Yext used an open-source machine-learning framework from Google to upgrade the natural-language processing (NLP) search algorithm that powers Yext Answers, Yext’s site search product.

Bidirectional Encoder Representations from Transformers (BERT) is designed to help NLPs better understand user searches.

Leveraging BERT within Named Entity Recognition, a process to locate and classify named entities mentioned in unstructured text into predefined categories, allows Yext Answers to improve its ability to distinguish locations from other types of entities, such as people, jobs, and events.

Marc Ferrentino, chief strategy officer at Yext, says today’s announcement marks the company’s latest major upgrade to the Answers algorithm, which leverages BERT.

Milky Way is the name of this update for the algorithm that powers Answers. Each future update will have a “galaxy” theme name. It’s similar to the candy-theme names that Google gives Android such as Lollipop and KitKat.

The English language is complicated. The word “cougar” could refer to a car, a woman dating a younger man, or an animal. “Turkey” could refer to a bird, food or a country.

Yext’s Milky Way update includes improved named entity Recognition to better understand the contextual relationship between search terms.

When someone uses a search term that can refer to multiple things, Answers will return a more relevant result by taking into account the correct classification, whether a location, person or product.

The update also include an improvement in location detection. It removes location biasing, which requires information like popularity and proximity to identify the location for which a person searches.

Yext Answers now filters through locations stored by a business in its knowledge graph — a database of millions of structured facts — to surface the best match. This is useful if a search term could refer to multiple places such as Paris, Texas vs. Paris, France.

The health-care taxonomy was also updated.

More than 3,000 new healthcare-related synonyms, conditions, treatments, and procedures have been added to the algorithm’s taxonomy. Whether a patient is searching for ailments in layperson’s terms like “pink eye” or a provider is searching with clinical terms like “conjunctivitis,” Yext Answers can understand the context and deliver the best answer.

The update also introduces improved stemming — NLP’s ability to recognize different forms of a root word — and fix typographical errors to better match a search term with results that include variations of that term, such as “integrate” and “integrations.”