When choosing two topics to merge (if they are similar), how does BERTopic aggregate them?

[tawfiqam]

I am considering merging topics that are similar. I am wondering how BERTopic aggregates the topic probabilities?

[MaartenGr]

It depends on the type of probabilities you have selected. If you have set calculate_probabilities=False, then flat topic probabilities are generated. These probabilities indicate the probability of a document belonging to its predicted topic. When merging topics, these probabilities are not changed.

If you have set calculate_probabilities=True, then a topic-document probability matrix is generated instead of flat probabilities. In this matrix, you can see the probability of a document belonging to any topic. When merging topics, these probabilities are then added to one another since we have information about the topics they are merged together with.

EDIT: Do note that there is a new version coming up that will fix some of the issues when merging topics.

[tawfiqam]

Fantastic! Thank you very much.