Framing karstology

We describe the creation of a knowledge base in the field of karstology using the frame-based approach. Apart from providing a new multilingual resource using manually annotated definitions as the source of structured information, the main focus is on exploring text mining methods to identify targeted knowledge structures in specialised corpora. The first stage of this process is the design of a domain model and its implementation in a definition annotation task. Once annotation is completed, an analysis of typical co-occurrence patterns between semantic categories and the relations describing them allows us to discern ideal definition templates. We demonstrate that such templates contribute to a more comprehensive and structured representations of concepts, but also help us design targeted text mining experiments to retrieve new semantic relations from text. Two such experiments are presented, the first using intersections of word embeddings to identify words expressing a specific semantic relation, and the second using the embedding of the semantic relation to extract multiword units which contain the target relation. Results suggest that the proposed methods are promising for capturing the semantic properties of relations in frame-based knowledge modelling.