Conversations Sentiment and Intent Categorization Using Context RNN for Emotion Recognition
Rifqi Majid, Heru Agus Santoso
Abstract
Nowadays, a chatbot is often found in daily life, and its role in business becomes more important. We develop Dinus Intelligent Assistant (DINA) chatbot for assisting student administration services. While having a text-based conversation, recognizing emotions in textual-based content is a challenging task. This study uses dataset which contains conversation dialogues. In this study, we preprocess conversation using sentiment analysis then collect the data by its conversations and sentiment analysis results. Then we apply RNN to categorize the emotions based on current conversations. The result of this approach is promising, with a precision accuracy of 0.76. From the result, this algorithm can indeed help recognize emotions from text-based conversations.