AppSearchSchema.EmbeddingPropertyConfig.Builder

Builder(propertyName: String)

Creates a new EmbeddingPropertyConfig.Builder.

fun build(): AppSearchSchema.EmbeddingPropertyConfig

Constructs a new EmbeddingPropertyConfig from the contents of this builder.

fun setCardinality(cardinality: Int): AppSearchSchema.EmbeddingPropertyConfig.Builder

Sets the cardinality of the property (whether it is optional, required or repeated).

If this method is not called, the default cardinality is CARDINALITY_OPTIONAL.

@RequiresFeature(enforcement = "androidx.appsearch.app.Features#isFeatureSupported", name = Features.SCHEMA_SET_DESCRIPTION)
@ExperimentalAppSearchApi
fun setDescription(description: String): AppSearchSchema.EmbeddingPropertyConfig.Builder

Sets a natural language description of this property.

For more details about the description field, see getDescription.

fun setIndexingType(indexingType: Int): AppSearchSchema.EmbeddingPropertyConfig.Builder

Configures how a property should be indexed so that it can be retrieved by queries.

If this method is not called, the default indexing type is INDEXING_TYPE_NONE, so that it will not be indexed and cannot be matched by queries.

@ExperimentalAppSearchApi
@RequiresFeature(enforcement = "androidx.appsearch.app.Features#isFeatureSupported", name = Features.SCHEMA_EMBEDDING_QUANTIZATION)
fun setQuantizationType(quantizationType: Int): AppSearchSchema.EmbeddingPropertyConfig.Builder

Configures whether the vector contents of this property should be quantized.

Quantization can reduce the size of the embedding search index, potentially leading to faster embedding search due to lower I/O bandwidth. Quantization is usually very reliable and in most cases will have a negligible impact on recall. Using quantization is strongly recommended.

If this method is not called, the default quantization type is QUANTIZATION_TYPE_NONE.