Emotion recognition model based on Indonesian sentiment text using machine learning and neuro-physiological approach /
Emotion Recognition in the Brain and Computer Interface (BCI) field is gaining popularity, not only in terms of volume or amount of incoming data but the variety of media used by netizens and the acceleration of increasing information (velocity) as well. Therefore, the development of techniques and...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis Book |
| Language: | English |
| Published: |
Kuala Lumpur :
Kulliyyah of Information and Communication Technology, International Islamic University Malaysia,
2022
|
| Subjects: | |
| Online Access: | http://studentrepo.iium.edu.my/handle/123456789/11375 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Emotion Recognition in the Brain and Computer Interface (BCI) field is gaining popularity, not only in terms of volume or amount of incoming data but the variety of media used by netizens and the acceleration of increasing information (velocity) as well. Therefore, the development of techniques and algorithm models with various approaches is a significant concern to recognize the netizens' emotions through writing. This study examined the introduction of text-based emotions in the Indonesian language by taking Twitter data as the dataset. The dataset is processed using two approaches; 1) Recognizing emotions automatically based on sentiment text, and; 2) In real-time viewing brain waves using machine learning and Electroencephalogram (EEG) tools by neuro-physiological approach. The output of these tasks is the accuracy of training data and testing data score. Knowing the results of the accuracy of the two approaches is important, as a reference recommendation to see how much emotion affects the writer and the status of the reader. Furthermore, we conducted preliminary research to obtain Indonesian words with raw data from Affective Norm English Words (ANEW) and classify them into four basic emotions: happiness, sadness, anger, and fear. The highest scored calculation for these four emotions are carried out as keywords in crawling Twitter data. After that, it processed using the Long Short-Term Memory (LSTM) model and also using two benchmark models (Random Forest and Support Vector Machine) at the emotion recognition stage based on sentiment analysis. Next, the dataset in the form of brain waves are processed using the same models. In the sentiment analysis approach, the LSTM model has the highest accuracy value than the two benchmarks. Whereas for data using EEG, Random Forest produces the best accuracy value. Consequently, this research contributed to a collection of datasets based on affective Indonesian words. Besides, it provided recommendations for several algorithm models that match the data and the case. This research's novelty value was to recognize emotions using brain waves with stimulation of reading text with a sentiment analysis approach. Future research was still very much needed to get maximum results to provide knowledge that human emotions can be affected by reading emotional texts. |
|---|---|
| Item Description: | Abstracts in English and Arabic.
"A thesis submitted in fulfilment of the requirement for the degree of Doctor of Philosophy in Computer Science." --On title page. |
| Physical Description: | xvii, 168 leaves : color illustrations ; 30cm. |
| Bibliography: | Includes bibliographical references (leaves 154-162). |