Typo tolerance

Typo tolerance helps users find relevant results even when their search queries contain spelling mistakes or typos, for example, typing phnoe instead of phone. You can configure the typo tolerance feature for each index.

Typo tolerance is enabled by default, but you can disable it if needed:

curl \
  -X PATCH 'http://localhost:7700/indexes/movies/settings/typo-tolerance' \
  -H 'Content-Type: application/json' \
  --data-binary '{ "enabled": false }'

With typo tolerance disabled, Meilisearch no longer considers words that are a few characters off from your query terms as matches. For example, a query for phnoe will no longer return a document containing the word phone.

We recommend keeping typo tolerance enabled for most uses. Massive or multilingual datasets may be exceptions, as typo tolerance can cause false-positive matches in these cases.

By default, Meilisearch accepts one typo for query terms containing five or more characters, and up to two typos if the term is at least nine characters long.

If your dataset contains seven, searching for sevem or sevan will match seven. But tow won't match two as it's less than 5 characters.

You can override these default settings using the minWordSizeForTypos object. The code sample below sets the minimum word size for one typo to 4 and the minimum word size for two typos to 10.

curl \
  -X PATCH 'http://localhost:7700/indexes/movies/settings/typo-tolerance' \
  -H 'Content-Type: application/json' \
  --data-binary '{
    "minWordSizeForTypos": {
      "oneTypo": 4,
      "twoTypos": 10
    }
  }'

When updating the minWordSizeForTypos object, keep in mind that:

• oneTypo must be greater than or equal to 0 and less than or equal to twoTypos
• twoTypos must be greater than or equal to oneTypo and less than or equal to 255

To put it another way: 0 ≤ oneTypo ≤ twoTypos ≤ 255.

We recommend keeping the value of oneTypo between 2 and 8 and the value of twoTypos between 4 and 14. If either value is too low, you may get a large number of false-positive results. On the other hand, if both values are set too high, many search queries may not benefit from typo tolerance.

You can disable typo tolerance for a list of query terms by adding them to disableOnWords. disableOnWords is case insensitive.

curl \
  -X PATCH 'http://localhost:7700/indexes/movies/settings/typo-tolerance' \
  -H 'Content-Type: application/json' \
  --data-binary '{
    "disableOnWords": [
      "shrek"
    ]
  }'

Meilisearch won't apply typo tolerance on the query term Shrek or shrek at search time to match documents.

You can disable typo tolerance for a specific document attribute by adding it to disableOnAttributes. The code sample below disables typo tolerance for title:

curl \
  -X PATCH 'http://localhost:7700/indexes/movies/settings/typo-tolerance' \
  -H 'Content-Type: application/json' \
  --data-binary '{ "disableOnAttributes": ["title"] }'

With the above settings, matches in the title attribute will not tolerate any typos. For example, a search for beautiful (9 characters) will not match the movie "Biutiful" starring Javier Bardem. With the default settings, this would be a match.

The typo ranking rule sorts search results by increasing number of typos on matched query words. Documents with 0 typos will rank highest, followed by those with 1 and then 2 typos.

The presence or absence of the typo ranking rule has no impact on the typo tolerance setting. However, disabling the typo tolerance setting effectively also disables the typo ranking rule. This is because all returned documents will contain 0 typos.

To summarize:

• Typo tolerance affects how lenient Meilisearch is when matching documents
• The typo ranking rule affects how Meilisearch sorts its results
• Disabling typo tolerance also disables typo

Meilisearch uses a prefix Levenshtein algorithm to determine if a word in a document could be a possible match for a query term.

The number of typos referenced above is roughly equivalent to Levenshtein distance. The Levenshtein distance between two words M and P can be thought of as "the minimum cost of transforming M into P" by performing the following elementary operations on M:

• substitution of a character (for example, kitten → sitten)
• insertion of a character (for example, siting → sitting)
• deletion of a character (for example, saturday → satuday)

By default, Meilisearch uses the following rules for matching documents. Note that these rules are by word and not for the whole query string.

• If the query word is between 1 and 4 characters, no typo is allowed. Only documents that contain words that start with or are of the same length with this query word are considered valid
• If the query word is between 5 and 8 characters, one typo is allowed. Documents that contain words that match with one typo are retained for the next steps.
• If the query word contains more than 8 characters, we accept a maximum of two typos

This means that saturday which is 7 characters long, uses the second rule and matches every document containing one typo. For example:

• saturday is accepted because it is the same word
• satuday is accepted because it contains one typo
• sutuday is not accepted because it contains two typos
• caturday is not accepted because it contains two typos (as explained above, a typo on the first letter of a word is treated as two typos)