Spell Correction & Query Segmentation
summary
A Java port of Symmetric Delete (Symmetric Delete (SymSpell)) adds keyboard-aware scoring (QWERTY distance), tunable operation weights, compound splitting/joining, and a context signal from query counts, and it reads index terms on commit to build its dictionary (Solr integration). On 3,695 indexed items and 8,060 queries Lucene produced 1,550 false negatives, vanilla SymSpell 545, and SymSpell+QWERTY 573; true positives rise versus Lucene, and the project is published to Maven Central.
AI-written summary · about
chapters (6)
related talks
49m‘Vertical Reality Glasses’: Fixing Typos in Search Queries
16mAutomata Invasion
37mSpeech to text with Elasticsearch
21mThe Impact of Word, Multiple Word, and Sentence Input on Virtual Keyboard Decoding Performance
33mFuzzy Search Algorithms How and When to Use Them
15mSpeling werds egspressively with rrkurrent nuril nedwirques!
37mLexical, Vector & Hybrid Search with Elasticsearch
36mUnlocking Smarter Typeahead Search