• by Laurie Sullivan , Staff Writer @lauriesullivan, November 23, 2020

What if between every question the person in conversation lost their train of thought and forgot the topic being discussed? Until recently, that’s kind of what would happen in Google Search.

“Many Google searches are part of a longer series of queries, but Google didn’t always carry over the context from one query to another,” writes Bobby Weber, Search PM.

It occurred especially when the next query had multiple interpretations. The results might have seemed a little off, leading the person searching to spend more time rewriting the query until they found what they wanted. 

Google this year introduced new ways to get those querying the information they need, using context from their recent search activity. It’s an improvement in what Weber calls “language understanding capabilities” that now make it easier for people to get to a more specific, on-topic search.

Search in the past were largely siloed. If someone searched for “turkey recipes” and then “carving,” Google would do nothing to associate or personalize “carving” to turkey recipes,” which seems kind of odd by itself. Google Search would use the most common interpretation in ranking results. Semantics didn’t seem to come into play.

Now, doing a search on the word “carving” after a turkey-related query offers up a new search for “carving turkey” that is more relevant than just offering a definition. With the latest change, Google can determine that some people are searching to learn more about preparing and serving a turkey.

This new understanding of language also can identify when someone is exploring a topic. Perhaps someone is searching for a family-type movie like "The Polar Express." Google may detect that they are exploring related ideas. Results would show a list of similar movies to help someone browse for the perfect pick.

This understanding of context also enables Google to surface more relevant follow-up questions to help take people to the next search. 

Prior to introducing these features, if you had searched for “how to make a napkin fan,” Google might have helped you discover additional information by showing you similar questions such as “How do you make a cone napkin?”

Weber explains that the last update will surface more relevant questions under the “People also ask” list.

Prior to introducing these features, if someone searched for things like "how to make a napkin fan," Google might have helped the person discover additional information by showing them similar questions such as “How do you make a cone napkin?”

With contextual learning, Google will source more relevant questions in the “People also ask” section, like “How do you make a turkey out of a cloth napkin?”

It seems like a simple task. Google should have been doing this all along.