Why is the number of imported Phrases/Glossary different from the actual number of registrations?

If there is the same sentence or a term in the registered data, it will be overwritten with a new translation.