What are Synonyms | Yext Hitchhikers Platform
What You’ll Learn
In this section, you will learn:
- What synonyms are
- How synonyms are considered in Search
- The difference between one way synonyms and synonyms sets
What are Synonyms
“Synonyms are two or more words which have the same meaning.”
At a high level, Search can recognize when a query includes a synonym (or synonyms), and will take the synonyms into account when determining which results to return.
For example, let’s say a client had “Healthcare Professional” entities in their Knowledge Graph, but they know people generally refer to their professionals as “Doctors”. They can define a synonym from “Doctor” to “Healthcare Professional”. That way, if a user searches “Doctor”, Search will understand that this word means the same thing as “Healthcare Professional” and will return the correct results.
Synonyms allow us to translate tokens within a query (for more information on tokens, please see the Search Algorithm module ) to different variations that mean the same thing. Synonyms are set at the experience level since business rules vary from client to client. They are defined directly by the admin in the backend configuration and apply at the Universal level (i.e. across all Verticals!).
There are three options to consider when setting up a synonym:
One way Synonym: One way relationship between tokens where term A means the same as term B, but term B does NOT mean the same as term A.
“square” → “rectangle”
- A square is a rectangle, but a rectangle is not a square! You want searches for “rectangle” to return squares, but you don’t want searches for “square” to return any old rectangle.
- A square is a rectangle, but a rectangle is not a square! You want searches for “rectangle” to return squares, but you don’t want searches for “square” to return any old rectangle.
“Windy city” → “Chicago”
- If someone searches for “windy city”, you want to surface “Chicago” results in case that’s what they’re looking for. However, the query “windy city” might have a lot of noise and you wouldn’t want a query for “Chicago” to also be searching for the tokens “windy” and “city”. It would return a lot of false-positives.
“Reputation management” → “reviews” (Yext example!)
- In this case, reputation management is just one use case for Reviews. If you have two terms where one is more specific than the other, you probably want to use a one-way synonym.
- In this case, reputation management is just one use case for Reviews. If you have two terms where one is more specific than the other, you probably want to use a one-way synonym.
- Synonym Set: Defines synonym groupings – e.g., multiple terms that all mean the exact same thing.
- [“location”,”branch”]
- [“doctor”,”dr”,”physician”,”healthcare professional”]
- [“customer success story”,”case study”,”customer showcase”]
- In these cases, we would want all terms to be considered equivalent and to return the same results. Searching “location” will yield results for “location”/”branch”, and searching “branch” will ALSO yield the same results for “location/”branch”.
- Normalization Synonyms: The inverse of a one way synonym where a set of many terms is mapped to one term.
- “Physician”,“MD” → “doctor”
- Doctors may be reffered to in multiple ways (e.g., nurse, physician, MD) and you want to be able to capture results for all doctors when searching those various terms.
- “Physician”,“MD” → “doctor”