Towards a Linguistic Ontology with an Emphasis on Reasoning and Knowledge Reuse

Towards a Linguistic Ontology with an Emphasis on Reasoning and Knowledge Reuse A. Parvizi, M. Kohl, M. Gonz`alez, R. Saur´ı Oxford University Press Oxford, United Kingdom {artemis.parvizi, matt.kohl, meritxell.gonzalezbermudez, roser.sauri}@oup.com Abstract The Dictionaries division at Oxford University Press (OUP) is aiming to model, integrate, and publish lexical content for 100 languages focussing on digitally under-represented languages. While there are multiple ontologies designed for linguistic resources, none had adequate features for meeting our requirements, chief of which was the capability to losslessly capture diverse features of many different languages in a dictionary format, while supplying a framework for inferring relations like translation, derivation, etc., between the data. Building on valuable features of existing models, and working with OUP monolingual and bilingual dictionary datasets, we have designed and implemented a new linguistic ontology. The ontology has been reviewed by a number of computational linguists, and we are working to move more dictionary data into it. We have also developed APIs to surface the linked data to dictionary websites. Keywords: lexical ontology, knowledge reuse, linked data, under resourced languages 1. Introduction For years, dictionaries, thesauri, and morphological re- sources were only intended for human consumption and, therefore, inaccessible to machines. Many Natural Lan- guage Processing (NLP) tasks such as machine translation, natural language generation, and word sense disambigua- tion require machine readable lexical resources. These lex- ical resources must be able to provide some level of in- terlinking between the data; WordNet1 is one such exam- ple. The structure and the combination of resources (dictio- nary and thesaurus) that a semantic network like WordNet presents have enabled many NLP tasks to be done at scale. The existence of WordNet as a successful resource encour- aged the creation of WordNet in many languages other than English, as well as resources such as FrameNet2, VerbNet3, and BabelNet4 with different functionalities to WordNet. Re-structuring these lexical resources using Semantic Web standards such as RDF and OWL offers even more oppor- tunities. An example would be BabelNet, which uses the lemon model (Ehrmann et al., 2014) as a lexical ontology. As a result, the linked data version of BabelNet builds on its many multilingual resources by linking together vari- ous datasets into a single resource, with added emphasis on supporting the Semantic Web community and linked data- based NLP applications like word sense disambiguation. A number of specific benefits of modelling linguistic re- sources as linked data have been identified (Chiarcos et al., 2013), such as: (i) structural interoperability, (ii) fed- eration, (iii) enhanced conceptual interoperability, (iv) a rich ecosystem of formalisms and technologies, and (v) dy- namic import (the possibility of creating resolvable links between resources that are maintained by various data providers). As described in (Chiarcos et al., 2013), the chal- lenge mainly arises from information integration; how dif- ferent types of resources can be combined in an efficient way. An upper ontology is one means for various natural 1https://wordnet.princeton.edu 2https://framenet.icsi.berkeley.edu 3http://verbs.colorado.edu/verb-index 4http://babelnet.org languages to be able to communicate, but upper ontologies are often sparse …

Content truncated to 500 words