, University of Wisconsin-Milwaukee (USA)
The impact of faceted analytical theory and of the implementation of “facets” have had major impact on the development of systems for knowledge organization and information retrieval, ranging from major general bibliographic classifications to ecommerce. The discourse of facet analytical theory as a research front—that is, the over-arching story line of the growth, concretization and shifting intension of the concept of the facet is represented in the literature produced by that research front. This research front often is associated with the Dorking conference in 1957, but of course has roots in knowledge organization that can be traced to the 19th century. Diverse meanings of the concept of facet range from “broad facet categories,” often used as the bases of bibliographic classifications, to “analytical categories,” often used in ecommerce. Outside of the domains of KO and information science usages range from geology to dentistry to philosophy and beyond. Insofar as facets represent dimensions of knowledge they also reveal cultural influences. How is it that such a concrete idea can lead to so many divergent implementations? The answer lies in the analysis of cultural synergy. Cultural synergy is the merging of perception and behavior that shapes knowledge within and among the intellectual nodes of the diverse domain of “facets.” This presentation begins with pointers to the discourse as they emerge from an informetric analysis of published research concerning facets and facet analytical theory, both within the knowledge organization and information science communities as well as across all other visibly active domains. We will explore the ways in which what we call classification theory has had effect much beyond our own domain. Faceted analytical theory is one 21st century realization of a multi-verse of cultural synergy, emerging from classification theory to provide impetus to improved information architecture at large.
, University College London (UK)
The Classification Research Group manifesto of 1955 proclaimed its members’ commitment to the techniques of facet analysis as a general methodology for organizational, indexing and retrieval systems. In the 1950s this was hardly the case, but sixty years later the influence of faceted classification can be seen in all kinds of representation and discovery tools, and goes far beyond the limits of the conventional bibliographic classification that many of the original CRG envisaged as their objective. However, the CRG’s purpose was not just to encourage the faceted approach to designing and constructing classifications, but to propose it as a fundamental theory of knowledge organization, at the core of the disciplines of library and information science. At the time faceted classification theory was in many respects poorly articulated; many of the elements of ‘classical’ facet analysis were yet to be properly identified and defined, and it would be the work of some years to arrive at a mature theory. Yet that rudimentary model would eventually provide a foundation for much modern information retrieval. What are the distinctive features of facet analysis that make it so compatible with current needs, particularly in a digital environment? Some of the truth resides in the integrated nature of the faceted model, its clear explication of categorization, order, and intra- and inter-facet relationships, which can be rolled out across different species of knowledge organization system. The logic of this structures is readily exploited in automated systems, and can in part be expressed by representation languages. The complexity of the fully faceted classification, while internally consistent, is, nevertheless, challenging to realise in the same way.
, University of Copenhagen (Denmark)
There has been a tendency within the community of knowledge organization (KO) to consider facet analysis the only approach to classification. Jack Mills, for example, wrote that he does not see faceted classification "as a particular kind of library classification but as the only viable form enabling the locating and relating of information to be optimally predictable". My own research findings, however, show that there are different “approaches", "paradigms" or theories to classification which remain relevant and that these theories of classification and KO basically corresponds to theories of knowledge. In my earlier research I argued that facet analytical theory should be viewed as a rationalist/logical approach (in contrast to empirical approaches, genealogical and hermeneutical approaches and pragmatic/critical approaches). I also observed that knowledge about the theory of knowledge can serve as an indicator of the relative strength and weakness of a given approach to classification, in this case facet analysis. In this talk I will revisit this problem again, and consider counterarguments to the argument that facet-analysis basically is a pragmatic approach and the argument that it represent eclecticism. Although it is well known that labels such as "rationalism", "empiricism", "historicism" and "pragmaticism" are polysemantic, it is possible to offer an understanding of these concepts, that would provide a much needed guidance in our field. The theory of classification needs to consider different perspectives.
, University of Washington (USA)
S. R. Ranganathan's conception of faceted classification was an amelioration to the problem of an "ever expanding universe of knowledge." It went a long way to solving many problems that strictly enumerative schemes created. However, there are many assumptions behind faceted classification and one of them is stability in semantics. This paper will explore this assumption in light of subject ontogeny research. While creating new facets and the ability to combine facets to create new classes is one way to accommodate change, I will point out where we must go beyond Ranganathan's designs in order for our faceted and analytico-synthetic schemes to retain their value over time.
, University of Pavia (Italy)
While occurring in many knowledge organization systems (KOS) of different types, the notion of facet has been used with different meanings and roles in time. Generally, it suggests the combination of several concepts concurring to specify the subject of a document. However, the syntax of such combinations varies considerably, as do the sources from which the possible values of a facet (foci) can be taken. This paper attempts a review of syntactical alternatives in faceted systems. Concepts can be simply juxtaposed without expressing the relationship between them (free combination); or the relationship can be expressed and link two concepts taken from any part of the scheme (freely faceted systems, phase relationships); or it can be characteristic of only a given basic class and allow to link it to a choice of other concepts according to a facet formula (classical faceted classification, special facets); or it can link any basic class to a choice of auxiliary concepts such as space, time or form (common facets). Foci, in turn, can be taken from any other part of the system, or typically from a certain part, or be defined in the context of the facet itself. When describing a KOS as "faceted", which now seems to be a fashionable attribute, the nature of such "facets" should be made explicit. Finally, the supposed "rationalistic" theoretical basis of facet analysis, as opposed to "empirical" or "pragmaticist", is briefly discussed.
, University of Arizona (USA)
A description is provided of basic faceted classification which involves combinations of foci across facets, where the foci within a facet are dependent (i.e. exclusive) and the foci across facets are independent (i.e. orthogonal). This is shown to be suitable for organizing the basic goods that Amazon, the online retailer, sells and for progressive filtering as a mode of search. However, on closer inspection, the Amazon case involves a sorted domain. This is problematical for basic faceted classification. Additionally, books from Amazon would typically carry subject classification which also is difficult for basic faceted classification. It does not support filtering as a mode of search. Then subject classification really requires relatively sophisticated linguistic and logical constructors and modifiers, such as adjectives, adverbs, functions, binary relations, and transitive verbs. These can be part of a synthetic subject classification scheme, but they pose a challenge for faceting.
, OCLC (USA)
Various fields, each with its own theories, techniques, and tools, are concerned with identifying and representing the conceptual structure of specific knowledge domains. This paper compares facet analysis, an analytic technique coming out of knowledge organization (especially as undertaken by members of the Classification Research Group [CRG]), with semantic frame analysis, an analytic technique coming out of lexical semantics (especially as undertaken by the developers of FrameNet). The investigation addresses three questions: (1) How do CRG-style facet analysis and semantic frame analysis characterize the conceptual structures that they identify? (2) How similar are the techniques they use? (3) How similar are the conceptual structures they produce? Facet analysis is concerned with the logical categories underlying the terminology of an entire field, while semantic frame analysis is concerned with the participant-and-prop structure manifest in sentences about a type of situation or event. When their scope of application is similar, as, for example, in the areas of the performing arts or education, the resulting facets and semantic frame elements often bear striking resemblance, without being the same: facets are more often expressed as semantic types, while frame elements are more often expressed as roles.
, University of Buffalo (USA)
Composionality (the idea that "The meaning of a complex expression is determined by its structure and the meanings of its constituents.") and entity-relationship modelling are intertwined structural principles underlying thought, language, and classification / knowledge representation / data modelling . Drawing on examples from many contexts, this paper illustrates common principles for representing and understanding reality, imagination, and conceptualization as they apply to thought, natural language, and systems designed for organizing and applying knowledge -- classification for organizing documents and document-like objects, knowledge representation for artificial intelligence, and data modelling for managing databases. Examples include the arrangement of the Greek alphabet, Chinese characters, sign language, frames and semantic networks as models for the organization of knowledge in the mind and in computer systems, faceted classification (including facets in the UDC), record structures and entity-relationship modelling (done properly) in databases. To make these ideas more concrete, the paper provides an entity-relationship model that represents the facet structure of the UDC. The examples demonstrate that the idea of facets -- if not known by this name -- has been around for a long, long time. Following the principle of compositionality and entity-relationship modelling through many contexts improves our understanding of faceted classification.
, University of Basel (Switzerland)
, University of Pavia (Italy)
This paper outlines the recently launched project of classifying top-ranked knowledge organization systems in the terminology registry BARTOC by the Integrative Levels Classification (ILC) and comparing the resulting organization of knowledge with that produced by Dewey Decimal Classification as applied to the same items. This is meant to provide a case study for evaluating phenomenon-based classification and comparing it to disciplinary classification. It addresses both technical aspects of importing ILC into Drupal CMS, and intellectual aspects of this subject indexing endeavour.
, DRTC, Indian Statistical Institute (India)
, University of Trento (Italy)
, DRTC, Indian Statistical Institute (India)
S.R. Ranganathan, is credited with developing analytic-synthetic faceted classification. He proposed the five ‘fundamental categories’ [P][M][E][S][T] which were deemed as necessary and sufficient to characterize all documents in a library. However, in the context of the web and knowledge management in general, resources are no longer limited to mere academic disciplines. We are in fact required to extend disciplines to a plethora of domains which can capture the immense variety of the world. While domains provide the context, the entities in a domain provide the conceptual infrastructure for classification which, therefore, should be entity-centric. Examples of entities are mind products, organizations, objects, e.g., physical books and people, or events. Entities can be either abstract and concrete; they are what the world is made up of. Within each domain, entities are described with a set of properties. Hence, there is a logical progression from books in the library to entities on the web. As there is a transition from subject to domain, we need to come up with ‘fundamental categories’ that will be necessary and sufficient to characterize the web resources. The collaborative work between the University of Trento and DRTC resulted in a faceted entity-centric approach, a new faceted knowledge representation model. The proposed methodology, called DERA (for Domain, Entity, Relation, Attribute) exploits Ranganathan's analytic-synthetic classification principled approach and exploits it towards building and reusing knowledge. At the same time, DERA is amenable to logical formalization. Any DERA statement can be directly translated in Description logics (DL) being, in practice, just a syntactic variation of a DL axiom. This allows for the full automation of reasoning, including, e.g., search, classification, generalization, subsumption and so on. The presentation aims at showcasing the mapping of Ranganthan’s fundamental categories to the faceted entity-centric model - DERA.
, UK Data Service, University of Essex (UK)
, UK Data Service, University of Essex (UK
Topic searches pose a challenge for web-scale discovery. A pilot study of the application of Universal Decimal Classification (UDC) to manage topic access to the collection was underway when the UK Data Service began to plan for the management of new and novel forms of data (NNfD) such as ‘big data’ or administrative data. This paper reports on the results of the pilot project and explores an application for managing end user access, not only to data with a clearly defined scope and accompanying metadata, but also to data which will challenge current curation procedures. NNfD will not have been collected for research purposes but nevertheless may be a rich source of primary data. Users of NNfD will want to evaluate the suitability of this data for a particular research project and may also wish to access similar data from our collection that was curated for the purpose of secondary data analysis. Application of a standard classification code such as UDC, we anticipate, will assist us to negotiate discoverability issues that will undoubtedly arise as researchers explore new and novel sources of data.
, The Courtauld Institute of Art (UK)
This paper considers a particularly knotty aspect of classifying notated music: the classification of instrumental ensembles, where the term ‘ensembles’ is defined as music written for multiple players with only one player per part. Facet analysis is used to examine this area of music classification and as the basis of a model for classifying ensembles. The conceptual analysis is aided by examples drawn from two classification schemes, British Catalogue of Music Classification and Flexible Classification. First, this exploration reveals that there are conceptually four sub-facets for classifying instrument ensembles, and that the omission of any of these sub-facets causes issues within classification schemes. Next, the different type of relationships between pairs of these sub-facets is delineated, including hierarchical and associative relationships. The classification of ensembles is depicted in a novel way, as a series of inter-connected relationships between sub-facets. Finally, the paper ascertains exactly what is being counted, including introducing potential extra sets of sub-facets pertaining to performers and hands. So, facet analysis helps to create a model for classifying instrumental ensembles which provides a novel solution to this historically problematic area of music classification, as well as suggesting a potentially generalizable new way of thinking about complex relationships between sub-facets.
, University of Debrecen (Hungary)
Analytico-synthetic and faceted classifications, such as Universal Decimal Classification (UDC) provide facilities to express pre-coordinated composite subjects by using syntactic relations. In this case the effective decisions regarding the relevance or the identification of the described object require extracting the meaning of the classmarks as precisely as is possible. In information retrieval, it is a central question: how the identification mentioned above can be supported by automatic means. Among other things, analysing the structure of the codes is an obvious requirement of this. The current research has focused on developing a machine-readable format that contains the whole syntactic structure of the composite UDC numbers to support their further automatic processing. An algorithm that can produce the representation of the numbers in such a format directly from their designations has also been developed and implemented. The research also includes implementing conversion methods to provide outputs that can be employed by other software directly and, as a service, make them available for other software. The current paper summarizes the status of the project, the developments that have been implemented since it was presented at the International UDC Seminar 2015 and outlines future research plans.
, UDC Editorial Team (Switzerland)
The UDC classmarks and their verbal representation in Network of Libraries and Information Centers in Switzerland (NEBIS) subject index provides both detailed subject access in the process of retrieval and a desktop tool for systematic indexing and classification. The subject authority control in NEBIS supports simple and advanced searching of UDC notation and their verbal representation comprising of both hierarchical navigation (semantic expansion to broader and narrower topics) as well as advance searching. This authority control tool enables searching for combinations of the main topics with general facet categories such as place, time, form and language of the document and makes the good use of UDC analytico-synthetic structure. Complex subjects that are expressed with complex UDC notations and elements of complex UDC expressions can be searched for in isolation or as combinations (e.g. via Boolean logical operators). Whereas most of the traditional catalogues depend primarily on Boolean logic, NEBIS search is based on an enriched subject index that contain additional, closely related terms and synonyms for each concept to assist the search. The advantage of NEBIS subject index is that it allows searching of descriptors in three languages as well as searching for elements of complex UDC notations.
, Cracow University of Economics (Poland)
, Cracow University of Economics (Poland)
The paper discusses two different approaches to similarity measurement between UDC expressions. The purpose of the research was to propose a general formula for calculation of similarity coefficient reflecting relatedness between two different UDC expressions which could contribute towards managing complex UDC notation in information retrieval. The first approach was used to measure the degree of the overlap between two UDC class tree structures representing given expressions . To achieve a measure of similarity between two complex UDC expressions first a process of their decomposition should be performed. This process allows to transform every complex UDC expression to a set of UDC classes located in the UDC knowledge tree. Finally two trees representing given UDC expressions are vectorised and compared with Jaccard coefficient. Its value can be treated as a similarity measure between original UDC expressions. The second approach was concerned with measuring information similarity between UDC classmarks using Lin's method (which compares the information content of two concepts with their lowest common subsumer). For this approach estimation of probability of occurrence for every UDC class is required. Aggregating Lin’s similarity coefficient between classes the final measure of similarity between complex UDC expressions can be expressed. The research, conducted, in part, on the catalogue of the National Library of Poland. provide an evaluation of strengths and weaknesses of each of the two tested similarity measurement methods. The authors explain the potential application of similarity measurement in implementing semi-automatic classification solutions and in performing cluster analysis of bibliographic resources.
, Universidade de Coimbra (Portugal)
, Universidade de Coimbra (Portugal)
, Universiade Estadual Paulista (Brazil)
The subject indexing of fiction is a complicated matter. In addition to the difficulties that affect the indexing of non-fiction, the determination of the aboutness of narrative fiction makes the process even more challenging. The disparity in the understandings of the concepts of subject and aboutness (sometimes across languages, such as in the case of Portuguese) complicates the matter even more. Some initiatives around the world have addressed this problem following a facet-analytic approach (e.g., Pejtersen, 1979; American Library Association, 1990; Beghtol, 1994; Jansson & Södervall, 1987; Saarti & Hypén, 2010; Hypén & Mäkelä, 2011; Fideli, 2015). Despite the multiplicity of views of these authors, none of them seems to have worked with the strict application of Ranganthan's facets (Personality, Matter, Energy, Space, and Time). In this paper, we study the possibilities and contribution of Ranganthan's PMEST formula for the determination of the aboutness of novels. We aim to determine the effectiveness of this approach for the indexing and retrieval of this kind of fiction in relation to the users' views. We distributed a questionnaire among the readers of a public library in Portugal in order to know their views on the PMEST facets of the novels they read.
, UDC Consortium (The Netherlands)
, Robert Gordon University Aberdeen (UK)
The paper will look into different patterns of facet analysis used in the UDC schedules and how these affect the scheme presentation, the underlying data structure and the management of the classification scheme both at its source and at the point of use. From the very beginning UDC was designed to represent the universe of knowledge as an integral whole allowing for subjects/concepts from all fields of knowledge to be combined, linked and the nature of their relationships made explicit. In Otlet's original design, the emphasis for his new type of classification was on the coordination of classmarks at the point of searching, i.e. post-coordination, which he firmly rooted in an expressive notational system. During its long history, while the main analytico-synthetic principle remained the same, knowledge fields in UDC grew organically, i.e. without a coherent theoretical framework that can be recognized across all fields. Thus, while some UDC classes exhibit all patterns of facet analytical theory proper, others, although used in an analytico-synthetic fashion, may still contain hierarchies enumerating complex subjects. The authors describe the history of two kinds of proposals for improving the UDC: proposals concerned with a more rigorous faceted theoretical framework and those arguing for further formality in notational representation and in synthesis. The authors argue that a distinction should be made between: a) theoretical requirements of an overarching facet analytical theory as a founding principle guiding schedule constructions; and b) practical requirements for an analytico synthetic-classification in terms of notational presentation and data structure that enables its use in indexing and retrieval, as well as its management online.
, Federal University of Minas Gerais (Brazil)
, Federal University of Minas Gerais (Brazil)
, Federal University of Minas Gerais (Brazil)
Digital libraries, especially those whose collections are composed of theses and dissertations, lack more adequate mechanisms for the exploitation of the collection. The users need to elaborate query expressions, and the inability to express clearly and precisely the information that is needed besides problems related to the ambiguity of the natural language, has a negative impact on the results of the search. This research aimed to facilitate the exploration and retrieval of information by users in digital libraries of theses and dissertations, as well as to improve their experience and satisfaction. This general objective was be pursued through the use of mechanisms to represent and organize the collection of a digital library of theses and dissertations, based on a faceted classificatory structure, as well as the implementation of information retrieval mechanisms based on dynamic taxonomies. As result of the research, a prototype, called TDF-Biblio, was developed as a proposal to represent and organize the collection of a digital library to facilitate information retrieval and to improve user satisfaction. This model used the facet analysis theory to represent and organize knowledge in a digital library of theses and dissertations, as well as dynamic taxonomies, with the purpose of composing the information retrieval mechanism. The faceted dynamic search provided support for a new access paradigm, allowing the guided exploration of the collection and bridging the search and navigation processes.
, University of Alberta (Canada)
Can we achieve the goals of facet analysis without actually performing facet analysis? There is, perhaps, an implicit assumption in the field that the answer to this question is no. This paper will suggest that the answer may well be yes. We review the goals of facet analysis and discuss ways in which these might be achieved. We then discuss some of the challenges faced in performing facet analysis – and thus the potential advantages of an approach that maintains the goals of facet analysis while foregoing the means. We then explore whether the use of basic grammar in subject headings, coupled with flat schedules of controlled vocabulary addressing (separately) nouns, verbs, and adjectives/adverbs, can achieve the goals of facet analysis. Classifiers might then move fairly directly from a sentence in a document description to a sentence-like subject description. Though different grammatical elements can be identified with different facets the classifier need not do so. We briefly apply this technique to a small sample of recent books. We close by drawing some lessons from Preserved Context Indexing System (PRECIS), an indexing approach employing grammar to a considerable degree that was applied in the British National Bibliography and elsewhere for a couple of decades from the 1970s.
, Kent State University (USA)
, Academia Sinica, Taipei, Taiwan (China)
This poster discusses the applications of faceted systems seen from end-user interface designed for facilitating the explorations of museum collections on the Web. With the incredible investments in digitization during the last two decades, many museum objects that may take over 100 years to be physically exposed to the on-site visitors of the exhibitions now are presented (through their digital replicates) to online visitors across the globe. How can the museum websites present the unique objects meaningfully across their collections, while also allowing visitors to explore the collections based on the objects’ unique and common properties? The authors conducted a series of visits (on-site and online) of selected museums, examined their websites and mobile apps, and sorted out the common and unique approaches used by museums in facilitating the explorations of museum collections on the Web. In addition to examining the common facets employed by these museum websites for exposing their collections (such as featuring by Who, What, Where, and When), the authors also traced the hidden facets behind (e.g., collections presented online according to culture, place, time, type, and style may reflect a museum’s setting of original collections, galleries, and exhibitions). This poster will share the findings of a part of the study.
, University of Alberta (Canada)
,University of Wisconsin-Milwaukee (USA)
Interdisciplinarity in knowledge organization is an increasingly critical component of the theory of how knowledge might be usefully clustered around particular phenomena rather than in disciplinary hierarchies. Gathering by discipline provides certain epistemic assurances concerning the treatment of phenomena, but concomitant scattering by discipline prevents the phenomenon-based knowledge discovery that is a hallmark of interdisciplinary research. This poster connects interdisciplinarity to facet analysis. We share results from an exploratory study that compares the approach to interdisciplinarity provided by the Universal Decimal Classification’s synthesis and faceted auxiliaries to that provided by the Basic Concepts Classification, which uses basic grammar to incorporate elements of facet analysis. A set of use cases was assembled for which complex multiple UDC strings were compared to grammatically structured BCC strings. The nodes, auxiliaries, and connectors in classified strings—in both UDC and BCC—constitute a network among elements of each classified string. We show how the network structures are comparable, not just as descriptive data, but as networks underlying classification as navigable pathways among concepts.
, VZG/GBV (Germany)
The availability of tools and standards has increased the use and exchange of KOS during the last few years. Concordances between these systems are however, rather rare. Project Coli-conc aims to address this gap by developing tools, methods and techniques to simplify and accelerate the intellectual creation of concordances. It also aims to ease the use and exchange of the same and at the same time provides quality monitoring that aids quality management. The project creates a set of reusable software modules to enable a uniform access to knowledge organization systems, concordances, and concordance assessments. These modules are provided as a web application to support effective processing of concordances. In addition, existing software have been evaluated and enhanced with new components for storage, access to, and analysis of different concordances.
, VINITI - All-Russian Institute of Scientific and Technical Information (Russia)
The poster represents an automatic mapping between Web of Science (WoS) subject categories and UDC using a pivot method. Semantic relationships between subject areas of WoS and UDC classes are established through relationships each of these two systems has with classes in the Russian official national classification for scientific and technological information, known as State Rubricator of Scientific and Technological Information (GRNTI). The work undertaken shows that in most cases, whenever UDC and WoS headings were linked to the same GRNTI category, a direct link between the UDC and WoS headings could be determined by an algorithmic inference. This method of mapping classifications provides a rough estimate i.e. an approximation of the headings' correspondence the advantage being that it is fully automated. This makes the task of matching complex classification schemes feasible within a reasonable time framework and limited resources. Our poster shows the schema and tables of the logical inference of the relations between headings. The parameters of the UDC and WoS concordance tables, produced by this method, are also shown. This work is a part of a larger project of automatic mapping of subject vocabularies used in information indexing in Russia which is undertaken by VINITI (All-Russian Institute of Scientific and Technical Information, department of the Russian Academy of Science).
, National and University Library Ljubljana (Slovenia)
The poster outlines an approach to using UDC's faceted structure in Slovenian libraries. Subjects are often context specific and related to a certain region, language or peoples. In UDC, various facets, whether denoting common concepts or concepts specific to certain subject fields, can be combined to describe complex subjects in a more specific and detailed way. For instance, basic subject classmark can be extended with any number of attributes to denote languages, ethnic grouping, places, time, form, etc. This makes classification culturally relevant and widely acceptable. In Slovenian libraries, examples of such use of facet-based indexing can be observed in bibliographic records, in different examples of classification summaries, in physical arrangement of objects, etc. The importance and the role of UDC facets in improving subject access is examined in a controlled list of UDC codes used in the field 675 (COMARC format) and which are extracted from the standard version of the scheme UDC MRF 11. Special attention is paid to the principles of construction and use of UDC classmarks for works of fiction (literature). The presence of common and special auxiliaries is observed in relation to ethnic aspects, which are often present in classification of literature.