Parquet file with embeddings to be ingested, will require the following columns: review_id review_text user_id book_id embedding