MCINext/synthetic-persian-chatbot-tone-user-classification

Dataset Summary

Synthetic Persian Chatbot Tone User Classification (SynPerChatbotToneUserClassification) is a Persian (Farsi) dataset built for the Classification task. It focuses on identifying the tone of user input in a chatbot conversation. This dataset is part of the FaMTEB (Farsi Massive Text Embedding Benchmark) and was synthetically created using the GPT-4o-mini language model.

• —Language(s): Persian (Farsi)
• —Task(s): Classification (Tone Detection)
• —Source: Synthetically generated using GPT-4o-mini
• —Part of FaMTEB: Yes

Supported Tasks and Leaderboards

The dataset evaluates a model's ability to classify user input tone (e.g., polite, aggressive, neutral) in chatbot environments. This is essential for emotion-aware and adaptive dialogue systems. Results can be found on the Persian MTEB Leaderboard under the classification category.

Construction

1. 1.Conversation snippets and user messages were generated in various tones using GPT-4o-mini.
2. 2.A predefined tone taxonomy (e.g., polite, rude, enthusiastic, sad) was used.
3. 3.Messages were labeled accordingly based on prompt conditioning.
4. 4.A sample set was manually validated to ensure accuracy and consistency.

Data Splits

• —Train: 87,208 samples
• —Development (Dev): 0 samples
• —Test: 21,799 samples