Categorical Embeddings in an Unsupervised Setting for Anomaly Detection

Question

Context: I am working on an unsupervised use case. The Dataset I have has the following fields: TimeStamp, UserName and eventName Eg. User A has done Event