What is Spelling Correction?
Spelling correction (a.k.a. Did You Mean) proposes corrected queries using edit distance, language models, and query logs, optionally auto-applying when confidence is high.
How It Works (quick)
- Candidates: Generate via Levenshtein, phonetics, and dictionaries.
- Score: Rank by frequency and context.
- Apply: Suggest or auto-correct; keep original as fallback.
- Measure: Track accept/override rates.
Why It Matters in E-commerce
- Reduces zero-results and frustration.
- Captures revenue otherwise lost to typos.
Best Practices
- Never auto-correct SKU/brand codes.
- Show both corrected and original results link.
- Localize dictionaries; learn from logs.
Summary
A robust “Did You Mean” boosts success with minimal risk—just protect codes and show fallback links.
