Semantic text field type

The semantic_text field type accepts text only. To index or search images, audio, video, or PDF files, use semantic with a compatible multimodal embedding endpoint.

For text-based semantic search, semantic_text provides sensible defaults that automate most of the manual work typically required for vector search. You don't have to manually configure mappings, set up ingestion pipelines, or handle chunking. The field type automatically:

• Configures index mappings: Chooses the correct field type (sparse_vector or dense_vector), dimensions, similarity functions, and storage optimizations based on the inference endpoint.
• Generates embeddings during indexing: Automatically generates embeddings when you index documents, without requiring ingestion pipelines or inference processors.
• Handles chunking: Automatically chunks long text documents during indexing.

Elasticsearch refers to semantic and semantic_text as inference fields: mapped fields that use inference endpoints and store generated embeddings in internal subfields.

Elasticsearch provides two field types that generate and store embeddings automatically. Choose the field type based on the content and embedding model you want to use.

semantic_text accepts text only. To index or search images, audio, video, or PDF files, use semantic with a compatible multimodal embedding endpoint.

The following example creates an index mapping with a semantic_text field, using default values:

PUT semantic-embeddings 
{
  "mappings": { 
    "properties": {
      "content": { 
        "type": "semantic_text"
      }
    }
  }
}

The following example creates an index mapping with a semantic_text field that uses dense vectors:

PUT semantic-embeddings
{
  "mappings": {
    "properties": {
      "content": {
        "type": "semantic_text",
        "inference_id": "my-inference-endpoint",
        "search_inference_id": "my-search-inference-endpoint",
        "index_options": {
          "dense_vector": {
            "type": "bbq_disk"
          }
        },
        "chunking_settings": {
          "strategy": "word",
          "max_chunk_size": 120,
          "overlap": 40
        }
      }
    }
  }
}

1. (Optional) Specifies the inference endpoint used to generate embeddings at index time. If you don’t specify an inference_id, the semantic_text field uses a default inference endpoint.
2. (Optional) The inference endpoint used to generate embeddings at query time. If not specified, the endpoint defined by inference_id is used at both index and query time.
3. (Optional) Configures how the underlying vector representation is indexed. In this example, bbq_disk is selected for dense vectors. You can configure different index options depending on whether the field uses dense or sparse vectors. Learn how to set index_options for sparse_vectors and how to set index_options for dense_vectors.
4. (Optional) Overrides the chunking settings from the inference endpoint. In this example, the word strategy splits text on individual words with a maximum of 120 words per chunk and an overlap of 40 words between chunks. The default chunking strategy is sentence.

The semantic_text field type documentation is organized into reference content and how-to guides.

The Reference section provides technical reference content:

• Parameters: Parameter descriptions for semantic_text fields.
• Inference endpoints: Overview of inference endpoints used with semantic_text fields.
• Chunking: How semantic_text automatically processes long text passages by generating smaller chunks.
• Pre-filtering for dense vector queries: Automatic pre-filtering behavior for dense vector queries on semantic_text fields.
• Limitations: Current limitations of semantic_text fields.
• Document count discrepancy: Understanding document counts in _cat/indices for indices with semantic_text fields.
• Querying semantic_text fields: Supported query types for semantic_text fields.

The How-to guides section organizes procedure descriptions and examples into the following guides:

• Set up and configure semantic_text fields: Learn how to configure inference endpoints, including default endpoints, Jina on EIS, ELSER on EIS, third-party models on EIS, external inference, and dedicated endpoints for ingestion and search operations.

• Ingest data with semantic_text fields: Learn how to index pre-chunked content, use copy_to and multi-fields to collect values from multiple fields, and perform updates and partial updates to optimize ingestion costs.

• Search and retrieve semantic_text fields: Learn how to query semantic_text fields, retrieve indexed chunks, return field embeddings, and highlight the most relevant fragments from search results.