Melania Berbatova

Melania Berbatova

The use of large language models (LLMs) for psychological and emotional support (ES) has rapidly evolved, becoming the most widely used application of generative artificial intelligence among consumers by 2025. This paper presents the results of an anonymous survey of 100 Bulgarian users, primarily high school, university, and doctoral students, to explore their attitudes toward and usage of chatbots for emotional support. Findings indicate that approximately one-half of the surveyed population utilizes chatbots for ES, with ChatGPT being the most dominant platform. Users primarily seek support for coping with stress in interpersonal relationships and work or study-related environments. While 71% of users perceive the technology as effective, non-users remain sceptical. Despite the growing adoption, significant concerns persist regarding data security, technology reliability, and the tendency of chatbots to provide excessive affirmation.

Melania Berbatova, Tsvetoslav Vasev

Toxic content detection in online communication remains a significant challenge, with current solutions often inadvertently blocking valuable information, including medical terms and text related to minority groups. This paper presents a more nu-anced approach to identifying toxicity in Bulgarian text while preserving access to essential information. The research explores two distinct methodologies for detecting toxic content. The developed methodologies have po-tential applications across diverse online platforms and content moderation systems. First, we propose an ontology that models the potentially toxic words in Bulgarian language. Then, we compose a dataset that comprises 4,384 manually anno-tated sentences from Bulgarian online forums across four categories: toxic language, medical terminology, non-toxic lan-guage, and terms related to minority communities. We then train a BERT-based model for toxic language classification, which reaches a 0.89 F1 macro score. The trained model is directly applicable in a real environment and can be integrated as a com-ponent of toxic content detection systems.