edited by Birger Hjørland and Claudio Gnoli


Integrative Levels Classification (ILC)

by Claudio Gnoli

Table of contents:
1. Introduction
2. History
3. The metaphysics of levels
4. Main classes
5. Subdivisions
6. Free combination of themes
7. Facets
8. Applications
9. Reception
10. Discussion

The Integrative Levels Classification (ILC) is a freely-faceted knowledge organization system. Instead of disciplines, as is usual in bibliographic classifications, it directly lists phenomena. The order of main classes and subclasses is based on the theory of levels of reality, particularly its formulation by Nicolai Hartmann and by James K. Feibleman. This approach is a continuation of research carried out by the Classification Research Group in the 1960s. ILC classes can be combined (1) by free relationships, where a base theme can be followed by particular themes, or (2) by free facets, or (3) by special facets exclusive to a class. Original features of ILC include deictics and such fundamental categories as form, disorder and perspective. The system has been tested in a variety of small databases in different domains and has arisen a certain interest among classification theorists.

[top of entry]

1. Introduction

The Integrative Levels Classification (ILC) is a → knowledge organization system (KOS) developed since 2004. It is meant as a general system covering all areas of knowledge. Although it can be ascribed to the tradition of → facet analysis, that has historically developed for libraries and bibliographies, ILC is conceived to be applied to any kind of documents, including museum specimens, websites or other collections of knowledge resources. While expressing and ordering concepts by a → notation like in → library classifications, it also includes synonyms and some associative relations (particularly dependence) like in → thesauri and ontologies.

Another main difference from classical faceted classifications, and an original feature of ILC, is that it lists phenomena rather than → disciplines. This is considered as an implementation of its ontological approach to → knowledge organization (KO) (Gnoli 2011b). Knowledge has ontological components, concerning the nature of the known objects, as well as epistemological components, concerning the ways objects are known. Although both these dimensions are relevant, the organization of knowledge can give priority to one or the other (Kleineberg 2020). Gnoli (2016, 406) lists the ontological and the epistemological as two basic dimensions in knowledge organization, also followed by a bibliographical, a bibliothecal/archival/museological, a cognitive and a sociological dimension (see section 7 for their expression in ILC).

Many disciplinary systems, like the Dewey Decimal Classification (DDC), primarily organize knowledge according to an epistemological perspective: knowledge is first divided e.g. by knowing ability into the “arts of reason”, including philosophy, theology, sociology, pure and applied sciences; the “arts of imagination”, including art and literature; and the “arts of memory”, including geography and history. Other systems, like the Bliss Bibliographic Classification (BC), can be described as more ontological, as they list main classes according to some principle concerning their objects, such as increasing specificity: philosophy, physics, chemistry, biology, psychology, sociology... Still, the same phenomenon, like “marriage”, can be found in these systems under various disciplines, depending on the perspective under which it is considered – e.g. marriage in ethics, in religion, in sociology, in history, etc. (Mills and Broughton 1977, 36-37; Slavic 2007).

In a classification based on phenomena, on the other hand, a phenomenon has a place of unique definition of its own, and what is scattered are disciplines rather than phenomena: while history of marriage is listed under rituals, history of republics is listed under polities. As classifications create linear orders of concepts, but each concept is related to several others, no classification can avoid to scatter some related concepts in a way or another. While disciplinary classifications prefer to scatter phenomena, phenomenon-based classifications prefer to scatter disciplines. Which one is the better option is an old question in KO (see section 2 and section 9), and the whole ILC project can be seen as a test of the alternative phenomenon-based approach (Hong 2005; Gnoli 2016). Disciplinary classifications are considered to better reflect the perspectives by which knowledge is produced and is expected to be searched: users start by deciding whether they are interested in, say, religion or sociology, before browsing the corresponding class. On the other hand, they may miss such connections as those between marriage in religion and marriage in sociology, that could suggest more innovative, interdisciplinary research directions (Szostak et al. 2016).

[top of entry]

2. History

General KOSs listing phenomena rather than disciplines were not uncommon in the past centuries, like in the cases of Wilkins' → Characteristica Universalis or Roget's → Thesaurus. On the other hand, modern library classifications as developed by Cutter, Dewey, Otlet, Bliss and → Ranganathan are usually based on disciplinary main classes. A modern library classification based on phenomena is → James Duff Brown's Subject Classification, published in 1906, which is considered as an interesting example of alternative approaches in KO (Beghtol 2004).

The idea of basing a classification on phenomena (as well as their properties and activities) was explored again during the 1960s by the Classification Research Group (1969), which used a NATO grant to draft a new general system, involving CRG members → D. J. Foskett, B. Kyle, H. Tomlinson, → E. Coates and D. Austin among others. Unfortunately this project was never completed, although it informed the later development of Austin's PRECIS subject indexing system and of the second edition of (disciplinary-based) Bliss Bibliographic Classification (BC2). To sort phenomena according to a principle independent from disciplines, the CRG looked at the theories of → integrative levels and of → general systems. The former (discussed in the next section) provides a way to arrange phenomena according to an order of increasing organization. A draft of the new system (Classification Research Group 1969, 127-128) listed the following main classes of entities:

A  general systems
B  phenomena and energy
C  matter
D  mineral systems
E  life support systems
G  astronomical universe
H  Earth as an environment
J  atmosphere
K  liquid layers
L  land forms
M  geo-centred living systems
N  viruses
P  organisms sharing characteristics of plants and animals
Q  plants
R  animals
S  man

General systems theory had been proposed in the 1950s by Ludwig von Bertalanffy, Kenneth Boulding and others, though it is anticipated in 1773 in d'Holbach's Système social (Bunge 2003, 41). The theory considers any phenomenon as a system composed of parts (also called “sublevels” in Austin's terminology) connected in a dynamic structure, which provides a basis for identifying facets of phenomena (Gnoli 2017b; see section 7).

Around year 2000, academic librarian Claudio T. Gnoli was also inspired by a theory of levels as a possible basis for classification: this was the version developed by philosopher Nicolai Hartmann (1940; 1952; see Poli 2016) and reported in Lorenz's Behind the mirror (1977). Gnoli thus got in contact with philosopher Roberto Poli, an expert in Hartmann's thought, and discussed with other colleagues. In the meantime he realized that a very similar idea had already been investigated in CRG's works, and found that such research was worth resuming. An output of this period is Gnoli and Poli (2004). A first draft was called Naturalistic Classification, as an application of naturalism (Gnoli 2004) meant as what we now call the ontological approach to KO. However, Lorena Zuccolo found that this name would erroneously suggest that the system could only be applied to the natural sciences, while the alternative name of Integrative Level Classification was more clear. “Level” was later changed to “Levels” for linguistic reasons after Rick Szostak's recommendation.

Figure 1: Right to left, Roberto Poli, Claudio Gnoli and Carlo Scognamiglio at a workshop on levels of reality, part of the International ISKO Conference, Rome, February 2010

The project has subsequently involved collaboration with other researchers, including Enzo Cesanelli, Hong Mei, Philippe Cousson, Tom Pullman, Mela Bosch, Rick Szostak, Rodrigo de Santis, Marcin Trzmielewski and Daniele P. Morelli. They have developed some subclasses or facets in more detail and tested the system by applying it to various little bibliographical databases (section 8). An important application involving Ziyoung Park of Hansung University in Seoul is to BARTOC, the online KOS directory edited by Andreas Ledl. An interdisciplinary approach to classification was advocated by various speakers at the Spanish ISKO Conference held in León in 2007, including María López-Huertas, Bosch, Gnoli and Szostak (also the author of a similar, contemporary project, the → Basic Concepts Classification): their ideas and later comments have been synthesized in the León Manifesto (ISKO Italia 2007).

As ILC is a newly developing system, its concepts and the notation for them are still relatively unstable. This makes it especially important to publish established schedules that can be referenced in a precise way. A first stable edition (ILC1) consisting of 7052 classes and facets was frozen in 2011, but development and experimentation have then continued, and a second edition (ILC2) consisting of 10,845 classes and facets has been published online in 2019. Changes introduced in ILC2 are described by Park et al. (2020). Also in 2019, long-term collaboration with Ceri Binding and Douglas Tudhope of University of South Wales Hypermedia Research Unit led to the publication of ILC2 as linked data in SKOS format (Gnoli et al. 2011; Binding et al. 2020). All these versions are freely and fully available online at the website (http://www.iskoi.org/ilc/) of the project, a working group of the Italian chapter of ISKO of which Cesanelli, Cheti, Gnoli and Park are members.

[top of entry]

3. The metaphysics of levels

The theory of levels of reality claims that the real world is structured in a series of levels of increasing organization. Each level depends on lower levels for its existence, but also has novel emergent properties that are not possessed by the lower ones. For example, living organisms need chemical substances as their material constituents, and there cannot be organisms without chemical substances; at the same time, chemical substances have not such properties as metabolism, growth or death, that only emerge at the level of organisms. In KO terms, dependence of a level on another is a kind of associative relationship, that can be recorded as classification data along with the classical hierarchical relationships: organisms depend on chemical substances, and plants are a type of organisms (De Santis and Gnoli 2016).

Among the levels that are identified by most authors are matter, life, mind and society, usually in this order (although Poli believes that mind and society co-evolve). Finer subdivisions have also been identified, for example matter is subdivided into particles, atoms, molecules and macroscopic bodies, or life is subdivided into cells, organisms, populations and ecosystems. Many different authors have proposed their own lists of levels, as reported in Kleineberg (2017, Appendix A).

Thinkers that have referred to this kind of stratified structure of the world belong to very different traditions (Juarrero and Rubino 2008; Gnoli 2017a, 41-43). Indeed, they include idealists like Hegel, materialists like Engels, empiricists like Mill, positivists like Comte and Spencer, spiritualists like Bergson and Smuts and even Christians like Teilhard de Chardin. Views of levels are often, although not always, coupled with evolutionary views, assuming that higher levels have evolved from pre-existing lower ones. Evolution was especially popular after the works of Lamarck and Darwin and was taken as a general view in philosophy by Herbert Spencer. At the beginning of the 20th Century, a philosophical school of evolutionary emergentism was championed by Samuel Alexander, Conwy Lloyd Morgan and George Conger (Blitz 1992).

The term integrative levels was introduced by biochemist Joseph Needham (1943) and discussed in biology by Alex Novikoff (1945). Psychologist James K. Feibleman then formulated an explicit “theory of integrative levels” (Feibleman 1951; 1954), which was a reference to the Classification Research Group, especially through Douglas J. Foskett (1961). This anglophone tradition was inspired by a generally materialistic worldview, which is reflected in CRG's list of levels (section 2) that mainly covers the natural sciences.

However, both Lorenz in his epistemological treatise and Dahlberg (1978, 28-31) in her theory of classification also refer to Baltic German philosopher Nicolai Hartmann, originally a scholar of neo-Kantianism who developed a “new ontology” using the terminology of continental metaphysics (“ideal being”, “real being”, “spirit” etc.). Hartmann’s system, though acknowledging the results of the natural sciences as a sound reference, devotes much room to the highest “stratum” of spirit. This is analysed into the “personal spirit” of each individual human as part of a culture, the “objective spirit” shared in societies and the “objectivated spirit” in the technological and intellectual products of human activity. In this way, levels can cover all knowledge domains rather than just the natural sciences.

In classifications, the subjects of knowledge (be they disciplines or phenomena) are → divided into classes, these are further divided into subclasses and so on. Most often, subclasses are → types of the things in the parent class: the class of plants is divided into the types of ferns, mosses, conifers, flowering plants etc. Types are often identified on a morphological basis: conifers have cones and lack flowers, flowering plants have flowers, etc. On the other hand, in a → genealogical perspective, types can often be found to have derived from other, more primitive types: thus one can say that flowering plants have evolved from pre-existing plants at a later time than conifers, and list them after conifers on this basis. In this case, the genealogical criterion is coupled with the morphological one.

This is the way many scientific taxonomies, like those of organisms or of languages, are often constructed. ILC also tries to reflect the principles of morphology and genealogy in the sequence and subdivision of its classes. While this is relatively clear for some classes, like organisms, it is less so in others, like musical instruments, where information on the origin is not easily available or cannot be applied in an immediate way. Also, morphology and genealogy can occasionally suggest different groupings, and decision has to be made on which one is given priority. Gnoli (2006) discusses some cases.

Hjørland (2017b, → section 4.2c) identifies four basic approaches in KO: rationalism, empiricism, historicism and pragmatism. ILC can be described as rationalist and historicist. It is rationalist in that its most general concepts, like main classes and their subclasses of higher ranks, are identified and ordered in a top-down way rather than on the basis of the amount of literature produced on them, as in the principle → literary warrant that informs many library classification schemes. (This does not exclude that, as applications of ILC are developed, missing concepts or relationships are identified bottom-up from individual cases and contribute to improve new editions of the scheme.)

ILC can also be described as historicist, in that it organizes knowledge objects following the genealogical principle, starting with the most primitive phenomena (energy, atoms...) before those derived from them (organisms, societies...) through cosmic, biological and social evolution. This can be seen especially in the series of integrative levels on which its main classes are based, but also within specific classes: for example, communities are divided into citizens, families, clans, castes, classes, civil society, public institutions, peoples and civilizations, again listed in an order of increasing organization (sections 4-5). Notice that the historicist approach “may be applied to both the object and the subject in classification” (Hjørland 2017b, section 4.2c). In the case of ILC, only the former appears to be relevant, as phenomena are listed according to their own genealogy rather than the genealogy of their study (molecules precede organisms though having been studied later).

[top of entry]

4. Main classes

ILC2 has 25 main classes, represented by lowercase letters a to y (z was also used in ILC1):

a  forms
b  spacetime
c  branes
d  energy; wave-particles
e  atoms
f  molecules
g  continuum bodies
h  celestial bodies
i  rocks
j  land
k  genes
l  bacteria; prokaryotes
m  organisms (eukaryote)
n  populations
o  instincts
p  consciousness
q  language
r  rituals
s  communities
t  polities
u  enterprises
v  technologies
w  artefacts
x  artworks
y  knowledge

As Pullman has observed, the choice of 25 main classes is biased by the notational base adopted, but there is no reason why reality must have 25 levels rather than less or more of them. Indeed, ILC also acknowledges more general levels, that roughly correspond to Hartmann's strata, and express them as spans of main classes (connected by the symbol W):

aWc  information
dWj  matter
kWn  life
oWp  mind
qWy  culture

Ontologically, main classes a to y can be understood as subclasses of these five major levels, which are → telescoped in notation. On the other side, another notational device of ILC, the emptying digit z, allows to express as many sibling classes as needed beyond the number of 25. Thus, class czp “preons” is technically a sibling of a, b, c etc. listed between c and d, although being assigned a longer notation (as it covers phenomena that are still hypothetical and poorly known in contemporary science). Potentially, more main classes can be interpolated in the same way, in case future science will identify them as first-rank levels. (Pronouncing rules for ILC notation are provided in Gnoli 2018a.)

The sequence of main classes in ILC is not very different from that of other general classifications, such as BC2 or Coates' Broad System of Ordering (BSO), except that what is listed is phenomena rather than the disciplines that typically study them.

Class a covers logical and mathematical entities, that are thus seen in a platonist way as basically inherent to reality, rather than as products of the human reason. Class c “branes” is unusual as it refers to string theory as part of contemporary fundamental physics; as such it has the potential of being developed or possibly renamed according to future results of science. The subsequent classes list natural phenomena as is usual in many versions of levels theory.

Class q language is placed at the beginning of cultural phenomena, as it emerges from human minds in p and triggers the development of organized civilization as represented in the subsequent classes. Rituals (which include religions), communities, polities, enterprises and technologies are interconnected “modal aspects” of the social life, to use a term from Herman Dooyeweerd — another thinker that has been associated to levels theory by Poli. Their interdependencies mean that they could have also been listed in slightly different orders, for example placing economies and technologies before rituals and polities, although the chosen order attempts at following a sequence where more primitive phenomena common to all human societies precede those typical of advanced civilizations.

Between v “technologies” and w “artefacts” one moves from organizations of active humans (Hartmann’s objective spirit) to their products (his objectivated spirit). These consist in artefacts and “mentefacts” (Gnoli 2018b), which can have a history independent from that of their creators. Knowledge is the highest level and includes yi disciplines themselves, which can thus be listed like in traditional bibliographic classifications though within the whole of a classification of phenomena. The last subclass of knowledge in ILC2 are yx knowledge organization systems.

[top of entry]

5. Subdivisions

As subclasses are also expressed by lowercase letters, any class can be divided in up to 24 subclasses, or more by the emptying digit -z-. For example, q “language” is divided into the following twelve subclasses:

qe  phones; segments
qf  phonemes
qg  combined sounds; sequences
qh  syllables
qm  morphs
qo  words; lexical categories
qp  phrases; syntagmas
qq  clauses
qs  sentences
qt  texts; discourses
qv  languages
qw  language transmission systems

Letter -a is never used used for subclasses in ILC2, as it is reserved to the attributes (such as processes, properties and parts) that are typical of a class:

qac  language change; evolution
qae  errors; mistakes
qaq  communicative functions
qas  senses
qax  language formalization

Attributes can in turn be subdivided with further lowercase letters. Attributes are a sort of → facets, although in this form they are meant to express an attribute concept autonomously: for example, qae means linguistic errors as such, while to express e.g. texts containing errors one has to combine class qt with a facet proper (see section 7; Park et al. 2020).

A feature original of ILC are special classes and subclasses expressed by capital letters A to Z (Gnoli 2018a). These are also called deictics, that is concepts whose meaning is not fixed but changes according to the context. They are used for a variety of purposes. A to T are used for classes that have to be defined in every specific application, as a shorthand representation of concepts of frequent use in a particular collection (Gnoli 2011a). For example, links from the category in the colophon of each article of this encyclopedia to its systematic index use this notation:

A   Knowledge organization: general and historical issues
AD        Discipline and adjacent disciplines
AM        Movements and bodies
AR        Biographical articles
C   Core concepts in KO
CC         Theoretical concepts
CS         Specific document types, genres and media
K   Knowledge organizing systems (KOS)
KA        KOS general issues
KD        KOS kinds
KG        KOS, specific (general/universal)
KN        KOS, specific (domain specific)
KS        Standards and formats for representing data
P   Knowledge organizing processes (KOP)
R   Methods, approaches & philosophies
T   KO in different contexts and applications

-U is the typical subclass: for example, the Sun is just a star among billions others, so should be some deep subclass of hl “stars”, but as it is of special interest to all users of the classification, it is hlU, the typically-discussed star. U as a main class is the typical subject overall, that is persons. V (entirety), W (conjunction) and Z (anaphora) are used for technical purposes to identify places in the schedules that are connected with others in specific combinations. X means “something”, so that e.g. qtX means “some texts”. Y means actually existing individual members of a class, so that qtY means an individual texts. For these and other details of ILC syntax see the webpage “How ILC works” (http://www.iskoi.org/ilc/how.php).

[top of entry]

6. Free combination of themes

ILC is a synthetic classification, which means that a knowledge resource can be assigned either a simple class or a combination of classes. The simplest way to combine classes is freely listing them separated by a blank space, e.g. hl r “stars [in some relationship with] rituals”.

Even in this simple case, one has to choose which of these themes is cited before the other(s). Following Alberto Cheti’s theory of subject analysis (in turn inspired by such verbal indexing systems as Austin’s PRECIS), for every knowledge resource a base theme can be identified, possibly associated with other particular themes (Cheti 1996). Thus, hl r implies that stars are the base theme and rituals are only discussed as for their relation to stars. If rituals are the base theme instead, the classmark must be written r hl.

If more than one particular theme has to be expressed, particular themes in turn have to be listed in some order. Gnoli (2011b), agreeing with previous considerations by Coates and Mills, suggests that themes of the same relevance should be cited in the inverted order of integrative levels, e.g. k “genes”, mqUH “horses” and vk “breeding” will be combined in the inverted sequence “breeding: horses: genes” — unless either horses or genes are focused as the base theme in the document, in which case they will be promoted to the leading position.

The notion of theme is imported into KO from text linguistics. A related notion is that of rheme, that is, the new information that is stated in the document concerning the theme. It is usually assumed that the → subject of a document corresponds to its theme(s) (Hjørland 2017a, → section 3.6). However, in some cases rheme can also be relevant in indexing. A rheme, like in “wolf diet is affected by cervid abundance” as distinguished from “wolf diet (as) affected by cervid abundance”, can be expressed in ILC by deictic Y after a facet indicator (Gnoli 2018a, 46). Treatment of themes and rhemes is another original feature of ILC as compared to other classification schemes.

[top of entry]

7. Facets

Another way of combining concepts in ILC is by facets. While the nature of the relationship is not specified in a free combination, it can be specified by facets: for example, “rituals, influenced by stars” can be expressed as r3hl. ILC is said to be freely faceted, as such facets can connect any two concepts from whatever part of the schedules, rather than only concepts within a particular discipline as it usually happens in BC2 or in → Colon Classification (CC) (in CC, a functional equivalent of free facets are phase relationships).

ILC facets are expressed by digits that follow a pattern of → fundamental categories, similar to Ranganathan’s PMEST. The ten categories of ILC facets are:

0  as for perspective
1  at time
2  in place
3  by agent
4  disturbed by disorder
5  with transformation
6  having property
7  with part
8  as form
9  of kind

Several of these categories, like part, property, transformation (that includes processes and activities), agent, place and time, are similar to other faceted systems. Form, disorder and perspective are more original, and are discussed in Gnoli (2008a; 2017b).

Categories can be combined by using two or more digits: 37 means a part (7) of an agent (3), that is a concurring factor. Multi-digit facets that begin by 9 have a special meaning limited to the class where they occur, much as in classical faceted classifications: hl99 means stars of some given spectral type, with spectral type obviously being only a facet of stars.

A numeral facet indicator is followed by one among the possible values (foci) of the facet: hl99 will be followed by literal notation for some specific spectral type. The foci for a facet that includes a 9 (either at the beginning or at the end as in common facets) have to be taken from another part of the schedules — often from the attributes of the same class. The corresponding notation, which is written between square brackets in ILC schedules, can be omitted to give a shortened faceted combination:

hl99 [hlat]  stars, of spectral type

hlat  spectral types
hlatf     F; yellow white
hlatg     G; yellow

hl99g  stars, of spectral type G

Finally, facets beginning with 0 express the perspective in which the phenomena are considered, or such other dimensions (section 1) as a document or a collection where knowledge of these phenomena is contained. For example, hl07yise means “stars, as studied in physical chemistry”. In this way, although phenomena are always the primary dimension, disciplines or other perspective facets (methods, theories etc.) can also be expressed (cp. Kleineberg 2020).

[top of entry]

8. Applications

The first tests with ILC were done on the catalogue of a small documentation centre, the Friuli Venezia Giulia Regional Laboratory of Environmental Education (LAREA) by Zuccolo and Gnoli; practice was also done with the Dandelion Bibliography of Facet Analysis by Hong (Gnoli and Hong 2006), with a gateway to Internet Resources for Chemical Research edited by Caterina Barazia and with news posts in the domain of communication and media indexed by Cesanelli (2008). Free combinations of general and local (deictic) ILC classes are applied to a bibliography of local culture and history in the website Where the Apennine Begins.

A more substantial application of detailed, freely faceted ILC classmarks was done with bibliographic records of the BioAcoustic Reference Database (BARD) set in collaboration with the Interdisciplinary Center of Bioacoustics and Environmental Research (CIBRA) at the University of Pavia (Gnoli et al. 2008). Bioacoustics is an interesting case of an interdisciplinary domain, connecting physics, zoology, ecology, management and technology; some hierarchies of such ILC subclasses as whales and sound types were developed in depth for this project. A similar, updated application is Traditional Europe: a selected videography indexed by ILC, where freely faceted ILC3 classmarks are used to organize online videos showing folk celebrations of various regions (Gnoli 2024).

Szostak et al. (2016) discuss the advantages of an interdisciplinary, phenomenon-based classification as compared to a traditional disciplinary one. A first comparison between the two kinds was made at the Land Ecology Department Library, University of Pavia, by assigning ILC classes to a small sample of documents already provided with Dewey Decimal Classification (DDC) classmarks (Szostak et al. 2016, 104-106). More recently, both DDC and ILC are used to index KOSs recorded in the Basel Register of Thesauri, Ontologies & Classifications (BARTOC). Ledl and Gnoli (2017, 114-5) observe that

Combinations of classes seem to be adequate to express the subject matters covered by KOSs both with DDC and with ILC. The degree of accuracy (co-extension) obtained by the combined classmarks looks similar in the two cases. [...] Clearly, what is different is the meaning of classes, as these express disciplines and subdisciplines in the case of DDC, but phenomena and their types or facets in the case of ILC. Users can perceive this especially in the display of scheme trees for browsing: while with DDC one has to start with 10 disciplinary macro-classes, such as philosophy, religion, social sciences etc., with ILC one is presented with 26 classes of phenomena sorted by integrative levels, such as molecules, rocks, cells, organisms, populations, civil society, etc. One can expect the effect to be cognitive, as in the latter case the classification scheme will guide users to the exploration of the universe of subjects by classes of phenomena rather than those of disciplines.

Gnoli et al. (2018, 659) also discuss the mapping of high-rank DDC and ILC classes, as connected to their application to BARTOC:

Another consideration is that some main classes of phenomena in ILC, such as d “elementary particles and waves” or l “prokaryote organisms”, are represented in DDC by several digits as they are only subclasses quite deep in the hierarchy; conversely, main DDC classes such as 800 “literature” or 410 “English language” are just subclasses (of x “arts” and qem “Germanic languages”) in ILC. Indeed, the top level classes are very different between the two classification schemes. A further point of interest is that some entries in ILC are mapped to more than one entry in DDC:
190, 140yy99m  modern Western philosophy
230/280rt  Christianity
400/410q  languages; idioms
510, 160a  forms; mathematical objects
660/680vt  industry
740, 760xg  drawings
800/890xl  literature
900, 940/990t91  historical periods

[top of entry]

9. Reception

The ILC project has arisen a certain interest among classification experts. Broughton’s (2015) textbook on classification mentions the ILC project as an attempt at phenomenon-based classification; the same author later acknowledges it as "another more complex and nuanced version of a faceted system that includes relationships among its facets" (Broughton 2023). Vickery (2008) applauds the ILC draft scheme, though suggesting a partially different approach that combines phenomena with human activities. Beghtol’s (2009) encyclopedia entry on “Classification theory” states that

the theory of integrative levels remained intriguing and became the basis for Dahlberg’s Information Coding Classification (ICC). More recently, the Italian chapter of ISKO […] has embarked on a series of investigations into the theory of integrative levels and has begun to develop the Integrative Levels Classification (ILC). These investigations appear to be fruitful new directions in research on the structure(s) of knowledge organization classification.

On a similar line, Movahedian and Cheshmeh Sohrabi (2017) and Moraes (2018) discuss integrative levels or their application in ILC as one relevant theoretical approach. Szostak discusses ILC and his BCC as promising examples of interdisciplinary classifications in several papers and especially in Szostak et al. (2016). Machado (2017) discusses ILC’s ontological approach and compares it with Barry Smith’s Basic Formal Ontology, which claims to do without concepts though also being ontologically based (Machado et al. 2020).

Michael Kleineberg reintroduces an epistemological approach when applying integrative levels as an ordering principle to “levels of knowing” instead of the known phenomena; he proposes to classify perspectives in this way in ILC (Kleineberg 2018; 2020). For a collective discussion, see Gnoli et al. (2013). However, Slavic (2007), the current editor-in-chief of the disciplinary Universal Decimal Classification, is critical on the abandonment of perspective or “aspect” as the primary ordering principle in classification:

Document indexing aims to group similar contents in the way books are likely to be sought and it is assumed that, for instance, a nutritionist looking for “fish cooking recipes” will not necessarily be interested in books on fishing, growing fish or the sport of fishing. Collecting all books about “fish” on a single library shelf does not seem to make sense in practice. Hence, although in the past there were libraries organized according to a classification of phenomena (for example, Subject Classification of Brown) this principle of organization is recognized as ill-suited for library users [Ranganathan 1961]. Thus, nowadays most of the widely used documentary classifications are disciplinary, i.e. aspect classifications.

To this, Gnoli replies that

even in a phenomenon classification, like the ILC experimental draft, there will be a variety of distinct classes including fish at different integrative levels, such as organisms, aquaculture, food, sport, etc., so not all documents about fish will be listed together, though they will be retrievable through synthetic notation. In the terms you describe, UDC and ILC are not going towards directions so different, apart for the structure of their main classes. (ISKO Italia 2007, comment 2008.01.08)

On a technical side, Green (2008) observes that dependence relationships of the levels theory “possess some characteristics of paradigmatic relationships […] but lack others”; Will (2013, section 3.4) discusses non-disciplinary facets in ILC; and Bianchini and Dousa (2024) introduce a paper (Gnoli 2024) acknowledging that it

demonstrates that one of Ranganathan's most cherished goals—the design of a "freely faceted classification"—can best be realised in practice by moving from a discipline-based classification, such as Ranganathan's Colon Classification, to a phenomenon-based classification, of which the most advanced—and most fully functional—example created to date is Gnoli's own Integrative Levels Classification (ILC).

The distinction between phenomena and perspectives has been applied by Frank (2018) in the NeDiMAH Methods Ontology which “distinguishes between research goals and activities, research techniques, and research objects”. Kos et al. (2018) consider the interdisciplinary approach of ILC as a promising way to address KO of bioethics, a domain where perspective is a characteristically ambiguous component.

Montoya (2022, 136-137) describes the approach of ILC and related research as “a static understanding of external phenomena (consistent enough, in practice, that the organization of all knowledge can perpetually be conformed to this schematic)” which could hardly follow the changing concepts of biological taxonomy. His qualification of “reductionist approaches” seems to be an odd reference to emergentism, which is just the opposite. He believes that such approaches “overlook the socially situated, culturally defined unfolding of our knowledge production practices”, although the latter are not denied in ILC theoretical premises.

[top of entry]

10. Discussion

As shown in the last section, the ILC project has stimulated a certain amount of debate in KO literature, by resuming experimentation on non-disciplinary classification that was attempted already by Brown and the CRG. At this stage, the debate is mainly theoretical, as we still lack any large corpus of documents indexed with ILC, and the system is still under development. On the other hand, the tests performed on a variety of small corpora of different nature (books, papers, KOSs, websites) and in different domains (local culture, bioacoustics, knowledge organization, general) already provide several hints. Phenomenon-based classification by ILC has been shown to be possible and to produce orderings that are different from those of disciplinary classifications, while still having several features in common with them (combination of concepts, need of cross references between different classes, e.g. organisms, aquaculture and food).

Such established classification systems as DDC, UDC and Library of Congress Classification (LCC) are applied already to millions bibliographic records across the world; this makes it unlikely that any new system can compete with them in the short term. For this reason, developing systems that are not radically different from the existing ones (like Ten Pillars of Knowledge (Zins 2011) or Step by Step Classification, to only mention two of recent appearance) is unlikely to produce any major impact. Of greater interest may be systems that are innovative in some aspects, like ILC claims to be, as they can be relevant for the study of performance and effects of such aspects. Room for their application can be open, among other kinds of collection, in museums and in websites, especially because they need to index objects or non-academic topics, which can be poorly served by the disciplinary approach.

[top of entry]


Beghtol, Clare. 2004. “Exploring new approaches to the organization of knowledge: the Subject Classification of James Duff Brown”. Library Trends 52, no. 4: 702-718.

Beghtol, Clare. 2009. “Classification theory”. In Encyclopedia of library and information sciences, 3rd ed. London: Taylor and Francis, 1045-1060.

Bianchini, Carlo and Thomas M. Dousa. 2024. “Editorial Introduction: Ranganathan and his Legacy: Past, Present, Future”. Annals of Library and Information Studies 71, no. 1: 3-8.

Binding, Ceri, Claudio Gnoli, Gabriele Merli, Marcin Trzmielewski and Douglas Tudhope. 2020. “Integrative Levels Classification as a networked KOS: a SKOS representation of ILC2”. In Knowledge organization at the interface: proceedings of Sixteenth International ISKO Conference, Aalborg, July 2020. Baden-Baden: Ergon, in preparation.

Blitz, David. 1992. Emergent evolution: qualitative novelty and the levels of reality. Dordrecht, Boston: Reidel.

Broughton, Vanda. 2015. Essential classification, 2nd ed. London: Facet.

Broughton, Vanda. 2023. "Facet analysis: the evolution of an idea". Cataloging and Classification Quarterly 61, https://www.tandfonline.com/doi/full/10.1080/01639374.2023.2196291.

Bunge, Mario. 2003. Emergence and convergence: qualitative novelty and the unity of knowledge. Toronto (etc.): University of Toronto Press.

Cesanelli, Enzo. 2008. Classificare il dominio della comunicazione: analisi di un corpus di fenomeni secondo la teoria dei livelli di integrazione. Degree thesis, University of Trieste, corso di laurea specialistica in Scienze della comunicazione, a.a. 2007-2008. In E-LIS, http://eprints.rclis.org/12280/ .

Cheti, Alberto. 1996. Manuale ipertestuale di analisi concettuale, realisator Serena Spinelli, University of Bologna, http://www2.sba.unibo.it/miac/.

Classification Research Group. 1969. Classification and information control. London: Library Association.

Dahlberg, Ingetraut. 1978. Ontical structures and universal classification. Bangalore: SRELS.

De Santis, Rodrigo and Claudio Gnoli. 2016. “Expressing dependence relationships in the Integrative Levels Classification using OWL”. In Knowledge Organization for a sustainable world: challenges and perspectives for cultural, scientific, and technological sharing in a connected society: proceedings of the Fourteenth International ISKO Conference, Rio de Janeiro, 27-29 September 2016, eds. José Augusto Chaves Guimarães, Suellen Oliveira Milani and Vera Dodebei. Würzburg: Ergon, 368-375.

Feibleman, James K. 1951. “The graded series of integrative levels”. In Ontology. Baltimore: Johns Hopkins, 332-356.

Feibleman, James K. 1954. “Theory of integrative levels”. British Journal for the Philosophy of Science 5, no. 17: 59-66.

Foskett, Douglas J. 1961. “Classification and integrative levels”. In The Sayers memorial volume, eds. D. J. Foskett and B. I. Palmer for the Classification Research Group. London: Library Association. Republished in Theory of subject analysis: a sourcebook, eds. Lois Mai Chan, Phyllis A. Richmond and Elaine Svenonius. Littleton: Libraries Unlimited, 1985, pp. 210-220.

Frank, Ingo. 2018. “Interdisciplinary research organization: superimposing linked library data, linked research information and research data with interdisciplinary knowledge organization”. In Challenges and opportunities for knowledge organization in the digital age: proceedings of the Fifteenth International ISKO Conference, 9-11 July 2018, Porto, Portugal, eds. Fernanda Ribeiro and Maria Elisa Cerveira. Baden-Baden: Ergon, 948-950.

Gnoli, Claudio. 2004. “Naturalism vs. pragmatism in knowledge organization”. In Knowledge organization and the global information society: proceedings of the Eighth International ISKO conference, London, July 13-16, 2004, ed. Ia C. McIlwaine. Würzburg: Ergon, 263-268.

Gnoli, Claudio. 2006 “Phylogenetic classification”. Knowledge Organization 33, no. 3: 138-152.

Gnoli, Claudio. 2008. “Categories and facets in integrative levels”. Axiomathes 18, no. 2: 177-192.

Gnoli, Claudio. 2011a. “Animals belonging to the emperor: enabling viewpoint warrant in classification”. In Subject access: preparing for the future, eds. Patrice Landry, Leda Bultrini, Edward T. O’Neill and Sandra K. Roe. Berlin: De Gruyter, 91-100.

Gnoli, Claudio. 2011b. “Ontological foundations in knowledge organization: the theory of integrative levels applied in citation order”. Scire 17, no. 1: 29-34.

Gnoli, Claudio. 2016. “Classifying phenomena, part 1: Dimensions”. Knowledge Organization 43, no. 6: 403-415.

Gnoli, Claudio. 2017a. “Classifying phenomena, part 2: Types and levels”. Knowledge Organization 44, no. 1: 37-54.

Gnoli, Claudio. 2017b. “Classifying phenomena, part 3: Facets”. In Dimensions of knowledge: facets for knowledge organization, eds. Richard Smiraglia and Hur-Li Lee. Würzburg: Ergon, 55-67.

Gnoli, Claudio. 2018a. “Classifying phenomena, part 4: Themes and rhemes”. Knowledge Organization, 45, no. 1: 43-53.

Gnoli, Claudio. 2018b. “Mentefacts as a missing level in theory of information science”. Journal of Documentation 74, no. 6: 1226-1242.

Gnoli, Claudio. 2024. “Ranganathan's principles and a fully "freely faceted" classification”. Annals of Library and Information Studies 71, no. 1: 113-120.

Gnoli, Claudio and Hong Mei. 2006. “Freely faceted classification for Web-based information retrieval”. The New Review of Hypermedia and Multimedia 12, no. 1: 63-81.

Gnoli, Claudio and Roberto Poli. 2004. “Levels of reality and levels of representation”. Knowledge Organization 31, no. 3: 151-160.

Gnoli, Claudio, Gabriele Merli, Gianni Pavan, Elisabetta Bernuzzi and Marco Priano. 2008. “Freely faceted classification for a Web-based bibliographic archive: the BioAcoustic Reference Database”. In Wissensspeicher in digitalen Räumen: Nachhaltigkeit, Verfügbarkeit, semantische Interoperabilität: Proceedings der 11. Tagung der Deutschen Sektion der Internationalen Gesellschaft für Wissenorganisation, Konstanz, 20. bis 22. Februar 2008, eds. Jörn Sieglerschmidt and H.-Peter Ohly. Würzburg: Ergon, 124-134.

Gnoli, Claudio, Tom Pullmann, Philippe Cousson, Gabriele Merli and Rick Szostak. 2011. “Representing the structural elements of a freely faceted classification”. In Classification and ontology: formal approaches and access to knowledge: proceedings of the International UDC Seminar, 19-20 September 2011, The Hague, eds. Aida Slavic and Edgardo Civallero. Würzburg: Ergon, 193-206.

Gnoli, Claudio, Michael Kleineberg, Riccardo Ridi and Rick Szostak. 2013. The blind knowledge organizers and the elephant: working notes on Kleineberg's levels of knowing. Version 7.1, 12 January 2016. http://www.iskoi.org/ilc/elephant.php. Includes Kleineberg's “The blind elephant”.

Gnoli, Claudio, Andreas Ledl, Ziyoung Park and Marcin Trzmielewski. 2018. “Phenomenon-based vs. disciplinary classification: possibilities for evaluating and for mapping”. In Challenges and opportunities for knowledge organization in the digital age: proceedings of the Fifteenth International ISKO Conference, 9-11 July 2018, Porto, Portugal, eds. Fernanda Ribeiro and Maria Elisa Cerveira. Baden-Baden: Ergon, 653-662.

Green, Rebecca. 2008. “Relationships in knowledge organization”. Knowledge Organization 35, no. 2-3: 150-159.

Hartmann, Nicolai. 1940. Die Aufbau der realen Welt: Grundriss der allgemeinen Kategorienlehre. Berlin: De Gruyter.

Hartmann, Nicolai. 1952 (orig. ed. 1942). New ways of ontology. Westport: Greenwood Press.

Hjørland, Birger. 2017a. “Subject (of documents)”. Knowledge Organization 44, no. 1: 55-64. Also available in ISKO Encyclopedia of Knowledge Organization, eds. Birger Hjørland and Claudio Gnoli, http://www.isko.org/cyclo/subject.

Hjørland, Birger. 2017b. “Classification”. Knowledge Organization 44, no. 2: 97-128. Also available in ISKO Encyclopedia of Knowledge Organization, eds. Birger Hjørland and Claudio Gnoli, http://www.isko.org/cyclo/classification.

Hong Mei [non-inverted Chinese name]. 2005. “A phenomenon approach to faceted classification”. Paper presented at the 53th Conference of the Japan Society of Library and Information Science (JSLIS), Keio University campus, 22-23 October 2005, http://www.iskoi.org/ilc/phenomenon.php.

ISKO Italia. 2007. The León Manifesto. http://www.iskoi.org/ilc/leon.php.

Juarrero, Alicia and Carl A. Rubino eds. 2008. Emergence, complexity and self-organization: precursors and prototypes. Goodyear: ISCE.

Kleineberg, Michael. 2017. “Integrative levels”. Knowledge Organization 44, no. 5: 349-379. Also available in ISKO Encyclopedia of Knowledge Organization, eds. Birger Hjørland and Claudio Gnoli, https://www.isko.org/cyclo/integrative_levels.

Kleineberg, Michael. 2018. “Reconstructionism: a comparative method for viewpoint analysis and indexing using the example of Kohlberg's moral stages”. In Challenges and opportunities for knowledge organization in the digital age: proceedings of the Fifteenth International ISKO Conference, 9-11 July 2018, Porto, Portugal, eds. Fernanda Ribeiro and Maria Elisa Cerveira. Baden-Baden: Ergon, 400-408.

Kleineberg, Michael. 2020. “Classifying perspectives: expressing levels of knowing in the Integrative Levels Classification”. In Knowledge organization at the interface: proceedings of Sixteenth International ISKO Conference, Aalborg, July 2020. Baden-Baden: Ergon, in preparation.

Kos, Denis, Sonia Špiranec and Ante Covic. 2018. “Mapping perspectival ambiguity in bioethics: revisiting the viewpoint warrant”. In Challenges and opportunities for knowledge organization in the digital age: proceedings of the Fifteenth International ISKO Conference, 9-11 July 2018, Porto, Portugal, eds. Fernanda Ribeiro and Maria Elisa Cerveira. Baden-Baden: Ergon, 959-961.

Ledl, Andreas and Claudio Gnoli. 2017. “Indexing KOSs in BARTOC by a disciplinary and a phenomenon-based classification: preliminary considerations”. In Faceted classification today: theory, technology and end users: proceedings of the International UDC Seminar, 14-15 September 2017, London, United Kingdom, edited by Aida Slavic and Claudio Gnoli. Würzburg: Ergon, 109-117.

Lorenz, Konrad. 1976 (orig. ed. 1973). Behind the mirror: a search for a natural history of human knowledge. London: Methuen.

Machado, Luís Miguel Oliveira. 2019. “Entre a organização do que é conhecido e o conhecimento da sua organização: um estudo comparativo entre as abordagens ontológicas da Integrative Levels Classification (ILC) e da Basic Formal Ontology (BFO): projeto de tese de doutoramento em ciência da informação”. Presentation at EDiCiC 2019, Barcelona, 9-11 November 2019.

Machado, Luís Miguel Oliveira, Graça Simões, Renato Rocha Souza and Claudio Gnoli. 2020. “Can an ontologically-oriented KO do without concepts?”. In Knowledge organization at the interface: proceedings of Sixteenth International ISKO Conference, Aalborg, July 2020. Baden-Baden: Ergon, in preparation.

Mills, J. and Vanda Broughton. 1977. Bliss Bibliographic Classification, 2nd edition. Introduction and auxiliary tables. London: Butterworths.

Montoya, Robert D. 2022. Power of position: classification and the biodiversity sciences. MIT Press. Freely available at https://doi.org/10.7551/mitpress/12245.001.0001.

Moraes, Rosana Portugal Tavares. 2018. A representação do conhecimento em espaços multidimensionais: uma investigação de princípios teóricos a partir de autores fundacionais da organização do conhecimento. PhD thesis, Universidade Federal Fluminense, doutorado em Ciência da informação. https://app.uff.br/riuff/bitstream/1/10205/1/....

Movahedian, G. and M. Cheshmeh Sohrabi. 2017. “Theoritical [sic] approaches in developments of library classification systems”. National Studies on Librarianship and Information Organization [Iran] 27, no. 4, http://nastinfo.nlai.ir/article_1471_en.html.

Needham, Joseph. 1943. “Integrative levels: a revaluation of the idea of progress”. In Time: the refreshing river: essays and addresses, 1932-1942. London: Allen and Unwin, pp. 233-272.

Novikoff, Alex B. 1945. “The concept of integrative levels and biology”. Science, 101: 209-215.

Park, Ziyoung, Claudio Gnoli and Daniele P. Morelli. 2020. “The second edition of the Integrative Levels Classification: evolution of a KOS”. Paper presented at DCMI-NKOS Workshop, Seoul, 2019. Journal of Data and Information Science, in prep.

Poli, Roberto. 2016. “Nicolai Hartmann”, revised version. In Stanford encyclopedia of philosophy, https://plato.stanford.edu/entries/nicolai-hartmann/.

Ranganathan. S.R. 1961. “Library classification on the march”. In The Sayers memorial volume, eds. D. J. Foskett and B. I. Palmer for the Classification Research Group. London: Library Association, 72-95.

Slavic, Aida. 2007. “On the nature and typology of documentary classifications and their use in a networked environment”. El profesional de la información 16, no. 6: 580-589.

Szostak, Rick, Claudio Gnoli and María López-Huertas. 2016. Interdisciplinary knowledge organization. Cham: Springer.

Vickery, Brian C. 2008. “The structure of subject classifications for document retrieval”. In Brian Vickery at home [personal website now closed], http://www.lucis.me.uk/classification.htm . Republished by ISKO Italia, http://www.iskoi.org/ilc/vickery.php.

Zins, Chaim. 2011. Ten Pillars of Knowledge. Jerusalem: Knowledge Mapping Research.

[top of entry]


Visited Hit Counter by Digits times.

Version 1.0 published 2020-02-13
Version 1.1 published 2020-02-19: references to Gnoli 2018a, IEKO index and Zins 2011 added
Version 1.2 published 2023-04-13: reference to Broughton 2023 added
Version 1.3 published 2023-09-27: reference to Montoya 2022 added
Version 1.4 published 2024-03-19: reference to Bianchini and Dousa 2024 and Gnoli 2024 added, mention of TradEU.

Article category: KOS, specific (general/universal)

©2020 ISKO. All rights reserved.