<ul><li>Tweets were processed and tagged with emotional tone using the facebook/bart-large-mnli model.</li><li>Each tweet was converted into a numerical vector using the SentenceTransformers model to capture context and meaning.</li><li>KMeans clustering was applied to detect latent structure in tweets based on the semantic embeddings.</li><li>UMAP was used to visualize the tweet clusters and sentiment layout in a 2D space for further analysis.</li></ul>

Case Study: An Unsupervised AI Pipeline to Cluster Tweets by Emotion and Meaning

Discover more