On the 19th of December 2017, MPN IPTS Cluster organized a seminar on “Sentiment Analytics on National Issues Through Social Media: A Novel Approach to transforming Unstructured Data for Predictive Modelling using Machine Learning Algorithms” at the Pullman Hotel in Kuala Lumpur. The Head of Project of MPN IPTS Cluster, Professor Ong Fon Sim, welcomed delegates from the private-sector and public-sector companies, government agencies and universities. Participants from Malaysia Airlines Bhd, Maxis Bhd, SAS Malaysia, Polis Diraja Malaysia, KWSP, Universiti Malaya, Universiti Kebangsaan Malaysia, Jabatan Perpaduan Negara dan Integrasi Nasional and others were among the 50 attendees of this seminar.
This seminar entailed a novel approach on how machine learning algorithms can gauge public sentiments on national issues via social media platforms. This novel approach is potentially a disruptive technology that will pave way to new insights in big data analytics. From using this approach, people’s opinion, sentiment, and attitude towards issues that affect their daily life can be gathered, analysed and organised into patterns that enable public policy makers to gain a deeper insight for better informed decision making, thus minimising unwarranted risks. It will also be useful to private sector companies in marketing and customer relationship management.
The session started with an introduction and aim of the presentation followed by the proof of concept and methodology on how sentiment analytics on National Issues through Social Media can be done. Then 3 cases which were general sentiments of FITMY in Malaysia, the SEAGAMES hosted by Malaysia in 2017 and the Budget 2018 were presented. Finally, discussion and conclusion were put forward to the audience for a Q&A session. The audience were intrigued by this technique and wonder if it is scalable in a larger dataset. Additionally, suggestions to improve the modelling process using multi-modelling and multigroup analyses were presented. The results in these cases were consistent to that found in various literature as well as other groups in MPN doing similar studies. Overall the session went well as there were opportunities to collaborate with other MPN clusters in this area.
MPN @ MEDIA SOSIAL