When dealing with mental health issues, precise emotion recognition is essential, particularly in real-time applications like chatbots. However, because sarcasm is so difficult to convey when expressing sentiments that are contrary to literal meanings, it continues to be a significant barrier to successful emotion recognition. The study demonstrates how crucial it is to recognise sarcasm as a first step in enhancing emotion recognition systems. We suggest a method to enhance the accuracy of sarcasm detection by including contextual information from parent comments. Our approach examines the interaction between comments and their prior context, providing a detailed understanding of tone and intent, using the SARC dataset and the XLNet pre-trained model for feature extraction. According to experimental data, sarcasm detection ability is much enhanced by using parent comment context, as seen by improved accuracy and superior evaluation metrics when compared to algorithms that only analyse individual comments. Building more reliable emotion identification algorithms is made possible by this study’s solution to the sarcasm detection problem. In addition to helping create chatbots that can recognise emotional distress, improved emotion detection enables real-time interventions for better mental health. The significance of context-aware natural language processing (NLP) models in developing sympathetic, helpful tools for mental health applications is highlighted by this work. Future research will concentrate on incorporating sarcasm-aware emotion detection into more comprehensive mental health solutions in order to support wellbeing and early intervention.