The Advanced Settings allow you to configure whitespace, which adjusts the search query to sort and match multi-term entries correctly.
- Cross Field Factor: The degree to which the score is adjusted to sort exact vs multi-field matches. Any number between 0 and 1 may be entered, up to two decimal places.
- Split on Whitespace Behavior (SOW): Allows "term-centric" text analysis that is invoked separately for each whitespace-separated word. This means that the individual terms only need to exist across any given fields to be considered a match.
- When No Multi-Term Synonym: Utilize SOW analysis only when there are no multi-term synonyms defined. Any new search configurations that are created will be set to this option by default.
- If enabled, then you can add up to three values to the Multi-Term Synonym MinMatch field. To do so, click the Add icon and enter a number from 0 to 100. This will apply MinMatch behavior to multi-term scenarios.
- Never: Never utilize SOW analysis. The text analysis will be "field-centric", where whitespace-separated sequences will be provided to text analysis as one term. This enables multi-word synonyms, but also means that one single field must contain all of the terms in order to be treated as a match.
- Always: Always utilize SOW analysis. This means that multi-term synonym expansion cannot take place, as the field analyzers will only look at individual terms instead of the entire query.
- When No Multi-Term Synonym: Utilize SOW analysis only when there are no multi-term synonyms defined. Any new search configurations that are created will be set to this option by default.
Note that even with SOW enabled, there is still relevancy weight given to phrases where the entire query is found in a single field. This "phrase boosting" is usually based on the entire query typed by the user, but you can use two-word phrase boosting to allow any two adjacent words in the phrase to be considered a partial phrase match and boost relevancy of that result.
Whitespace Example
Example search term: "blue suede shoes"
Example product with attributes:
- color: "blue"
- material: "suede"
- product_name: "men’s high comfort casual shoes"
Without SOW, the query is not tokenized and relevancy will be determined based on the entire term “blue suede shoes.” This means that a 100% match must have one single field that contains all three terms, so using a high MinMatch threshold will result in the product not being returned as a match.
With SOW enabled, the query is tokenized and relevancy will be analyzed for “blue” and “suede” and “shoes." This means that the product will be successfully returned as a match.