Yext is constantly improving the Answers algorithms to support the interesting new use cases our customers are finding for Document Search. This means we are fine-tuning our models to optimize its performance across additional document types and styles. With the June release, we have trained the Extractive QA algorithm to understand rich text, and improve its ability to surface featured snippets for things like ordered lists or instruction steps.
These algorithm improvements will be especially useful for Support Search, where content from sources like Zendesk are often stored as rich text. Some other common types of unstructured documents with rich text include recipes, instructions, biographies, blog posts, articles, and ordered lists.