Implementation:Lance format Lance Java InvertedIndexParams

Knowledge Sources	Lance
Domains	Java_SDK, Indexing
Last Updated	2026-02-08 19:33 GMT

Overview

Description

InvertedIndexParams is a final Java class in the org.lance.index.scalar package that provides a builder-style configuration for inverted (full-text) scalar index parameters. It constructs a ScalarIndexParams instance with the index type set to "inverted". The builder exposes a comprehensive set of text analysis options including tokenizer selection, language configuration, stemming, stop word removal, ASCII folding, N-gram settings, and position storage. All parameters are optional and serialized to JSON internally.

Usage

Use InvertedIndexParams.builder() to configure inverted index options. The build() method returns a ScalarIndexParams object. This is the primary way to configure full-text search indices in the Lance Java SDK, supporting multiple tokenizer types (simple, whitespace, raw, ngram, lindera, jieba) and language-aware text processing.

Code Reference

Source Location

java/src/main/java/org/lance/index/scalar/InvertedIndexParams.java

Signature

public final class InvertedIndexParams {
    public static Builder builder();

    public static final class Builder {
        public Builder baseTokenizer(String baseTokenizer);
        public Builder language(String language);
        public Builder withPosition(boolean withPosition);
        public Builder maxTokenLength(Integer maxTokenLength);
        public Builder lowerCase(boolean lowerCase);
        public Builder stem(boolean stem);
        public Builder removeStopWords(boolean removeStopWords);
        public Builder customStopWords(List<String> customStopWords);
        public Builder asciiFolding(boolean asciiFolding);
        public Builder minNgramLength(int minNgramLength);
        public Builder maxNgramLength(int maxNgramLength);
        public Builder prefixOnly(boolean prefixOnly);
        public Builder skipMerge(boolean skipMerge);
        public ScalarIndexParams build();
    }
}

Import

import org.lance.index.scalar.InvertedIndexParams;

I/O Contract

Builder Inputs

Parameter	Type	Required	Default	Validation	Description
baseTokenizer	String	No	Not set (Rust default: "simple")	Non-null, non-empty	Tokenizer identifier: "simple", "whitespace", "raw", "ngram", "lindera/*", "jieba/*"
language	String	No	Not set	Non-null, non-empty	Language for stemming and stop words (e.g., "English")
withPosition	boolean	No	Not set	None	Whether to store token positions in the index
maxTokenLength	Integer	No	Not set	Must be positive	Maximum token length
lowerCase	boolean	No	Not set	None	Whether to lower case tokens
stem	boolean	No	Not set	None	Whether to apply stemming
removeStopWords	boolean	No	Not set	None	Whether to remove stop words
customStopWords	List<String>	No	Not set	Non-null	Custom stop word list overriding built-in language defaults
asciiFolding	boolean	No	Not set	None	Whether to apply ASCII folding (e.g., accented characters to ASCII)
minNgramLength	int	No	Not set	Must be positive	Minimum N-gram length (ngram tokenizer only)
maxNgramLength	int	No	Not set	Must be positive, >= minNgramLength	Maximum N-gram length (ngram tokenizer only)
prefixOnly	boolean	No	Not set	None	Generate only prefix N-grams (ngram tokenizer only)
skipMerge	boolean	No	Not set	None	Skip partition merge after indexing (for distributed builds)

Build Output

Return Type	Description
ScalarIndexParams	A scalar index params object with type "inverted" and JSON configuration

Usage Examples

import org.lance.index.scalar.InvertedIndexParams;
import org.lance.index.scalar.ScalarIndexParams;
import org.lance.index.IndexParams;
import java.util.Arrays;

// Create a basic inverted index with defaults
ScalarIndexParams basic = InvertedIndexParams.builder().build();

// Create an English full-text search index with stemming and stop word removal
ScalarIndexParams english = InvertedIndexParams.builder()
    .language("English")
    .stem(true)
    .removeStopWords(true)
    .lowerCase(true)
    .withPosition(true)
    .build();

// Create an N-gram index for substring matching
ScalarIndexParams ngram = InvertedIndexParams.builder()
    .baseTokenizer("ngram")
    .minNgramLength(2)
    .maxNgramLength(4)
    .prefixOnly(false)
    .build();

// Create an index with custom stop words
ScalarIndexParams custom = InvertedIndexParams.builder()
    .language("English")
    .removeStopWords(true)
    .customStopWords(Arrays.asList("the", "a", "an", "custom"))
    .asciiFolding(true)
    .build();

// Wrap in IndexParams
IndexParams params = IndexParams.builder()
    .setScalarIndexParams(english)
    .build();

Related Pages

• ScalarIndexParams - The output type produced by InvertedIndexParams.Builder.build()
• IndexParams - Top-level container that holds scalar index params
• BTreeIndexParams - Another scalar index builder for B-Tree indices
• IndexType - Contains the INVERTED and NGRAM enum constants