Google has been employing a technique known as neural matching to gain a better understanding of synonyms, reportedly affecting 30% of queries.
According to Danny Sullivan from Google, the search engine has been using neural matching for several months now.
"Over the past few months, Google has been using neural matching—an AI method to better connect words to concepts. It’s like having super synonyms, impacting 30% of queries. Don’t know what ‘soap opera effect’ means to search for it? We can better figure it out."
Sullivan is discussing how neural matching allows Google to provide results related to the "soap opera effect" when a user queries something like "why does my TV look strange?"
This is noteworthy because "why does my TV look strange?" seems unrelated to soap operas, yet Google can still understand what the user might be searching for.
The neural matching technique was highlighted today during Google’s 20th anniversary event, where various other search features were also announced.
Thanks to advancements in AI like neural matching, these new search features become feasible.
Activity Cards
Google’s new activity cards will help users resume their search journeys.
When a user enters a query related to a previous one, Google will present a card with relevant pages already visited, along with the queries that led to those pages.
Activity cards will be available in search later this year and will appear only when deemed useful.
Search Collections
Collections in Google Search will let users manually track content they’ve visited, such as websites, articles, or images.
Content found in Activity Cards can also be added to collections.
To help discover new content, Google will suggest based on saved items in collections.
This feature will launch in the fall.
Dynamic Organization of Search Results
Google is introducing a dynamic way to organize search results to guide users on what to explore next.
"Rather than presenting information within fixed categories, we can intelligently show the subtopics most relevant to your search and make it easy to explore information from the web, all with a single search."
For example, if a user searches for Pugs, they’ll see dynamically generated subtopics like breed characteristics and names.
These tabs will stay updated, reflecting the most relevant aspects of the topic.
New Topic Layer in the Knowledge Graph
The new developments are complemented by a new layer in Google’s Knowledge Graph, referred to as the Topic Layer.
The Topic Layer is designed to deeply understand a topic and how interests evolve over time.
"The Topic Layer is built by analyzing all existing web content for a topic, developing thousands of subtopics. For these, we identify the most relevant articles and videos… We then analyze patterns to understand subtopic relationships, allowing us to more intelligently suggest content you might explore next."
With this technology, Google aims to meet users’ ongoing information needs into the future.
