Analysis of sentiment optimization on social networks based on statistical data

As technology is much more developed, it has already merged with the daily life of people. Because of the characteristic of the network— no limitation, part of people believe that they are closer to their family and friends, while it makes people feel lonelier since companions are virtual. Therefore, it is easier to cause negative emotions and then lead to a more serious downside such as depression if those emotions cannot be alleviated or even eliminated on time. In this paper, the author focuses on the analysis of how to minimize the negative emotions so that to avoid more serious problems. All data are collected from COVID-19 Real World Worry Dataset which is related to Twitter. The author utilizes the comparison of P-value for each variable and Stepwise. Selection method to identify the most effective factor for causing negative emotions (anxiety, worry, fear, anger, disgust, and sadness). The author found that the frequency of participants on Twitter is the most influential variable. In other words, it is important to study ways to relieve negative emotions from Twitter emotions cannot be alleviated or even eliminated on time.