Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich.
mehr Informationen...
Automatically building semantic resources is essential to low resource-languages like Uyghur. However, Uyghur suffers from a lack of publicly available evaluation dataset for automatically building semantic resources like WordNet. To cope with this problem, first, we build the largest Uyghur-English and English-Uyghur dictionaries by exploiting many possible online and offline resources. Then by using Princeton WordNet (PWN) 3.0 and Contemporary Uyghur Detailed Dictionary (CUDD), we construct an English-Uyghur WordNet evaluation dataset which is publicly available (https://github.com/kaharjan/uywordnet). In this dataset, more than 73,000 English synsets are mapped Uyghur automatically, in which over 20,000 are annotated manually. And the corresponding Uyghur words include definition and examples in Uyghur language context. We also propose a Synset Mapping based on Word Embeddings (SMWE) method. The experimental results on the dataset are promising.