A Facebook Sentiment Analysis based on Malay Text for UTeM Services
DOI:
https://doi.org/10.53840/myjict8-1-2Keywords:
Facebook, machine learning, classification, Linear Support Vector MachineAbstract
In recent years, many large companies around the world have used information from social media to identify customer needs, in addition to obtain immediate feedback from marketed products or services. Feedbacks in the form of text, voice and pictures are classified into sentiment to identify the percentage of customer satisfaction before any decision is made. However, responses written in short text and native language make the classification process become difficult. The aim of this study is to classified social media text data from Universiti Teknikal Malaysia Melaka (UTeM) Facebook which is mostly written in Malay, into positive, neutral, or negative sentiments. Facepager software that facilitates Facebook Page data extraction through Facebook’s Application Programming Interface (API) is used to extract page posts, user comments to posts, replies to comments, and engagement. The model is developed using an exactly established technique that includes pre-processing ticket data, stemming words, feature vectorization, creating training tickets, and tuning machine learning algorithms. We use variety of techniques such as Naïve Bayes, Linear SVC, and Logistic Regression to demonstrated the highest accuracy among the chosen model by using of two datasets retrieved from UTeM Facebook data, (i) dataset 1 (cleaned dataset) consists of 7,005 Facebook data, and (ii) dataset 2 (resampled dataset) consist of 6,519 Facebook data. The experimental findings revealed that the model used Linear Support Vector Classifier achieved the highest performance with 93% accuracy. To visualize the sentiment, we also built an interactive dashboard to monitor the positive, neutral and negative sentiment in each post.
Downloads
References
Balaji, T.K., Annavarapu, C. S. R. & Bablani, A. (2021). Machine learning algorithms for social media analysis: A survey, Computer Science Review, 40: 100395.
Behrens, J. T. (1997). Principles and procedures of exploratory data analysis. Psychology Methods, 2(2), 131-160.
Chen, Y.-F., & Huang, S.-H. (2021). Sentiment-influenced trading system based on multimodal deep reinforcement learning, Applied Soft Computing, 112: 107788.
Chen, L., Davoudian, A. & Liu, M. (2022). A workload-driven method for designing aggregate-oriented NoSQL databases, Data & Knowledge Engineering, 142: 102089.
Flack, J. C. & D’Souza, R. M. (2014). The digital age and the future of social network science and engineering. Proceeding of IEEE, 102(12), 1873-1877.
Mashiah, I. (2021). “Come and join us”: How tech brands use source, message, and target audience strategies to attract employees, The Journal of High Technology Management Research, 32(2): 100418.
Mehmood, N., Culmone, R. & Mostarda, L. (2017). Modeling temporal aspects of sensor data for MongoDB NoSql database. Journal of Big Data, 4:8.
Rafique, A., Van Landuyt, D., Beni, E. H., Lagaisse, B. & Joosen, W. (2020). CryptDICE: Distributed data protection system for secure cloud data storage and computation. Information Systems. 96: 101671.
Tran, M.-Q., Doan, H.-P., Q.Vu, V. & T.Vu, L. (2023). Machine learning and IoT-based approach for tool condition monitoring: A review and future prospects, Measurement, 207: 112351.
Vatrapu, R., Mukkamala, R. R., Hussain, A. & Flesch, B. (2016). Social set analysis: A set theoretical approach to big data analytics. IEEE Access, 4, 2542-2571.
Wanniarachchi, V. U., Mathrani, A., Susnjak, T. & Scogings, C. (2020). A systematic literature review: What is the current stance towards weight stigmatization in social media platforms? International Journal Human-Computer Studies, 135:102371
Witten, I. H., Frank, E., Hall, M.A. & Pal, C. J. (2016). Data Mining: Practical Machine Learning Tools and Techniques. Morgan Kaufmann.
