Quick Index Field Options

reference

When you create a Search index with the Quick Editor, you must set options for each field you add to the index.

For more information about how to create a Search index with the Quick Editor, see Create a Search Index with the Quick Editor.

The following options are available for fields in the Quick Editor:

Option Description

Index this field as an identifier (Text Fields Only)

To index this field with the keyword analyzer, select Index this field as an identifier.

To use other analyzers or settings on this text field, clear Index this field as an identifier.

For more information about analyzers, see Customize a Search Index with the Web Console

Language (Text Fields Only)

Select the language for the content inside a text field.

The Search Service automatically applies an analyzer to the field’s contents based on the selected language.

For more information about the available language options, see Quick Index Supported Languages.

Dimension (Vector Fields Only)

For a vector child field, enter the total number of elements in the vector embedding array.

From Couchbase Server version 7.6.2 and later, Vector Search indexes can support arrays with up to 4096 elements. Arrays can be an array of arrays.

For more information about Vector Search indexes, see Use Vector Search for AI Applications or Create a Vector Search Index with the Server Web Console.

Similarity Metric (Vector Fields Only)

For a vector child field, choose the method to calculate the similarity between the vector embedding in a Vector Search index and the vector embedding in a Vector Search query.

It’s recommended to choose the same similarity metric for your Search index as the one used in your embedding model.

dot_product: Calculated by adding the result of multiplying a vector’s components, or the product of the magnitudes of the vectors and the cosine of the angle between them. The dot product of 2 vectors is affected by the length and direction of each of the vectors, rather than just taking a straight-line distance.

Dot product similarity is commonly used by Large Language Models (LLMs). Use dot_product to get the best results with an embedding model that uses dot product similarity.
l2_norm: Also known as Euclidean distance. Uses the straight-line distance between 2 vectors to calculate similarity. Smaller euclidean distances mean that the values of each coordinate in the vectors are closer together.

It’s best to use l2_norm similarity when your embeddings contain information about the count or measure of specific things, and your embedding model uses the same similarity metric.
cosine: From Couchbase Server version 7.6.4 and later, the cosine similarity metric is calculated by adding the result of multiplying a vector’s components, or the product of the magnitudes of the vectors and the cosine of the angle between them. This metric is not affected by the size of the vectors being measured.

Use cosine similarity to get the best results with an embedding model that uses cosine similarity. Cosine similarity works well for semantic search, document classification, and recommendation systems.

The Search Service will normalize any vectors in your documents before indexing when using cosine similarity. It will also normalize any vectors in your queries if the field for those queries uses cosine similarity. Use dot_product similarity if your vectors are already normalized.

For more information about Vector Search indexes, see Use Vector Search for AI Applications or Create a Vector Search Index with the Server Web Console.

Optimized For (Vector Fields Only)

For a vector child field, choose whether the Search Service should prioritize recall, latency, or memory efficiency when returning similar vectors in search results:

recall: The Search Service prioritizes returning the most accurate result. This may increase resource usage for Search queries.

The Search Service uses an nprobe value to calculate the number of centroids to search when using recall priority. This value is calculated by taking the square root of the number of centroids in the index.
latency: The Search Service prioritizes returning results with lower latency. This may reduce the accuracy of results.

The Search Service uses half the nprobe value calculated for recall priority.
memory-efficient: From Couchbase Server version 7.6.4 and later, choose this option to prioritize reducing memory usage and optimize search operations for less resources. This may reduce both accuracy (recall) and latency.

The Search Service uses either an inverted file index with scalar quantization, or a directly mapped index with exact vector comparisons, depending on the number of vectors in your data.

For more information about Vector Search indexes, see Use Vector Search for AI Applications or Create a Vector Search Index with the Server Web Console.

Include in search results

To include content from the field in search results, select Include in search results.

To exclude the field’s content from search results, clear Include in search results.

Support highlighting

The Search Service can highlight matching search terms in search results from an index.

To enable highlighting in search results, select Support highlighting.

To turn off highlighting in search results, clear Support highlighting.

To enable Support highlighting, you must also enable Include in search results.

Support phrase matching

To support searches for whole phrases, select Support phrase matching.

To turn off phrase matching, clear Support phrase matching.

Support field agnostic search

To search the field’s contents without specifying the field name in a search query, select Support field agnostic search.

To turn off field agnostic search, clear Support field agnostic search.

Support sorting and faceting

To sort search results and use facets with the field’s contents, select Support sorting and faceting.

To turn off sorting and facets, clear Support sorting and faceting.

Searchable As

Set a different name that you can use to search the field’s contents in a query.

The default value is the field’s name.