Effekten av sentimentanalyse i sosiale medier

Abstract

This dissertation examines the effect of sentiment analysis on social media engagement, with a focus on Norwegian crisis centers. We analyzed 164 posts from three crisis centers: Stavanger, Molde, and Bodø using The Lexical Suite to measure sentiment variables such as valence, extremity, emotionality, and certainty. ANOVA tests were conducted to assess the differences in sentiment across centers, as well as correlation tests to see how the variables interacted with each other. Additionally, a Poisson regression was also conducted to further view how the variables influenced the number of likes received per post. The findings suggested that higher levels of positive valence and consistent emotions are associated with an increase in likes highlighting the impact of positive and neutral sentiment in promoting engagement. Furthermore, posts with either low or high certainty levels have interacted with more likes, meaning that keeping an overall assertive sentiment does not impact likes positively. Notably, extremity and emotionality did not show any significant correlations with likes, which tells us that an excess of strong emotions and words, whether positive or negative, did not necessarily increase engagement levels. What can be concluded from this study is that a strategically planned management of sentiment in posts can improve the overall engagement that the crisis centers receive on their Instagram communications, which most likely could be applied to other social media platforms. For institutions such as crisis centers that have an inherent sensitive nature, it can be especially relevant to provide a basis for more effective digital communication strategies.

This dissertation examines the effect of sentiment analysis on social media engagement, with a focus on Norwegian crisis centers. We analyzed 164 posts from three crisis centers: Stavanger, Molde, and Bodø using The Lexical Suite to measure sentiment variables such as valence, extremity, emotionality, and certainty. ANOVA tests were conducted to assess the differences in sentiment across centers, as well as correlation tests to see how the variables interacted with each other. Additionally, a Poisson regression was also conducted to further view how the variables influenced the number of likes received per post. The findings suggested that higher levels of positive valence and consistent emotions are associated with an increase in likes highlighting the impact of positive and neutral sentiment in promoting engagement. Furthermore, posts with either low or high certainty levels have interacted with more likes, meaning that keeping an overall assertive sentiment does not impact likes positively. Notably, extremity and emotionality did not show any significant correlations with likes, which tells us that an excess of strong emotions and words, whether positive or negative, did not necessarily increase engagement levels. What can be concluded from this study is that a strategically planned management of sentiment in posts can improve the overall engagement that the crisis centers receive on their Instagram communications, which most likely could be applied to other social media platforms. For institutions such as crisis centers that have an inherent sensitive nature, it can be especially relevant to provide a basis for more effective digital communication strategies.