Adapting GermaNet for the Semantic Web

Adapting GermaNet for the Semantic Web Claus Zinn Department of Linguistics University of Tuebingen Germany claus.zinn @uni-tuebingen.de Marie Hinrichs Department of Linguistics University of Tuebingen Germany marie.hinrichs @uni-tuebingen.de Erhard Hinrichs Department of Linguistics University of Tuebingen Germany erhard.hinrichs @uni-tuebingen.de Abstract GermaNet1 (Hamp and Feldweg, 1997) is a lexical-semantic net that relates German nouns, verbs, and adjectives semantically. For this purpose, it groups lexical units that ex- press the same concept into synsets and it de- fines semantic relations between them. Ger- maNet has been developed since 1997, and its most recent edition contains over 200,000 lexical units and about 160,000 synsets. The GermaNet resource is of high quality as all its entries have been manually entered with great care. GermaNet has been linked with the In- terLingual Index and with Wiktionary, and it is our goal to increase such linkage with other resources such as the Leipzig Corpora Collec- tion and the DWDS-W¨orterbuch. For this pur- pose, GermaNet is converted to RDF, a format that facilitates the interlinking of data sources significantly. 1 Introduction GermaNet is a rich lexical resource that describes German vocabulary as a light-weight ontology. While GermaNet has been inspired by the Prince- ton Wordnet, the German resource deviates from it by a number of design decisions aimed to better represent the German language, e.g., by giving an adequate account of German compounds. The cre- ation of GermaNet started in 1997 and it has been maintained and extended ever since. The latest version of GermaNet (release 17.0, April 2022) of- fers about 205,000 lexical units and nearly 160,000 synsets. It defines 173,742 conceptual relations be- tween synsets, and 12,204 lexical relations between lexical units; the number of segmented compounds is 115,366. GermaNet already has some substan- tial linking to external data sources such as 28,564 pointers to the interlingual index and 29,546 links to Wiktionary. 1https://uni-tuebingen.de/en/142806 GermaNet data is stored in a relational database from which an XML-based serialisation can be gen- erated. Although the database is part of the yearly GermaNet releases, its main purpose is to serve as a reliable way to store and manage continuous and simultaneous updates by the GermaNet team. The XML representation, which is stored in sev- eral XML files, gives programmers easy access to the data, as Java and Python libraries are available to read and access all information. However, it is not practical, nor intended, to extract informa- tion about synsets and their lexical entries from the XML representation using a text editor. In this paper, we describe how we map GermaNet’s XML- based format to RDF, the standard format for data interchange in the Semantic Web. The new format gives users a compact, human-readable representa- tion, as all information about a synset (or a lexical unit) is directly attached to it. The RDF format also makes it possible to easily link such information with external knowledge sources such as Babelnet, Wikidata, DWDS, or the Leipzig Corpus Collec- tion. 2 Background 2.1 GermaNet In many ways, GermaNet’s XML serialisation …

Content truncated to 500 words