Farsi Mental State and Emotional Needs Dataset for Psychological Text Mining

The dataset consists of 993 validated text samples annotated manually by a human expert in Persian. Each sample contains free-form user-generated text expressing emotional distress, mental health concerns, or existential reflection.

The samples were collected and labeled between June and July 2025 from public online Persian-language mental health forums and social discussion platforms. The raw texts were anonymized and selected based on depth of psychological content and relevance to emotion-focused analysis. Texts involving offensive or highly clinical content were excluded.

Each entry in the dataset has the following structure:

{
"id": 75,
"text": "دوست دارم یکی توی این دنیا باشه التماسش کنم... چرا اینجوری شد؟...",
"mental_state": ["hopelessness", "regret"],
"emotional_need": ["closure", "forgiveness"]
}
id: a unique numeric identifier

text: the original Persian user-generated sample

mental_state: a list of inferred emotional or psychological states (e.g., hopelessness, numbness, anxiety)

emotional_need: a list of underlying emotional or psychological needs (e.g., support, validation, escape)

The labels were assigned by a native Farsi-speaking expert with a background in psychology and AI. The annotation schema was designed to reflect both observable emotional states and implied psychological needs. The labels were selected from a predefined taxonomy derived from psychological literature, but expanded inductively to include emergent patterns.

The dataset has been cleaned and formatted as a UTF-8 encoded JSON file. All records have been reviewed to remove duplicates and structural inconsistencies. No personally identifiable information (PII) is present.

Applications
This dataset is intended for use in Natural Language Processing (NLP) applications involving emotion detection, psychological profiling, affective computing, and Farsi language modeling. It is particularly valuable for research in AI for mental health, empathetic chatbots, and emotion-aware dialogue systems.

Potential users include:

Researchers in psychology and AI

NLP practitioners working on low-resource languages

Developers of mental health screening tools or therapeutic AI