“”black lab*“ ”pet shop“”~5 in Lucene (proximity search with multi-word phrases)

问题

How can I do a proximity search for two multi-word phrases in Lucene. For example, I want to find all black lab* (black labrador, black labradoodle, etc) withing 5 words of the phrase "pet shop". Which analyzer should I be using? Which query parser would be recommended? I'm working with Lucene.NET. I've ported the ComplexPhraseQueryParser from Java to C#, but that parser doesn't seem to be doing the trick (or perhaps I'm just using it wrong). I'm just getting started with Lucene, so your help is much appreciated.

回答1:

You can use a SpanQuery for this:

new SpanNearQuery(
    new SpanQuery[] {
        new SpanNearQuery(
            new SpanQuery[] {
                new SpanTermQuery(new Term(FIELD, "black")),
                new SpanMultiTermQueryWrapper<WildcardQuery>(new WildcardQuery(new Term(FIELD, "lab*"))),
            },
            0,
            true),
        new SpanNearQuery(
            new SpanQuery[] {
                new SpanTermQuery(new Term(FIELD, "pet")),
                new SpanTermQuery(new Term(FIELD, "shop")),
            },
            0,
            true),
    },
    5,
    true);

The default Lucene QueryParser doesn't support span queries, but you could try the Surround query parser. I couldn't find much else in the way of documentation.

You may also find this answer and this blog post useful.

回答2:

You just need to set the slop.

来源：https://stackoverflow.com/questions/10800357/black-lab-pet-shop5-in-lucene-proximity-search-with-multi-word-phrases