Problem addressed

Suicidal ideation is one of the leading mental health issues among young people in Hungary and in the EU. The individuals suffering in presuicidal syndrome communicate about their mental problems manily in online texts (i.e., in forum posts, social media posts and email consueling). These texts constitute a vast amount of linguistic material for the research in discursive suicidology. However, we do not have either enough human resource or appropriate knowledge about what the typical patterns of metaphorical language use are in these discourses. As a result, the linguistic material on suicidal ideation, even if it is available for analysis, cannot be used in large-scale research and in the development of innovative tools in suicide prevention.

Innovative solution

To address this problem, our project aims at creating an online platform where any input texts can be analysed in terms of metaphorical language use. The project has two primary innovative aspect. First, it uses large language models (i.e., generative artificial intelligence) for automatic metaphor identification. Second, it develops a natural language processing pipeline that integrates a lexical database, automatic linguistic preprocessing and promtping techniques to provide a fast and effective solution for finding metaphors in online discourse. This methodological innovation targets the Hungarian language, but it can be implemented in other languages as well. 

Key results and benefits

The most important result of the project is the online platform for metaphor identification. It can be used in large-scale corpus building and metaphor analysis, thus, it enhances the exploration of metaphor markers in presuicidal discourse (or in other discoursive fields related to mental health problems). The key benefit of the project lies in the acceleration and support of creating a representative linguistic database. On the basis of it we can explore systematically the role of metaphors in presuicidal discourse, and with this knowledge effective automatic alarming systems can be developped in the future.

Potential for mainstreaming

Since suicide affects every group of the societies, and the rehabilitation after a suicide attempt incurs high costs for the state and its health insurance system, the project can contribute to decrease the use of community resources and to incerase social equality. The integration of generative AI into the planned platform ensures the sustainability of the project both infrastructurally and in economically.