Open AccessPosted prediction in social media base on Markov chain model: twitter dataset with covid-19 trends
Author(s) -
Windu Cahyaningrum Handayani Notonagoro Suryaningrat,
Devi Munandar,
A Maryati,
A S Abdullah,
Budi Nurani Ruchjana
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1722/1/012001
Subject(s) - social media , crawling , markov chain , computer science , field (mathematics) , covid-19 , microblogging , distribution (mathematics) , big data , data mining , mathematics , machine learning , world wide web , medicine , pure mathematics , anatomy , mathematical analysis , disease , pathology , infectious disease (medical specialty)
The influence of social media is very attractive in disseminating information; even social media analysis is one of the focuses in the field of research in terms of data mining. In its development not only the field of social science that exists but many studies of social media that can be solved stochastically to calculate the trend of the emergence of a discussion on social media. In this paper, we investigated calculations and predictions using Markov Chains on the emergence of discussions on Twitter media related to coronavirus disease tweets or better known as covid-19. The tweet data obtained is a random sample of the tweet posts that are crawled at the specified time. The tweet data is crawled at three different observations each day for thirteen days continuously. The results of data crawling are calculated to determine the transition from one observation to the next observation. The stages of the process are; crawling tweet data with keywords coronavirus and covid-19; data cleaning process; data processing; Markov Chain modeling; n-step distribution and long-term prediction; interpretation of results. The computational results used are opportunity distribution conditions for the number of tweets. As a transition between two states, namely low (0) and high (1) relative to mean or median. The results of the opportunity distribution obtained in the next 145-time steps (0.28767, 0.71233) and (0.47368, 0.52632) in the probability distribution of the number of tweets are respectively the mean and median values. The results of the modeling show that the conversation on Twitter for 145-time steps in the next prediction is estimated to remain high along with the outbreak of coronavirus or covid-19 before this epidemic subsides.