00000nam a22000004a 4500 UP-1686042739784911759 Buklod 20240520133716.0 m |o d | ta 240426s2024 xxu r |||| u|eng d (Buklod)UP-1685954869149757829 DSTC rda DSTC eng LG 995 2024 S8 A73 Arcala, Marianne Angela D. author Identifying Changes In Trends of Twitter Conversation About Covid-19 Vaccines in the Philippines Marianne Angela D. Arcala. Quezon City University of the Philippines January 2024 120 pages with illustrations ABSTRACT This research identified the changes in trends of Twitter conversation about COVID-19 vaccines in the Philippines and relate these changes to key events, news, or vaccination campaigns. Exploratory data analysis using word cloud and term weighting were conducted to describe the data as well as four purposes of pre-processing. Latent Dirichlet Allocation (LDA) was used to identify topic. Since the length of the documents (or in this case, tweets) plays an important role in the performance of LDA, applying this technique to short texts like tweets results in inaccurate topics. In this paper, the tweets were grouped before applying LDA to overcome the limitation of the algorithm. To analyze sentiments, a lexicon-based approach based on positive and negative counts. Change point analysis was also performed on the results of the topic model and sentiment analysis. Results from this project suggest that a well-executed public campaign is highly beneficial in steering the conversations and managing the perception, at least in the same platform. The approach and results of this project may be used as a model or prototype on how to assess high-impact events/information and to monitor public conversations/perceptions in a social medial platform. COVID-19 Vaccine. Twitter. Topic Detection. Topic Modelling. Latent Dirichlet Allocation. Sentiment Analysis. Lexicon-based. Change Point Analysis. FI Thesis. UPD DSTC LG 995 2024 S8 A73 Book