inverted word indexing

An inverted index is a high-performance data structure that maps unique terms to their specific locations in a document collection to enable near-instant full-text retrieval.

Modern search engines like Elasticsearch and Solr rely on inverted indexes to bypass the inefficiency of linear scans. Instead of searching every document for a keyword (a forward index approach), this system maintains a dictionary of every unique token (word) and its corresponding postings list (the document IDs and positions where that word appears). When a user queries a term like 'latency', the engine performs a single dictionary lookup to find the exact records containing that term. This architecture reduces query complexity from O(n) to O(1) or O(log n) for the initial lookup: a critical optimization for handling millions of records in sub-second timeframes.

https://www.elastic.co/guide/en/elasticsearch/guide/current/inverted-index.html