Automatic thesaurus development: Term extraction from title metadata
Journal of the American Society for Information Science and Technology - Research Articles
| Hi-index | 0.00 |
Modern information systems such as the WWW and Digital Libraries contain more data than ever before, are globally distributed, ease their use and, therefore, become accessible to huge, heterogeneous user groups. On the other hand, the potentially enormously large amount of heterogeneous information requires powerful tools for the user to find the relevant pieces of data. One such tool are thesauri. They are a proven means to provide a uniform and consistent vocabulary for indexing and retrieval of information bearing objects (IBOs). Modern multilingual and multi-subject information systems require more than the traditional single-language, narrow-focus thesauri. The broad clientele of information systems demands thesauri that can be used by non-specialists. To achieve this goal, we introduce the framework of thesaurus federations, i.e., a loose compound of distributed, multi- or monolingual thesauri that goes beyond the already known concepts of multi-thesaurus systems. We classify multi-thesaurus systems into multi-thesaurus environments, thesaurus switching systems and thesaurus compounds.Our architecture is based on a mediation layer and wrappers for the integration of heterogeneous, distributed thesauri. We present a Java-based prototype system which enables integrated access to several thesauri available through a SQL or HTML interface via a comfortable thesaurus federation browser. This system has been used for the retrieval of metadata records managed by the Catalogue of Data Sources of the European Environment Agency (EEA).