This research presents experiments carried out to improve the precision and recall of Dutch hypernym detection. To do so, we applied a data-driven semantic relation finder that starts from a list of automatically extracted domain-specific terms from technical corpora, and generates a list of hypernym relations between these terms. As Dutch technical terms often consist of compounds written in one orthographic unit, we investigated the impact of a decompounding module on the performance of the hypernym detection system.
In addition, we also improved the precision of the system by designing filters taking into account statistical and linguistic information.
The experimental results show that both the precision and recall of the hypernym detection system improved, and that the decompounding module is especially effective for hypernym detection in Dutch.