Lucene.Net使用WhitespaceAnalyzer搜索名称无结果，求排查问题

Let's break down what's going wrong here and get your searches working as expected.

The Core Issues #

Your setup has two key problems that are blocking your search results:

1. WhitespaceAnalyzer's Limited Behavior
   WhitespaceAnalyzer only splits text into tokens based on spaces—it doesn't handle case conversion (e.g., "Baileys" vs "baileys") or stemming (reducing words to their root form, like turning "baileys" into "bailey"). So if your indexed transcript has "Baileys" (capitalized) or "baileys" (plural), those are stored as distinct, unmodified tokens.

2. TermQuery Requires Exact Matches
   TermQuery doesn't pass your search term through an analyzer—it looks for a character-for-character exact match with tokens in the index. That means:

   • If you search for "bailey" but the indexed token is "baileys", no match.
   • If you search for "baileys" but the indexed token is "Baileys", no match.
   • Even tiny differences (like trailing spaces) will break the match.

Fixes to Try #

1. Use QueryParser Instead of TermQuery

QueryParser automatically passes your search term through the same analyzer you used for indexing, ensuring consistency between indexed tokens and search terms. Here's how to adjust your code:

// Reuse the same analyzer from indexing (WhitespaceAnalyzer here)
var analyzer = new WhitespaceAnalyzer(LuceneVersion.LUCENE_48); // Match your Lucene.Net version
var queryParser = new QueryParser(LuceneVersion.LUCENE_48, "transcript", analyzer);

// Parse your search value into a properly analyzed query
Query searchQuery = queryParser.Parse("<search value>");
booleanMiniQuery.Add(searchQuery, rule);

This way, if your indexed text has "baileys" and you search for "baileys", the analyzer will process both the same way, and you'll get matching results.

2. Switch to a Stemming Analyzer (For Singular/Plural Matches)

If you want "bailey" to match "baileys" (and vice versa), you need an analyzer that handles stemming. Try using SnowballAnalyzer (for English) or StandardAnalyzer with a stem filter:

Indexing with SnowballAnalyzer:

var analyzer = new SnowballAnalyzer(LuceneVersion.LUCENE_48, "English");
document.AddField("transcript", <transcript value>, Lucene.Net.Documents.Field.Store.YES, Lucene.Net.Documents.Field.Index.ANALYZED);

Searching with the same analyzer via QueryParser:

var analyzer = new SnowballAnalyzer(LuceneVersion.LUCENE_48, "English");
var queryParser = new QueryParser(LuceneVersion.LUCENE_48, "transcript", analyzer);
Query searchQuery = queryParser.Parse("<search value>");
booleanMiniQuery.Add(searchQuery, rule);

SnowballAnalyzer will reduce both "bailey" and "baileys" to the same root token, so searches for either term will find matching documents.

3. (Not Recommended) Force Exact Matches

If you absolutely must use TermQuery, you need to ensure your search term exactly matches the indexed token:

• Match the exact case (e.g., search "Baileys" if the indexed term is capitalized)
• Match the exact pluralization (e.g., search "baileys" if that's what's in the transcript)
  This is rigid and not user-friendly, so it's not a good long-term solution.

Quick Recap #

The main issue is that TermQuery skips analyzer processing, while WhitespaceAnalyzer doesn't normalize text or handle word variations. Using QueryParser with a matching analyzer fixes consistency issues, and switching to a stemming analyzer lets you match related terms like singular/plural forms.

内容的提问来源于stack exchange，提问作者e03050