Glossary term
Glossary term
Memory and Retrieval
Sparse retrieval relies on traditional keyword matching methods to retrieve content. It is fast and effective for exact matches, but often less flexible than semantic search when queries are vague or varied.
Apache Lucene, the engine behind Elasticsearch and Solr, is the most widely deployed sparse retrieval system.
SPLADE and uniCOIL are learned sparse retrievers that improve on BM25.
Pinecone hybrid search and Weaviate's BM25 combine sparse retrieval with dense vector search.