Analysing ChatGPT User Tweets Using Hierarchical Clustering Technique

Artifical Intellgience (AI) Tools as Chatgpt is increasing in our digital world. Chatgpt made a significant impact once introduced by openAI company. Chatgpt capablities impress the world and changes the concept of text writing and processing. This study will determine the perceptions of twitter user in using Chatgpt to understand whether they accept or reject Chatgpt application.The study also will define most frequent keywords in each cluster after analysing the user tweets collected. Tweets will be anlayzed to understand the opinion of users on using Chatgpt through text mining methods. Chatgpt sentiment model will be proposed, the model will pass by the following steps: extraction, cleaning, and clustering. Using hierarchical cluster analysis, the tweets were clustered into five clusters: Like, dislike, not sure, game, fantasy. After using the elbow technique, it minimized clusters to only three clusters: Like, Dislike, and neutral. Chi-squared test was used to define the features that should be selected in the cluster. The results showed that positive opinions were 53,743 and negative opinions including 44,207. Sentiment analysis disclosed the acceptance of using Chatgpt, users understand the Chatgpt emerging benefits and its valuable support. Hierarchical cluster analysis revealed that the words reflecting the most frequently used words in “Like” cluster are “like” 8394, “help” 3722, “power” 3021, “good” 2905 and “ease” 2416, also the most frequently used words in “Dislike” cluster are “low” 3865, “horrible” 3344, “down” 2894, “threat” 2024 and “break” 914.