public static IList<CorpusDocument> GetMoreLikeThis(string indexName, int indexDocumentId, int maxDocs)
{
// See: http://lucene.apache.org/java/2_2_0/api/org/apache/lucene/search/similar/MoreLikeThis.html
var mlt = new MoreLikeThis(Searcher.GetIndexReader());
mlt.SetAnalyzer(new StandardAnalyzer(Lucene.Net.Util.Version.LUCENE_29));
mlt.SetFieldNames(new[] { "Title", "Content" });
mlt.SetMinWordLen(4); // improve relevancy
var query = mlt.Like(indexDocumentId);
var tsdc = TopScoreDocCollector.create(maxDocs, true);
Searcher.Search(query, tsdc);
var hits = tsdc.TopDocs().ScoreDocs;
var ret = new List<CorpusDocument>(maxDocs);
foreach (var hit in hits)
{
var d = Searcher.Doc(hit.doc);
ret.Add(new CorpusDocument
{
Id = d.Get("Id"),
Title = d.Get("Title"),
});
}
return ret;
}
public override Query Rewrite(IndexReader reader)
{
MoreLikeThis mlt = new MoreLikeThis(reader);
mlt.SetFieldNames(moreLikeFields);
mlt.SetAnalyzer(analyzer);
mlt.SetMinTermFreq(minTermFrequency);
if (minDocFreq >= 0)
{
mlt.SetMinDocFreq(minDocFreq);
}
mlt.SetMaxQueryTerms(maxQueryTerms);
mlt.SetStopWords(stopWords);
BooleanQuery bq = (BooleanQuery)mlt.Like( new System.IO.StringReader(likeText));
BooleanClause[] clauses = bq.GetClauses();
//make at least half the terms match
bq.SetMinimumNumberShouldMatch((int)(clauses.Length * percentTermsToMatch));
return bq;
}
public override Query Rewrite(IndexReader reader)
{
MoreLikeThis mlt = new MoreLikeThis(reader);
mlt.SetFieldNames(moreLikeFields);
mlt.SetAnalyzer(analyzer);
mlt.SetMinTermFreq(minTermFrequency);
if (minDocFreq >= 0)
{
mlt.SetMinDocFreq(minDocFreq);
}
mlt.SetMaxQueryTerms(maxQueryTerms);
mlt.SetStopWords(stopWords);
BooleanQuery bq = (BooleanQuery)mlt.Like(new System.IO.StringReader(likeText));
BooleanClause[] clauses = bq.GetClauses();
//make at least half the terms match
bq.SetMinimumNumberShouldMatch((int)(clauses.Length * percentTermsToMatch));
return(bq);
}
