Speakers
Description
Phraseological resources contain complex semantic structures that are often difficult to represent in interoperable lexical models. Although digital phraseological dictionaries increasingly use hyperlinks, thematic indices, and semantic groupings to improve navigation, many of these relations remain implicit and are primarily intended for human interpretation. This study explores the expressive capacity of the DMLex model for representing conceptual and semantic relations in a Croatian phraseological resource originally developed in Lexonomy. The resource consists of the Online Dictionary of Croatian Idioms and a manually curated thematic index grouping phraseological units into semantic domains and subdomains. The study adopts a scenario-based modelling approach focused on recurring phraseological structures observed in the resource. Three modelling scenarios are analysed: (1) variant and constructional relation networks, (2) polysemy and sense-level thematic classification, and (3) multiple thematic affiliation. The analysis demonstrates how lexical variation, constructional relations, synonymy, and thematic organisation can be represented through explicitly encoded semantic relations within DMLex. The results show that DMLex can accommodate much of the semantic complexity of phraseological resources while also revealing challenges related to overlapping classifications and conceptual hierarchies. The proposed approach contributes to the interoperable modelling of phraseological data for lexicographic and NLP-oriented applications.