Lexical Search - Keyword / Full Text Search

Text Preprocessing / Stemming / Normalization

Ranking Functions

TF-IDF

BM25

Indexing

Inverted Index

Production Systems

Lucene

ElasticSearch

Semantic / Neural Search

Embedding Methods

BERT

ModernBERT

Neural Retrieval

Two Tower / Bi-Encoder

Cross Encoders

BiEncoder

Late Interaction Models

ColBERT

Hybrid Methods

SPLADE - Sparse Lexical and Expansion Model for First Stage Ranking

Query - Understanding, Expansion, Parsing


Learning to Rank (LTR)

Models

GBDT - Gradient Boosted Decision Trees

LambdaRank

LambdaMART

Linear Models - Logistic Regression

Neural Methods

Ranking Types

Pointwise

Pairwise

Listwise

Ranking Signals


Data

Synthetic Data

Documents LLM Generated Queries Retrieved Documents LLM Relevance Ranking / Judgment

Query Lists + Results LLM as Judge Scoring

Hard Negative Mining

Document / Index Expansion


Datasets


Evaluation

Links