Development of an Emotion Driven Chatbot for IT Career Guidance

Abstract

The increasing complexity of IT career options and the emotional challenges faced by students and early-career professionals in navigating career decisions have highlighted the need for intelligent, emotion-aware guidance systems. This project presents the design and implementation of an Emotion-Driven Chatbot for IT Career Guidance, a web-based conversational system that detects the emotional state of users in real time and delivers personalised, empathetic IT career recommendations accordingly.
The system was developed using Python 3.10 and integrates a RoBERTa-based transformer model (SamLowe/roberta-base-go_emotions), pre-trained on the GoEmotions dataset comprising 58,009 annotated text samples across 28 emotion categories, for fine-grained emotion classification. Detected emotions are mapped to seven broader emotional groups, anxiety, frustration, joy, curiosity, sadness, positive, and neutral which drive the career recommendation logic. The Groq LLaMA 3.3-70b large language model API was integrated to generate dynamic, contextually aware conversational responses informed by the detected emotion. The system was deployed as a web application using the Flask framework, featuring a professional landing page and an interactive chat interface with real-time emotion badge display, career recommendation cards, career deep dives, and curated learning resources covering nine IT career paths.
The RoBERTa emotion detection model was evaluated against the GoEmotions test split of 5,427 samples, achieving an overall accuracy of 64% and a weighted F1-score of 0.64, consistent with published benchmarks for 28-class fine-grained emotion classification. A qualitative user evaluation was conducted with 6 participants using a five-point Likert scale survey, producing an overall mean satisfaction score of 3.97 out of 5.00 (79.3%). Ease of use received the highest rating (M = 4.50), while emotion detection accuracy received the lowest (M = 3.50), indicating an area for future refinement.
The results confirm that the system successfully achieved all four project objectives and demonstrates the viability of emotion-driven conversational AI as a practical and accessible tool for personalised IT career guidance. Future work should address production deployment, larger-scale user evaluation, and the integration of multimodal emotion detection.