Faculdade de Letras da Universidade do Porto - OCS, 15th INTERNATIONAL ISKO CONFERENCE

Font Size: 
Claudio Gnoli, Andreas Ledl, Ziyoung Park, Marcin Trzmielewski

Last modified: 2018-02-19



An important trend in recent research in knowledge organization (KO) is the critic of academic disciplines as the primary division principle of classification schemes (Beghtol 2010, p. 1056 ; Broughton 2015; Gnoli 2016). There have been many claims that disciplinary classifications may act as an obstacle to interdisciplinary research (Szostak et al. 2016), may leave out such important applications as services for citizens (Bonner 1982) particularly in e-government, or serious leisure, or vendors' catalogues of products, and anyway may not be the best foundation in ontological terms for general systems. Experimental phenomenon-based classification systems have been developed since the times of the Classification Research Group (Austin 1969), by which the Integrative Levels Classification was inspired, having been developed for a decade and currently offering a general scheme with more than 7,000 classes (Gnoli 2017).

On the other hand, the classification schemes used in most libraries and their online catalogues still are the classical disciplinary ones, like the Dewey Decimal Classification (DDC), the Universal Decimal Classification (UDC) or the Library of Congress Classification (LCC). This situation suggests that a comparative evaluation of phenomenon-based vs. disciplinary indexing and an examination of possible ways to create cross-references and mappings between the two kinds of systems may have important implications for the future of classification. This paper explores such possibilities by especially considering DDC as a case of a disciplinary system and ILC as a case of a phenomenon-based system.



Preliminary experiments of comparison between DDC and ILC have been performed by indexing small samples of items by both systems and considering how the resulting orderings and retrieval possibilities are different. A first sample consisted of books in nature conservation (Szostak et al. 2016). Experiments have recently continued by applying both DDC and ILC to items recorded in the Basel Register of Thesauri, Ontologies and Classifications (BARTOC) maintained by the Basel University Library (Ledl and Gnoli 2017). These preliminary results show that, while both DDC and ILC may be used in a post-coordinate way to assign each item a plurality of classes, the general ordering of items and the display of search results are very different if phenomena are considered as primary grouping criterion instead of disciplines.

Indeed, DDC as a disciplinary system scatters a given phenomenon, such as “water”, in a variety of different disciplinary classes (chemistry, hydrology, hydraulic engineering, architecture, …). This can easily be observed by checking a term for the phenomenon in the DDC Relative Index, where references are provided to its occurrence in the various classes. One of these classes can be considered prior to the others and work as the “interdisciplinary number” for that phenomenon, where interdisciplinary works on it (e.g. a book on “water” seen from many perspectives) should be classified. This reminds of Farradane’s notion of place of unique definition which is at the basis of phenomenon-based systems.


Exploring ways for comparing, mapping and linking

Recently, an editor of DDC has started considering “topics” as another relevant unit in this system complementary to its disciplinary classes: “As a knowledge organization system, Dewey is an analytico-synthetic classification system first and foremost organized by disciplines, in which topics and concepts are scattered throughout. What can be done with Dewey as a classification system (that is, because it groups topics into classes) that couldn’t be done if it were strictly topic/concept-based, like the typical thesaurus? […] Many of the topics within a class occur as Relative Index terms. Within WebDewey, these are formatted in Marc 21 for Authority data and include references. Can/should Relative Index terms be used (as supplements to subject headings or independently) in indexing?” (Green 2017, http://edug.pansoft.de/tiki-index.php?page=2017+meeting).

Although “topics” are not defined and are labelled by a term other than “phenomena”, it seems that the two units can work in similar ways, at least functionally. Thus Green’s suggestion opens to the possibility of using DDC in some phenomenon-like ways. For example, main classes of phenomena in ILC could be also represented by the DDC classes corresponding to their interdisciplinary number, so that the resulting arrangements can be compared:


c spacetime 333.79 energy

d energy 520 celestial objects

e atoms 530.1 spacetime

f molecules 531 continuum bodies

g continuum bodies 539.7 atoms

h celestial objects 541.22 molecules

i rocks 552 rocks


Topics can work as a better reference than disciplines in mapping DDC to different systems, including ILC itself or subject headings and thesauri, which usually index phenomena rather than disciplines. Indeed the experts who look for the most suitable DDC class to be recorded for mapping a term in Nuovo Soggettario, the Italian national subject heading system developed by the National Library of Florence, usually choose the DDC interdisciplinary number as listed in the Relative Index (Lucarelli pers. com.).

Topics also are an implicit basis for cross-references within the DDC as “see-also” references provided in WebDewey and as links established between classes in online applications. The latter kind of relationships is unfortunately not common in library OPACs, but has been successfully implemented in the SciGator web-based browsing interface (Lardera et al. in prep.). This application allows to navigate between DDC main classes and a selection of their subclasses before launching a search by DDC notation in the University of Pavia online catalogue. SciGator’s most original feature is query expansion using cross-references between DDC classes or their mapping with local shelving schemes. Phenomena are indeed an intuitive basis for the selection of these cross-references in SciGator: e.g., behind the links librarians creates between fluid dynamics in physics, hydrology in Earth sciences and hydraulic engineering in applied sciences stands the concept of water as a phenomenon. Such references have the potential of being more formalized and connected with topics in the Relative Index.



The example applications considered by us for indexing books (first test in Pavia), library catalogue items (SciGator) and KOS directory items (BARTOC) both by disciplinary DDC and by phenomenon-based ILC look promising. In particular, some notions implicit in existing disciplinary classifications, such as topics and interdisciplinary numbers in the Relative Index, may be better analized and formalized to provide brigdes towards phenomenon-based systems like ILC or Nuovo Soggettario. These experiences suggest that continued research in phenomenon-based classification is an important requisite both for the development of innovative classifications schemes and for more powerful applications of existing disciplinary systems.