In Memory Full Text Search with Kotlin
summary
Builds an in-memory inverted index in Kotlin using tokenization, lowercasing and Porter-style stemming, then stores tokens → document lists and ranks with TF‑IDF scoring. Uses Kotlin Coroutines for concurrent document processing and a compressed trie (a Patricia trie) for memory efficiency. Publishes the result as the Lucilla library and reports a stress test of ~1,000,000 documents using about 500 MB for 1M docs, with fuzzy search and boosts planned.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 0:51Why Care
- 5:08Full Text Search
- 8:12Text Extraction
- 12:35Reducing Text
- 14:43Building the Index
- 16:51Searching the Index
- 23:36Questions
