What goes into a great search experience?

The Elements of Search [Infographic]

Are you still adding synonyms to your search index? You shouldn’t have to anymore. With newer AI models, it’s no longer needed. Real AI (as opposed to fake AI) eliminates the need to manually add synonyms, use synonym generators, or mine clickstream and search data for synonyms. 

  • No more creating synonyms for general terms like “pants,” “trousers,” or “slacks.”
  • You also no longer need to understand the difference between one-way and two-way synonyms. 
  • You don’t need to tweak synonyms to improve search queries. 
  • And, you don’t need to worry about polysemy — words and phrases that have many different meanings. 

AI solves synonyms

Traditional search engines need to be told that the terms are equivalent. Companies have spent huge amounts of time writing synonym libraries or search rules to overcome the limitations of keyword search. 

Newer AI models can all do that for you without all of the manual effort. In fact, vector-based search has been around since 2013 and enabled search to understand the meaning of queries with synonyms — but it never took off due to speed, scale, and cost issues.

We released our own AI search last December built on neural hashing technology which delivered the same great results with none of the costs, scaling constraints, and overhead.  

Neuralsearch understands concepts so, without any training, Neuralsearch knows that “pants” and “trousers” are the same thing.  It also knows that someone searching for “jeans” may be interested in “pants” but are less likely to be interested in “trousers.” 

This is different from using a technology like word2vec to generate common synonyms for your site index. Word2vec solutions are a popular feature for many search engine companies; they generate a list of synonyms which are then added to your index. However, today that's one step too many! Neuralsearch doesn’t generate new synonyms — it understands synonyms from the get-go.  

How it works

Neural hashes uses mathematical expressions to compare terms. The distance between words or concepts can be measured. Terms that are closer together have more similar meanings. 

In the example above, the query contains a misspelling and the term “vertical.” Traditionally we would need to manually add a synonym for “vertical” to “upright” and some type of spelling or fuzzy matching to pick up the spelling issue (“cc” vs “uu”). With Neuralsearch, this and many other corner cases simply just work out of the box. 

Synonyms allow people to express the same ideas with different words. For example, “Waterproof clothing” and “Gortex jacket” contain similar ideas. “Jacket” is related to clothing, and “Gortex” is related to the idea of “waterproof.” 

There are other similar words that connect conceptually, too. “Rain,” “wet,” and “winter” are terms that are related to “water-proof jacket.” With concept search, someone searching for “raincoat” (or “rain coat” or “winter parka for winter”) can get equally good results as any other similar search. 

In an example from word2vec, if the vector representation of “man” is subtracted from “king” and then “woman” is added, we get a vector very similar to “queen”. This same idea built into Neuralsearch enables us to classify words — even if we’ve never encountered them before. Image credit: TowardDataScience

Manually creating synonyms doesn’t seem that bad for small catalogs, but for larger sites, it’s painfully slow. Sites are constantly changing. Many companies only create synonyms for the top 10% or 20% of their catalog because it would be too expensive to do it for all of them. Real AI supports long-tail searches equally well as “head” queries so 100% of your catalog is supported. 

Neuralsearch gives you your time back to work on building your business. If you’re still writing synonyms, it’s time to make the change. Sign up for a free 14-day trial or contact us for a custom demo for your business use-case.

site search CTA

Similar articles

User Experience (UX)
Best Practice

Drastically Improve Your Website's UX by Improving Site Search

Best Practice

An Introduction to Custom Search Engines

Best Practice

How to Get a Competitive Edge with Smart (and Surprisingly Easy) E-commerce Site Search Solutions