loder

Trends of Keywords, Sentiments, and Hoaxes Related to COVID-19 on Twitter

It has been a year since patient zero COVID-19 was first discovered in Wuhan, China. In Indonesia, the COVID-19 pandemic has been going on since March 2, 2020, when the government announced the first 2 confirmed cases of COVID-19 patients for the first time. Now, the number of confirmed cases COVID-19 in Indonesia has passed the 1 million cases as of January 26, 2021.

For almost 1 year of this pandemic, Prosa.ai managed to collect more than 240,000 data on keyword trends, sentiments, and hoaxes regarding COVID-19 on twitter using COVID-19 Social Media Monitoring based on artificial intelligence technology developed with the AI Center of the Bandung Institute of Technology (ITB)

Tren Kata Kunci.png

The interesting thing that can be seen from the data is that the most prominent keyword during the eleven months of the COVID-19 pandemic in Indonesia is the regional election, as can be seen in the picture above. Total election keyword data were 17.627 divided into 3 sentiments; positive, negative, and neutral. The identified data with positive sentiment related to the regional elections were 9,860, or the equivalent of 55,9%, while the data with negative sentiment were 6.605, or the equivalent of 37,4%. The rest, as much as 1.162, or the equivalent of 6,6%, are keywords with neutral sentiment. The election keyword itself first appeared in October 2020 with a total of 589 data until finally peaking in November 2020.

Aside from keyword trends, the COVID-19 Social Media Monitoring also helps users to monitor public sentiment regarding keywords related to the COVID-19 pandemic. This can be done because Prosa.ai and the ITB AI Center are using Natural Language Processing (NLP) technology in the form of concept-based sentiment analysis. With this technology, the COVID-19 Social Media Monitoring can extract concepts from sentences taken from Twitter and then assess the sentiment on the concept. This can be useful for regulators to map and determine policies in accordance with the existing situation in society.

Sentimen.png

From the total data taken since early March 2020, negative sentiment data made up the majority of the data as much as 44,3%. Then followed by 32% positive data and 23,7% neutral data. It can be seen that the most negative data appear in 2020 on the psbb keyword as many as 7,496 data.

In 2021, tweets with neutral sentiment began to dominate the majority of the data. Especially on the vaccine keyword. This is in line with the development of the Sinovac vaccine itself to news such as the first vaccination carried out by President Jokowi earlier this year. During the last 2 months, the keyword vaccine has become the most popular keyword on Twitter. There are 6.725 vaccine keyword data from 30.170 total data for the last two months, or the equivalent of 22.3% of the total data. A total of 4,160 data has a neutral sentiment, 1,922 data is positive, and 643 data is neutral.

Data Twitter.png

The COVID-19 Social Media Monitoring platform is also equipped with NLP technology in the form of hoax detection so that it is able to analyze whether a sentence has a hoax tendency or not. The spread of hoax on social media such as Twitter which can confuse and disturb the public can be tackled using this COVID-19 Social Media Monitoring.

hoax.png

A total of 23.755 data or equivalent to 9,8% of the total data have a high probability of hoaxes, which is >60% hoax. Meanwhile, 49.298 data or the equivalent of 20,4% were confirmed as facts. The platform can also help to map the areas where hoaxes are spreading the most. Referring to the picture above, it can be seen that the majority of hoaxes spread in Jakarta. From the total tweets on Twitter that had a hoax tendency, 52,81% or the equivalent of 12.545 tweets came from Jakarta.

The COVID-19 Social Media Monitoring platform also has various other features. Users can determine the timeframe for retrieving the required data so that they can get data that is more suitable for their needs. This platform is able to compare trends from the existing data. With this data trends comparison, it makes it easier for users to map the existing and current narratives in society. In addition to that, this COVID-19 Social Media Monitoring can also source not only from twitter, but from other social media or online news portals.

The summary of the data that has been described above can be seen in more detail through the link https://covid19-socmed.id/. Apart from the COVID-19 Social Media Monitoring which was developed together with the ITB AI Center, Prosa.ai also developed several AI solutions to help mitigate the COVID-19 pandemic, such as e-govt chatbot, AI-based People Management, and so forth. The complete AI-based solution can be seen by visiting the official prose website at https://prosa.ai/ or by emailing business@prosa.ai.

None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None