I unfortunately did not find a good field to sort by in the original dataset. So it's just sorted by text match relevance scores, then the order that they appear in the dataset.