HamRaz: A Culture-Based Persian Conversation Dataset for Person-Centered Therapy Using LLM Agents

TL;DR

HamRaz is a culturally adapted Persian-language dataset for AI mental health support, combining script-based dialogues and LLM role-playing to improve empathy and realism in therapy simulations.

Contribution

It introduces HamRaz, the first culturally specific Persian dataset for AI therapy, and a dual-framework for evaluating therapeutic conversations.

Findings

HamRaz outperforms baselines in empathy, coherence, and realism.

Human evaluations confirm improved conversational quality.

The dataset bridges language, culture, and mental health in AI applications.

Abstract

We present HamRaz, a culturally adapted Persian-language dataset for AI-assisted mental health support, grounded in Person-Centered Therapy (PCT). To reflect real-world therapeutic challenges, we combine script-based dialogue with adaptive large language models (LLM) role-playing, capturing the ambiguity and emotional nuance of Persian-speaking clients. We introduce HamRazEval, a dual-framework for assessing conversational and therapeutic quality using General Metrics and specialized psychological relationship measures. Human evaluations show HamRaz outperforms existing baselines in empathy, coherence, and realism. This resource contributes to the Digital Humanities by bridging language, culture, and mental health in underrepresented communities.