The search engine I am working on is to index lists of sale item listings that include fields like title, description, item ID and seller user ID. The relevancy and word matching using case-insensitivity and stemming in the following Analyzer works well. But I just do not want to tokenize and stem the seller user ID field, which is an alphanumeric, no space string that would need exact phrase matching.

I notice that Lucene's IndexWriter only take one analyzer per index. So I cannot index the seller user ID with exact phrase match. I have heard a suggestion that I may create another index with no-stemming and do multi-index search when I need to do a query with the seller ID in the search criteria. But I'm not sure of the performance this multi-index search. What are some good solutions to this problem?

			

			Analyzer analyzer = new TextSearchAnalyzer();
			

			IndexWriter writer   = new IndexWriter(indexDir, analyzer, newIndex);
			

		

			

			public class TextSearchAnalyzer extends Analyzer
			

			{
			

			public final TokenStream tokenStream(final Reader reader)
			

			{
			

			TokenStream result = new StandardTokenizer(reader);
			

			

			result = new StandardFilter(result);
			

			result = new LowerCaseFilter(result);
			

			result = new StopFilter(result, stopTable);
			

			result = new PorterStemFilter(result);
			

			

			return result;
			

			}
			

			}