lexir

Lexical IR on top of postings lists.

Status: experimental. This repository is public as a reference implementation; it is not currently packaged for crates.io.

Feature Selection

• default: Includes persistence.
• In-memory only: disable default features.

What it is

lexir is the scoring/ranking layer. Candidate generation and storage live in postings.

Building

lexir is not on crates.io yet; use a git dependency:

[dependencies]
lexir = { git = "https://github.com/arclabs561/lexir" }

Notes:

• postings, rankfns, and durability are pulled as git dependencies.
• gramdex and textprep are pulled from crates.io (and are only used when their features are enabled).

Usage (library)

BM25 (default):

use lexir::bm25::{Bm25Params, InvertedIndex};

let mut idx = InvertedIndex::new();
idx.add_document(1, &["hello".to_string(), "world".to_string()]);
let hits = idx.retrieve(&["hello".to_string()], 10, Default::default()).unwrap();
assert_eq!(hits[0].0, 1);

TF-IDF (requires multiple docs for non-zero IDF):

use lexir::tfidf::{TfIdfParams, retrieve_tfidf};
use lexir::bm25::InvertedIndex;

let mut idx = InvertedIndex::new();
idx.add_document(1, &["hello".to_string(), "world".to_string()]);
idx.add_document(2, &["other".to_string()]);  // IDF(hello) > 0
let hits = retrieve_tfidf(&idx, &["hello".to_string()], 10, TfIdfParams::linear()).unwrap();
assert_eq!(hits[0].0, 1);

Query Likelihood (Dirichlet (\mu=1000) via QueryLikelihoodParams::default()):

use lexir::query_likelihood::{QueryLikelihoodParams, retrieve_query_likelihood};
use lexir::bm25::InvertedIndex;

let mut idx = InvertedIndex::new();
idx.add_document(1, &["hello".to_string(), "world".to_string()]);
let hits = retrieve_query_likelihood(&idx, &["hello".to_string()], 10, QueryLikelihoodParams::default()).unwrap();
assert!(!hits.is_empty());

Features

• persistence (default): save/load via durability + postings/persistence
• recordlog: append-only operation logs for rebuildable indexes (CLI uses this)
• cli: enables the lexir CLI (debugging + end-to-end validation)
• fuzzy: fuzzy query expansion via gramdex — expands only OOV terms (terms not in the index); in-vocabulary terms are used as-is

CLI (with --features cli)

cargo run --features cli -- <subcommand>

Subcommands: index, search-index, search for indexing and search. Record-log operations: log-add, log-delete, log-search, log-checkpoint, log-compact, log-status, log-doctor, log-prune, log-scan, log-validate, log-serve.