Concepts in Context

Cologne Conference on Interoperability and Semantics in Knowledge Organization



A Semantic Web View on Concepts and their Alignments – From Specific Library Cases to a Wider Linked Data Perspective / Antoine Isaac
In the last couple of years, the linked data approach to data publication, sharing and interlinking has gained momentum. A growing amount of data is being openly released on the web using RDF. This includes a number of reference knowledge organization systems, especially the ones used from the largest libraries. The release of the SKOS vocabulary to represent and network knowledge organization systems, combined with the launching of many pilot or production-level application projects and standardization initiatives on the same theme, has been a key enabler in that respect. However, some argue that linked data still requires more links than what is currently available to fully deliver on its promises, especially at the conceptual level. A brief presentation will be given on landmark projects that position themselves in the linked data paradigm, as well as their choices for concept publication and mapping. What is being mapped? How? For which purpose? This talk will also give an overview of some issues that remain to be tackled, echoing some of the other presentations of that workshop, and--hopefully--paving the way for more synergies.

Conceptual Foundations for Semantic Mapping and Semantic Search / Dagobert Soergel
This talk presents a method for mapping between Knowledge Organization Systems or between KOS and natural language based on semantic analysis. Concepts from different KOS and word senses are represented through canonical expressions constructed from a core classification of elemental concepts and semantic relationships, a method akin to componential analysis in linguistics. From these canonical expressions a reasoner can infer mapping relationships. Two main advantages of this approach are: (1) It can be applied across languages. (2) The derivation of the canonical expressions can be done locally in a distributed fashion for individual KOS or even individual concepts, as long as they use a central extensible core classification that is continuously updated by many contributors. Once the canonical expression is known, the concepts can be related globally to all other concepts in the system through the conceptual hub. Existing lexical knowledge bases that give canonical expressions can be used. The derivation of canonical expressions can be supported through algorithms that use linguistic analysis and exploit conceptual structures available in individual KOS and through the entire system. The talk will illustrate these ideas through many examples.

FRSAD: Challenges of Modelling the Aboutness / Maja Žumer
Functional Requirements for Subject Authority Data (FRSAD) is the last model of the FRBR family. It focuses on the 'is subject of/has subject' relationship. The dilemmas during the development of the model are discussed as well as some of the comments received during the review process. Particular emphasis is given to the future task of harmonising the FRBR family.

In Pursuit of Cross-Vocabulary Interoperability: Can We Standardize Mapping Types? / Stella Dextre Clarke
In the last few years projects such as DESIRE, MACS, CrissCross and KoMoHe have demonstrated the benefits as well as the challenges of mapping between controlled vocabularies. Each project has taken a slightly different approach to the definition and implementation of appropriate types of mapping. The mapping types supported by SKOS are slightly different again. In an ideal world, all major databases would be interconnected; all widely used thesauri, classification schemes and subject heading schemes would map to each other; and all of us would use the same basic types of mapping to enable universal interoperability. Or would we? This presentation will discuss the feasibility and desirability of agreeing and implementing standardized mapping types.

Insights and Outlooks: A Retrospective View on the CrissCross Project / Jan-Helge Jacobs, Tina Mengel, Katrin Müller
Providing appropriate access to heterogeneously indexed collections is a contemporary goal for libraries worldwide. In order to take up this challenge, the project CrissCross has been initiated by the German National Library (DNB) and the Cologne University of Applied Sciences. Following a conceptual mapping approach, topical headings of the German subject headings authority file Schlagwortnormdatei (SWD) are being mapped to notations of the Dewey Decimal Classification (DDC) to improve subject access to heterogeneously indexed documents. The end of the project in autumn 2010 offers an opportunity to present an overall review of the project’s aims, methods and benefits. Special focus is given on project-specific retrieval concepts.

Interoperability and Semantics in RDF Representations of FRBR, FRAD and FRSAD / Gordon Dunsire
The presentation will describe recent work on registering Resource Description Framework (RDF) versions of the entities and relationships from the Functional Requirements for Bibliographic Records (FRBR) and Functional Requirements for Authority Data (FRAD) models developed by the International Federation of Library Associations and Institutions (IFLA). FRBR was developed several years before FRAD, and is under-developed in areas which FRAD was expected to cover; FRAD therefore makes significance reference to FRBR. Similarly, FRAD leaves a full treatment of subject authority data to the ongoing development of Functional Requirements for Subject Authority Data (FRSAD) which is expected to be finalised during 2010. Although the FRBR Review Group is charged with consolidating all three models in due course, the RDF versions of FRBR and FRAD are being created in separate namespaces, with a separate Web ontology Language (OWL) ontology to connect the two models. The presentation will discuss interoperability issues arising from this work, and take into account the requirements of FRSAD and the consolidated Functional Requirements model. Such issues include class definitions and sub-classes, reciprocal properties, and disjoint classes and properties. The presentation will also discuss related issues arising from the RDF representation of the metadata element set of RDA: resource description and access, which is based on the FRBR and FRAD models.

Intersystem Relations: Characteristics and Functionalities / Jessica Hubrich
As a result of the methodological support of the CrissCross project and the research conducted within the RESEDA project, a tiered model of semantic interoperability was developed which correlates methods of establishing interoperability resp. the resulting types of intersystem relations to specific search functionalities in retrieval scenarios. The presentation will describe this model and discuss characteristics of conceptual intersystem relations as well as their impact on enhancing access to information in detail.

Searching in a Multi-Thesauri-Scenario – Experiences with SKOS and Terminology Mappings / Philipp Mayr
With the standardization of SKOS in August 2009 a data model has been offered to publish controlled vocabularies and taxonomies on the web in a technical and semantically interoperable way. The heterogeneous environment of various vocabularies worldwide can be harmonized prospectively and especially the content of traditional databases can be made accessible and connectable for applications of the Semantic Web, i.e. as Linked Open Data. This presentation proposes a search scenario with three topic-related thesauri, which have been connected with cross-concordances as part of a major terminology mapping initiative in the project KoMoHe. The thesauri have already been or will be converted to SKOS and in order to not omit the relevant crosswalks, the mapping properties of SKOS will be used for modeling them adequately. The presentation will highlight vocabularies in SKOS format can play an eminent role for search, because they can serve as a bridging hub for the inter-linking of different published and indexed data sets.

Translingual Retrieval: Moving between Vocabularies – MACS 2010 / Helga Karg and Yvonne Jahns
Within the multilingual framework of the CrissCross project, MACS (Multilingual Access to Subjects), has continued its work. MACS has developed a prototype of mappings between three vocabularies: the LCSH (Library of Congress Subject Headings), RAMEAU (Répertoire d'autorité-matière encyclopédique et alphabétique unifié) and the SWD (Schlagwortnormdatei). A database with a Link Management System (LMI), which allows for an easy linking between English, French and German subject headings, was created. The database started working with headings from the disciplines sports and theatre, but by now headings from all other fields of knowledge have been included as well. In 2008–2010, equivalencies between English and French headings which had been produced by the Bibliothèque nationale de France have been completed with the most important German SWD topical terms. Thus, more than 50.000 trilingual links are now available and can be used in different retrieval scenarios. It is planned to use them in The European Library (TEL) in order to support multilingual searches over all European National Library collections. Helga Karg and Yvonne Yahns, Deutsche Nationalbibliothek, demonstrate in their presentation the project workflow, methodology of mapping and future applications.

Would an Explicit Versioning of the DDC Bring Advantages for Retrieval? / Claudia Effenberger and Julia Hauser
The DDC is constantly changing. In order to keep the classification up-to-date with scientific advancement and literary warrant, the editorial process regularly revises specific areas in the tables or schedules, and as a result particular topics in a class are relocated into other classes or new, subordinate classes are created. In the German National Library the DDC is the most important system for classification and indexing. Strictly spoken it would be necessary to regularly review the correctness of the DDC notations, since their new meaning may not correctly reflect the contents of the bibliographic medium any more. For economical reasons, this is not possible with the result, that a search for literature on a specific topic may return improper resources, since that topic might not be represented by the used DDC notation any more. The objective of CrissCross is to enrich the search vocabulary for DDC by linking terms from the Schlagwortnormdatei (German Subject Headings, SWD) to the corresponding DDC classes, and thus to improve the possibilities to find DDC-classified resources. The constant revision of the DDC makes it necessary to periodically update the links from the SWD to the DDC, since the parts of the subject heading’s topic might not be accurately reflected by the DDC class. This link, however, is regularly updated and older versions of the relation between the subject heading and the DDC class are versioned within the SWD authority record. In a small research project, the DNB currently investigates if it is possible to solve this problem by giving each version of a DDC class a unique identifier. By doing that it would be possible to explicitly label which version – and thus which topics contained – of a DDC class was used for the classification of a particular resource. If those identifiers conform to the generic URI syntax we can model the relations between the bibliographic resources, the subject headings and the different versions of the DDC classes as a semantic network using RDF and then investigate if this approach can improve retrieval in heterogeneously indexed collections. This talk will present some preliminary results.