Stress Detection from Chat Using Natural Language Processing Embeddings and Tokenization Using OCIR Machine Learning Method

Abstract

Stress is a subjective sensation that is difficult to accurately define. Stress can have a variety of biological and psychological consequences on one's health, which can be defined and quantified. We only get half of the information if we focus solely on what someone says and ignore what their demeanor tells us. In this project, we will use random forest and decision tree algorithms to detect mental stress. The existing system does not work in real time and is inaccurate and inefficient in terms of loading and implementation durations. Furthermore, the appropriate test-train split ratio is not applied during testing and training. The OCIR proposed method is implemented in real time and has a high level of accuracy. In comparison to the current system, the suggested solution has incredibly fast loading and execution times. The OCIR proposed method can be improved for more complex use cases and is highly effective and scalable.