| Abstrakt: |
Emotion plays a vital part in the life of humankind. Emotion Recognition (ER) is the necessary task to find one's inner Emotion through speech, text or face. ER process has inevitable places in applications such as education, healthcare, defence, forensics, automation and scientific-based purposes. However, the recognition models face various issues during the process. Due to the large feature sets, the model's computational time and complexity have been increased; it also produced high errors in recognition, which degrades the efficiency of the system. So, a novel Monkey-based V-Net Framework (MbVF) was developed for this research work to classify the Emotion and sentiment patterns using healthcare data. Initially, the healthcare Artificial Intelligence (AI) conversation data from the Coronavirus Tweets NLP-Text Classification dataset was collected and trained for the system. Next, stemming, stop-words removal, tokenization, and noise feature elimination process were performed in the preprocessing step. Then, the Term Frequency based-Inverse Document Frequency (TF-IDF) features were used to choose the best features depending on the fitness features of the proposed MbVF approach. Finally, the ER process was performed to recognize the emotions, such as positive, neutral and negative, using sentiment analysis with the emotion state transition and emotion pattern. The model attained 99.98% accuracy in predicting emotions, and is higher than the other model. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink