National Library of Poland Descriptors model as an attempt of opening library data for reuse

Abstract

National Library of Poland introduced the Descriptors model to the structure of its authority data in the bibliographic database in order to allow better data segmentation within authority and bibliographic data and in a consequence – to enable the shift from unstructured data to structured information and to create additional links between defined entities in the National Library database thus improving possibilities for data retrieval and linking with other datasets. Data atomization and usage of standard controlled vocabularies were prerequisites for data segmentation.

In accordance to FRBR model we organized our data model basing it on the notion of entities instead of headings – they emerged from the shared pool of merged name and subject authority files. Every entity according to the entity type has a set of attributes allowing phrase based linking. The new MARC 21 fields allowed assigning additional attributes to entities, which are currently being populated in the database using the variety of methods – from manual and semi-automatic data processing based on use of regular expressions to more automatic use of matching algorithms. Changes applied to the controlled vocabulary itself were both prerequisite for data atomization and consequence of taking this approach. The achieved better data segmentation is supposed to allow populating additional facets in the faceted search of library catalog thus strongly improving user experience and possibilities of information filtering as well as to extract data with specific attributes and attribute based datasets in various data formats. To clarify and exemplify these benefits the special web based data extraction tool “data.bn.org.pl” is presented.

Furthermore, the explanation is provided of how the National Library Descriptors model supports the simplicity, interoperability and Semantic Web compatibility of the National Library’s metadata