Hybrid search refers to the merging of search technologies, but the technology used to create hybrid search engines is evolving.

Here is where it is today.

Traditionally, hybrid search referred to a mix of directory-based and crawler-based search engines.

Now the meaning changes to refer to a combination of artificial intelligence (AI) and neural hashing.

AI replaces the complex rule-based understanding of traditional site search, while neural hashing makes vector search as fast as keyword search.

It might sound a bit complicated, but the result is effortless user research.

Today’s hybrid search technology allows a query like “stay cool in the summer” on an electronics store website to produce instant results for fans and air conditioners.

Without hybrid search, results may be limited to products containing words used in the query.

In addition to providing greater speed and relevance, hybrid search is more accessible for businesses to implement on their own websites.

A company called Search.io is one organization that is paving the way for a new future for hybrid search, and I recently had the opportunity to speak to CEO and co-founder Hamish Ogilvy.

He updated me on recent developments in this area and how his company’s new tool makes it easier for customers to search company websites. This, in turn, can lead businesses to make more sales.

AI + neural hashing = modern hybrid research

In search, AI replaces strict keyword matching with dense vectors encapsulating text meaning. It has proven to be superior to keywords in terms of relevance, but comes with slower results.

Neural hashing, sometimes referred to as “deep hashing,” helps make vector search as fast as keyword searching. It gets its name from its ability to use neural networks to hash vectors.

Neural hashes compare terms using mathematical expressions. They measure the difference between words and concepts and assign meaning to those that are closer to each other.

What does this mean in practice?

Ogilvy explains:

“In practical terms… AI language understanding can be easily deployed in search technology. Ironically, for many queries, despite the vastly improved relevance, it’s actually faster than keyword research as well.

Neural hashing delivers 99% of the performance of dense vector searching while being over 100x faster and using a fraction of the space.

Ogilvy showed me several examples of natural language queries with and without neural hashing to show how modern solutions are failing retailers and customers.

A query like “something to keep my beer cold” returns a single, irrelevant result on retailer Best Buy’s website. By comparison, the same query with neural hashes applied would return a results page full of beer fridges.

It could be the difference between a customer leaving empty-handed and a business making a quick sale.

Hybrid search – more accessible than ever for businesses?

Search.io is launching a tool called Neuralsearch, which would combine the speed of traditional keyword research with the precision of vector search.

Simply put, it allows websites to return results more like Google does, without employing an army of search engineers.

It eliminates the need for retailers to add synonyms to their search index, which is time-consuming that can be better spent elsewhere.

Neuralsearch is now in public beta, after private testing with select organizations. Businesses can add it to their sites for free with a 14-day trial.

For an example of how this works in action, Ogilvy tells me that the following company websites already use Neuralsearch:

• Unity.com
• BBC.com
• LockheedMartin.com
• Sennheiser.com
• Catch.com.au
• RentPath.com

---

Source: search.io
Featured Image: LookerStudio/Shutterstock