What is Synonyms in search?

When a shopper types a word that differs from the terms used in product data, synonym rules bridge that gap. Without them, a catalog full of 'trainers' returns nothing when someone searches 'sneakers', even though the products are identical. Synonym handling can be one-way (a maps to b) or two-way (a and b are fully equivalent), and it applies at query time without changing the underlying product data.

How does synonyms in search work?

• A search engine receives the raw query and checks it against a synonym dictionary or expansion rules.
• Matching words are expanded or replaced with their equivalents before the query runs against the index.
• The engine then retrieves and ranks results as if the shopper had used the catalog's preferred terminology.
• Some systems apply synonyms at index time instead, writing multiple word forms into the index entries themselves.

Why does it matter?

Shoppers use the language they know, not the language a catalog was built with. Missed synonyms mean missed revenue - a visitor who sees zero results is likely to leave. For dealerships, the same problem appears when a buyer searches 'used trucks' but inventory is labeled 'pre-owned pickups'. Closing those gaps directly lifts conversion without touching product data or ad spend.

Nobi's semantic matching reduces the burden of hand-maintaining synonym lists by understanding query intent rather than relying on exact word matches.

Frequently asked questions

What is the difference between a synonym and a stemming rule? Stemming normalizes word forms - 'running', 'runs', and 'ran' all reduce to the same root. Synonyms map entirely different words to one another, such as 'couch' and 'sofa'. Most search engines apply both techniques independently.

Should synonyms be one-way or two-way? It depends on the relationship. 'TV' and 'television' are true equivalents, so two-way works. But 'dress' and 'gown' are not always the same thing, so a one-way rule - 'gown' expands to also search 'dress' - is safer and avoids surfacing unrelated results.

How many synonym pairs does a typical ecommerce site need? There is no universal answer, but high-traffic sites often maintain hundreds of pairs. The practical starting point is zero-result queries - any search that returns nothing is a candidate for a new synonym rule or a catalog data fix.