in the past, but in the future, what
  can be accomplished by the speaker by preferring
  one meaning for another.
• Any library or database is a part of an
  organization with a given purpose (whether
  explicated or not) and this purpose is the key to
  the justification of such an interpretative layer
  as done by controlled vocabularies.

34
Theoretical progress in Knowledge Organization

• Controlled vocabularies have normally been
  developed to specific databases and/or
  collections. A given database and collection is a
  tool, that is designed to support certain tasks
  and functions.
• The pragmatic theory of knowledge seeks the
  criteria for selecting and describing informative
  objects in the goals that they are intended to
  support. The widespread ideology of objectivity
  and neutrality and universal solutions may be
  counterproductive in developing our field.

35
Theoretical progress in Knowledge Organization

• This insight, however, is just what makes the
  dream of a cumulative work like the UDC somewhat
  naïve
• Different purposes and interests in different
  social systems need different kinds of
  classifications.
• One reaction to this insight may be a skeptical
  attitude towards all kinds of controlled
  vocabularies.
• Another reaction has been a tendency to develop
  many specific information languages, which
  tends to make interoperability worse, not better.

36
Theoretical progress in Knowledge Organization

• Information specialists look at the paper in the
  context of the other papers in a given
  collection/database. If qualified, it is possible
  to add value, to add structure and semantic
  information to bibliographic records and to
  develop KOS that are supporting the activities
  done by the author producing the information.
• This activity is by principle a kind of
  meta-study of the domain, for example, a
  historian would describe the development of a
  field, for example, relating concepts to
  different theories and traditions within a field.

37
Theoretical progress in Knowledge Organization

• Let us consider literary history as an example.
  Such a history is always subjective, it is
  always reflecting its author, its time and a
  certain world-view. It may be, for example,
  traditional of feminist. Because of this it
  is considered valuable by many people.
• Such a work on literary history in reality
  classifies the single books and labels them in
  ways, which are not usually parts of the books
  themselves (e.g. by genres such as romanticism
  or magical realism). Such labels are useful for
  some information needs, although not for all.

38
Theoretical progress in Knowledge Organization

• A controlled vocabulary can do the same kind of
  job
• Provide conceptual access to documents not
  already accessible in this way (and tools such as
  histories and controlled vocabularies may serve
  each other, they are both instruments for the
  study of, e.g. fiction, as well as products of
  such studies).

39
Semantics

• Any kind of knowledge organizing system (KOS) is
• a selection of concepts with indication of,
  first of all, their semantic relations.
• I believe we should avoid reification and
  essentialism when we speak about, for example,
  thesauri.
• A thesaurus is a kind of KOS that may be further
  developed and in this process lose its identity
  compared to other kinds of KOS, or as I prefer
  semantic tools.

40
Semantics

• The essence of KO is thus
• To identify and define terms and concepts
• To provide information about those terms and
  concepts, essentially different kinds of semantic
  relations.
• In order to understand its own theoretical basis
  must KO develop proper theories on those issues.

41
Semantics

• In general I find KO informed by problematic
  semantic theories. They are too positivist and
  too little hermeneutic. We have lots of thesauri,
  but historical dictionaries (e.g. in the
  tradition of Begriffsgeschichte) are almost
  unknown in our field.

• Begriffsgeschichte is based on the view that
• the meaning of a term develops historically
• many meanings exist
• meanings are related to different theories and
  ideologies etc.
• Also concepts such as Harris concept of
  intralingual heterogeneity seems important.

42
Semantics

• Semantic relations, such as synonymy are not a
  question of neutrality or objectivity, but of
  goals and consequences. Let us consider an
  example
• Are library science and information science
  synonymous?
• Clearly some authors do consider them as
  synonyms, while other argue they are not. Only by
  developing an argument can this issue be solved.
  Until it is solved and some consensus is reached,
  the best thing would be to describe the different
  views (a la historical dictionaries).

43
Conclusion

• It is important to reconsider the bibliographic
  paradigm in LIS. Studies of literatures cannot
  be substituted by, for example, studies of users.
• Some of the criticisms raised against this view
  may be related to problematic philosophical
  premises The bibliographic paradigm does not
  necessarily imply a positivist description of
  documents, but may imply a consideration of what
  documents can do, and how LIS can support
  documents in doing important tasks, i.e. a
  critical and pragmatic perspective.

44

• Thank you for your attention!

45
References

• Feinberg, M. (2007). "Hidden bias to responsible
  bias an approach to information systems based on
  Haraway's situated knowledges" Information
  Research, 12(4) paper colis07. Available at
• http//InformationR.net/ir/12-4/colis/colis07.html
  

46
References

• Hjørland, Birger (2007). Arguments for 'the
  bibliographical paradigm'. Some thoughts inspired
  by the new English edition of the UDC.
  Information Research, 12(4) paper colis06.
  http//informationr.net/ir/12-4/colis/colis06.html

47
References

• Schneider, Jesper W. Borlund, Pia (2007a).
  Matrix Comparison, Part 1 Motivation and
  Important Issues for Measuring the Resemblance
  Between Proximity Measures or Ordination Results.
  Journal of the American Society for Information
  Science and Technology, 58(11), 1586-1595.
• Schneider, Jesper W. Borlund, Pia (2007b).
  Matrix Comparison, Part 2 Measuring the
  resemblance between proximity measures or
  ordination results by use of the mantel and
  procrustes statistics. Journal of the American
  Society for Information Science and Technology,
  58(11), 1596-1609.

48
References

• Talja, S,. Hartel, J. (2007). "Revisiting the
  user-centred turn in information science
  research an intellectual history perspective"
  Information Research, 12(4) paper colis04.
  Available at http//InformationR.net/ir/12-4/colis
  /colis04.html