From Atomic Elements to Fantastical Machines: The “Concept” in International Classification

The concept—an idea, a notion—is accepted as the core entity in knowledge organization (KO). The founder of the science of KO, Ingetraut Dahlberg, defined a concept repeatedly over time. Concepts have been well-discussed in the literature of both KO and information and have even been described as elementary particles in a theory of knowledge interaction. But an interesting question is what did the concept mean to these original thinkers in the nascent KO a century or more ago? An earlier series of papers about the evolution of the concept in information science based on the discourse of the concept in American Documentation led irrevocably to the notion of the concept as an element that could be isolated for analysis alongside frequent references to fantastical machines. This short paper describes an ongoing research project to undertake the same level of discourse analysis in the foremost evolutionary journal of KO, International Classification . A simple narrative of the occurrence of the “concept” in IC over the course of its run shows evolving definitions but also reveals usage of the notion of the concept as core (or atomic ) element.

In 1978 she had defined the concept as "a unit of knowledge" where "knowledge may be regarded as the totality of propositions about this world, existing-in documents or in the heads of persons" and "a concept is a knowledge unit, comprising verifiable statements about a selected item of reference, represented in a verbal form (143).It is clear that the establishment of knowledge organization on concept theory relied on a common understanding of the idea of a "concept" promulgated by Dahlberg in her own writing but also in the journal she edited.
Concepts, of course, have elsewhere been well-discussed in the literature of both KO and information science (see for example Hjørland 2009;Szostak 2011).Indeed, concepts have been described as elementary particles in a theory of knowledge interaction (Smiraglia and van den Heuvel 2013;Smiraglia, van den Heuvel and Dousa 2011).
But an interesting question for those of us approaching the first quarter mark of the 21 st century is what did the concept mean to these original thinkers in the nascent KO a century or more ago?In Smiraglia (2014;2018), a series of papers about the evolution of the concept in information science was based on specific analysis of the discourse of the concept in the foremost journal American Documentation.The analysis was broad-ranging, but the discourse analysis led irrevocably to the notion of the concept as an element that could be isolated for analysis.Frequent references to fantastical machines littered the writings of early 20 th century information scientists, including references to the "fantastical knowledge engine" in Gulliver's Travels (Smiraglia 2018, 128), to Cross's "Universal Index" (Duncan 2022, 211 ff.), and to Otlet's Grinder (Smiraglia and van den Heuvel 2013, 363).classification and the "ordering of knowledge" for "interdisciplinary common understanding" about the methods of classification.Specifically, the programme was (2): on an international scale … 1) to draw attention to the existence of a new, autonomous field of knowledge whose purpose it is to bring about order, whose objects are concepts and concept systems, and whose methods may be regarded as consisting in the construction of concept systems and the relating of concepts from such systems to elements of reality and vice versa" as well as "2) to provide … a common platform" and "3) to unite the specialists from the various fields …to bring about a common basis for their efforts."Thus, from the beginning, the critical role of the concept as core entity was acknowledged.
The founding of ISKO (International Society for Knowledge Organization), probably the pinnacle of Dahlberg's attempts to formulate a science of the ordering of knowledge around the concept as core entity, took place in July 1989.This was announced in an unsigned editorial in v. 16, no. 2 (71): Yes, on this very day of July22nd, 1989, a group of German and non-German members of the German Society for Classification founded the International Society for Knowledge Organisation (ISKO) at the Intercity Restaurant of Frankfurt's central Railway Station.
Beginning in that issue IC carried the following notice of its link to the new ISKO on its masthead: As of August 1989, International Classification has become the official organ of the International Society For Knowledge Organisation (ISKO) (founded on July 22, 1989) and is included for every ISKO member.
The first issue of IC carried seven articles, shown in Table 1.Eighteen years later, the final issue of IC before the journal evolved into Knowledge Organization, also carried seven articles (during the complete run, most issues carried three or four articles), shown in Table 2. Thus, in Figure 1 we see that in article titles the most prominent cluster is "indexing language" which is proximate to "search and retrieval," concept and concepts anchor a distinct cluster, and in the distance is a cluster for social sciences and thesauri.
Richard P. Smiraglia. 2023.From Atomic Elements to Fantastical Machines: The "Concept" in International Classification.NASKO, Vol. 9. pp.32-42.Similarly, the abstracts are dominated by a cluster representing indexing languages.The other largest prominent cluster is social sciences integrated thesaurus, which is somewhat distant indicating the two are distinct topics.Note that "social science" also occurs in conjunction with "comprehension and communication" at some distance, also distant ar clusters for relevance and concept systems.We see, then, that the concept is a core entity in the programme of the newly founded ISKO and in Dahlberg's journal.But, the most prominent research reported in the journal is that concerning indexing languages and thesauri.

Narrative analysis
Using the WordStat KWIC feature we are able to weave a simple narrative of the occurrence of the "concept," "concepts," and "conceptual" in IC over the course of its run.For this presentation we have ignored all casual usage and focused instead on articles in which the theory of the concept is either developed explicitly or moved forward in some manner.To save space (and meet the word requirements for a short paper a table of article citations and quotations from the relevant abstracts was constructed as a figure (Table 4)."A concept is regarded as the common element of both classification system and thesauri.Reality and knowledge are not represented by words or terms but by the meanings "behind" these tokens.A concept of, say, an object, a property of an object, a process, etc. is derived from verbal statements on these as subjects and may therefore be defined as the whole of true and possible predicates that can be collected on a given subject.It is from these predicates that the characteristics of the corresponding concepts can be derived.Discussion of the concepts "documentary informing" and "informeme" … the "informeme-theory" of Diemer is supplemented by the concepts "significance" and "efficiency of information" … Referring to the creative aspect in the use of language (Humboldt and Chomsky), it is shown how a description based on the principles of concept-synthesis may be successfully applied for the representation of informemes.Discussion of the concepts "documentary informing" and "informeme" … the "informeme-theory" of Diemer is supplemented by the concepts "significance" and "efficiency of information" … Referring to the creative aspect in the use of language (Humboldt and Chomsky), it is shown how a description based on the principles of conceptsynthesis may be successfully applied for the representation of informemes.(Dahlberg, Ingetraut. 1975."The Terminology of Subjectfields." 2, no.1:31-37) So far terminological work has been mainly directed towards defining very special concepts … it is intended to assemble their definitions in a dictionary and to build a general concept-system … Some application-possibilities for a general concept-system (e.g. a broad system of ordering) are given.(Dahlberg, Ingetraut. 1976."Classification Theory, Yesterday and Today" 3 no.2: 85-90) After clarification of the elements of classification systems as well as of the basis of concept relationships it is possible to apply a number of principles in the evaluation of existing systems as well as in the construction of new ones and by this achieving relatively predictable and repeatable results.(Dahlberg, Ingetraut. 1978."A Referent-Oriented, Analytical Concept Theory for INTERCEPT.5, no.2: 91-92.)"The concept theory presented, meant to serve as a basis for conceptual analyses of all terminological efforts, implies that every concept has a referent (be this a set of objects, a single object, an activity, a fact, a topic, etc.) about which verifiable statements determining the properties and relationships of the referent in question can be made.The totality of all the verifiable and necessary statements on a referent may be summarized and/or synthesized by a term which will then represent a concept in any communication process.A concept is thus regarded as a knowledge unit, and the statements about its referent are found to be the knowledge elements, also known as the characteristics, of the given concept.The possibility of thus determining the characteristics of concepts permits the analysis, construction, reconstruction, correlation, categorization and definition of concepts as well as the formation and control of adequate terms and the construction and comparison of concept systems."(Dahlberg, Wolfgang. 1979."Towards a Geometry of Basic Concepts."6, no.2: 83-84.) "The essential pillars of man's thinking and knowing are the basic concepts which structure all his knowledge so far attained and his future knowledge.It is, however, only with great difficulty that these basic concepts can be made accessible for everyone through conceptual clarification, definition and classification."(Wille, Rudolf. 1984. "Line Diagrams of Hierarchical Concept Systems." 11, no. 2:77-86.)The possibility of visualizing conceptual relationships by graphic representation of hierarchies … •is demonstrated here using a set-theoretical model for hierarchical concept systems.For a fixed context it is provable that the concepts in generic relationship (subconcept -superconcept) form -as a mathematical structure -a complete lattice called the "concept lattice" of the context.Methods and results of order and lattice theory can thereby be used for concept analysis.… building blocks of modified line diagrams may be understood as scales in the sense of a conceptual measurement theory.(Riggs, Fred. 1979."A New Paradigm for Social Science Terminology" 6, no.3: 150-58.) Terminology is concerned with the naming of concepts, and hence both with the meanings of words and with the need to signify concepts unambiguously.(Dahlberg, Ingetraut. 1982 Artificial intelligence research … besides the surface structure of facts there is a deep structure of concepts which provide access to the entire system.This means that terminological principles have to be observed.(Wilfried Lex, Wilfred. 1987."A Representation of Concepts for their Computerization."14, no.3: 127-32.) A lattice theoretical description of concept hierarchies is developed using for attributes the terms "given", "negated", "open" and "impossible" as the truthvalues of a four-valued logic.Similar to the theory of B. Ganter and R. Wille (6) so does this framework permit a precise representation of the usual interdependences in a field of related concepts -such as superconcept, subconcept, contrary concepts etc. … we have here a tool to deal with concepts and their interrelations on a computer, which may be of importance for some applications in artificial intelligence: automatic classification, information retrieval, data bases, expert systems, automatic theorem proving and machine translation.(Alexiev, Boyan. 1992 the issue of determining the proper tertium comparationis (TC) for contrastive equivalent terminological units from two languages.… Hence the aim of contrastive analysis is reduced to assessing quantitatively and qualitatively the various linguistic means used in the two languages for reflecting the characteristics (genus and differentia) of the same concept.
(Gilreath, Charles T. 1992 "Harmonization of Terminology -An Overview of Principles." 19, Harmonization is the process in which diverse positions are largely reconciled and assimilated into a single unified position.In terminology work there are four basic things needing to be harmonized: 1. concepts, 2. concept systems, 3. definitions, and 4. terms.These are briefly discussed, along with the related notions of extension and archetype (intension).
The narrative is as follows: In the first issue of IC Dahlberg lays out clearly her theory of the concept and of how concept systems (KOSs as we know them now) might be formed.It is critical to understand the distinction between the concept, which is a meaning, and the "words" or "verbal statements" that represent it.The concept itself is the "whole of true and possible predicates" that represent it.Relationships in the system are formed among "common characteristics of different concepts."In the same issue, Soergel follows the same path presenting suggested empirical methods for "determination of terms and concepts and of relationships between and among them."In 1975 Kashyap shows implementation of this understanding of concepts for education.Haendler puts forward "principles of concept-synthesis" for the development of informemes; he quotes Dahlberg alongside Chomsky, Diemer, Humboldt and Wittgenstein.
Dahlberg presents an implementation for concept-system generation (1975) and shows how incorporation of the "concept" in classification theory allows for replication, a key element of a science (1976).
In 1978 Dahlberg reiterates her concept theory and demonstrates its utility for the generation of theory-based concept systems.Notice we move from the concept itself now to the "knowledge unit" and that "statements about its referent are … the knowledge elements" or "characteristics of the given concept."A year later Wolfgang Dahlberg supports the general concept theory and, citing primarily I. Dahlberg but also Diemer creates a "geometry" of basic concepts.It is a bold attempt to denote another empirical approach to concept identification.In the same year Riggs shows usage of the concept theory and incorporation in relationship to terminology.
In 1982 Dahlberg introduces International Coding Classfication (ICC) using her concept system; she also introduces her "systematifier." In 1984 Wille advances the theory of the concept using a set-theoretical model to develop a "concept lattice" mathematical structure of the concept with its context.In 1985 Nedobity makes reference to early AI research, and relates the "deep structure of concepts" to the need for harmonization with terminological principles.In 1987 Lex draws together and demonstrates the use of the "lattice" theory introduced by Wille, and relates it to its potential use for computer applications, including AI.
In 1992 Alexiev moves the theory of the concept along theoretically into "contrastive terminology" to allow "contrastive equivalent terminological units from two langues."and Glireath describes harmonization of concepts and concept systems in terminology work.

Discussion: Definition, Dissemination, Replication
We find definition of a "concept," of course, but also revealing usage of the notion of the concept as core (or atomic) element.It is clear that the journal IC played a critical role in the evolution of Dahlberg's concept theory over time.IC was not just a vehicle for dissemination but also for evolution of the theoretical construct, which over the course of the 18 years proved useful for the construction of classifications, but also moved into information retrieval, thesaurus construction, automatic indexing, and most notably terminology and artificial intelligence.It is clear from the beginning that the role of empirical science was on the minds of these authors who exert processes of replication as proof of their theoretical points of view.As noted, this narrative avoids the regular occurrence of the term "concept" in regular usage, which accounts for the apparent gap in the narrative between 1985 and 1992, during which time it is clear the definition has settled into common use in the domain.
Next steps: Fantistical machines?IC also contained advertisements and notices of future meetings as well as summaries of meetings that had taken place.One such note appeared in 1990 (17, no.2: 76) under the headline "Meeting on Concept Relationships."According to the report, 74 persons attended the meeting hosted by Rudolf Wille in Darmstadt to receive and discuss 18 papers, although "there were only a few concerned expressly with concept relationships in a more or less philosophicallogical, psychological and linguistic way."Nonetheless, it shows the growth of the notion of the concept as a core atomic element of a theory of knowledge organization.
The next step for this research is to construct a discourse analysis around the papers referenced above.Elements of the discourse already emerge as the atomic role of the concept in a theory of KO, and the critical requirement that such a theory appear to be "scientific," primarily by making concepts and their structures replicable.We are particularly intrigued by the informeme, the concept-latttice, the systematifier, the geometric structure of concepts, and the deep structure of concepts.2
Common characteristics in different concepts lead to relations between concepts, which relations in turn are factors for the formation of concept systems.Development of a framework for the description and classification of statistical and graph-theoretical methods for the determination of terms and concepts and of relationships between and among them.Discussion of the problem of terms versus concepts in this context … methods … to construct global classificatory structures.