Using the modular property of BERTopic

Because BERTopic is modular, you can employ alternative techniques for a model. This section will show you the code structure and then explain what other techniques you can implement. First, you will import the following classes:

from umap import UMAPfrom hdbscan import HDBSCAN
from sentence_transformers import SentenceTransformer
from sklearn.feature_extraction.text import CountVectorizer
from bertopic import BERTopic
from bertopic.representation import KeyBERTInspired
from bertopic.vectorizers import ClassTfidfTransformer

Your BERTopic model, topic_model, is a collection of modules as follows. This view is clear because it tells us the components of BERTopic:

topic_model = BERTopic(        # word embedding
    embedding_model=embedding_model,
        # dimension reduction
    umap_model=umap_model,
   ...