We use custom BERTopic hyperparameter which is indo-sentence-bert-base for the embeddings, KMeans for the clustering, and KeyBERTInspired for the representatin model.