Empathy-Driven Chatbots for the Arabic Language: A Transformer Based Approach

Developing chatbots for the Arabic language presents unique challenges due to its complex grammar, rich vocabulary, and diverse dialects. Therefore, tailoring chatbot development methodologies and models specifically for Arabic is essential to ensure accurate understanding and generation of responses. This paper presents an empathy-driven chatbot which is a transformer-based model specifically designed for the Arabic language. The model is trained using a corpus of Arabic conversational pairs and is compared against a Seq2Seq model based on Bi-LSTM and attention mechanism. Empathy-driven chatbots aim to understand and respond to users’ emotions, needs, and concerns in a sensitive and human-like manner. By integrating empathy into chatbot design, we enhance the conversational experience, making interactions more personalized, engaging, and satisfying for users. While empathy has been extensively studied in English based chatbots, its application in other languages, such as Arabic, presents unique challenges and opportunities. Our research focuses on the development of an empathic chatbot that can understand and respond to user input in a contextually relevant and empathetic manner. The transformer based chatbot exhibits several advantages over the Seq2Seq model, including improved perplexity and BLEU score, indicating enhanced language modeling and generation capabilities.