Please use the form below to provide your feedback. Because your feedback is valuable to us, the information you submit in this form is recorded in our issue tracking system (JIRA), which is publicly available. You can track the status of your feedback using the ticket number displayed in the dialog once you submit the form.

Child Field Options

  • reference
    +
    You can configure a child field from the Couchbase Server Web Console to control how its data is added to your Search index.

    For more information about how to create a child field, see Create a Child Field.

    You can configure the following options for a child field, based on its type:

    Option Description

    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.

    Searchable As

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

    The default value is the value set in the Field field when you create a child field.

    Analyzer

    Choose an analyzer for the content in the child field.

    You can choose a default analyzer or a custom analyzer.

    For more information about the available default analyzers, see Default Analyzers.

    For more information about how to create a custom analyzer, see Create a Custom Analyzer.

    Index

    To include the child field in the index, select Index.

    To remove the child field from the index, clear Index.

    If you remove the child field from the index, it cannot appear in search results.

    Store

    To store the values from the field in the index and return them in search results, select Store. This increases your index’s size and indexing time.

    You can return the values from this field when you include the fields array in a Search query.

    To remove the field’s values from the index, clear Store.

    Include in _all field

    The _all field is a composite field that has the content from multiple fields in an index. It allows searches to query the content of a field without specifying the field’s name. This option increases your index’s size.

    To include this field in the _all field, select Include in _all field.

    To exclude this field from the _all field, clear Include in _all field.

    To change the name of the _all field, see Default Field.

    Include Term Vectors

    Term vectors store the location of terms in a field for an index. You can use term vectors to highlight matching search terms in search results, and perform phrase searches. Term vectors increase your index’s size and indexing time.

    To enable term vectors for this field, select Include Term Vectors.

    To turn off term vectors, clear Include Term Vectors.

    To enable term vectors, you must also enable Store.

    Doc Values

    Use doc values for a field for Search Facets and sorting your search results.

    To store doc values and enable faceting and sorting for this field, select Doc Values.

    To exclude doc values from the index for this field, clear Doc Values.

    See Also