public StoryCollection Find(int hostId, int storyId)
{
int? docId = ConvertStoryIdtoDocId(hostId, storyId);
if (docId.HasValue)
{
IndexSearcher indexSearch = SearchQuery.GetSearcher(hostId);
IndexReader indexReader = indexSearch.GetIndexReader();
MoreLikeThis mlt = new MoreLikeThis(indexReader);
mlt.SetAnalyzer(new DnkAnalyzer());
//mlt.SetFieldNames(new string[] { "title", "description" });
//these values control the query used to find related/similar stories
//
//-we are only using the title and tags fields,
//-the term must appear 1 or more times,
//-the query will only have 3 terms
//-a word less than 3 char in len with be ignored
//-the term must appear at in at least 4 doc
mlt.SetFieldNames(new string[] { "title", "tags" });
mlt.SetMinTermFreq(1);
mlt.SetMaxQueryTerms(5);
mlt.SetMinWordLen(3);
mlt.SetMinDocFreq(4);
mlt.SetStopWords(StopWords());
mlt.SetBoost(true);
Query mltQuery = mlt.Like(docId.Value);
Hits hits = indexSearch.Search(mltQuery);
List<int> results = new List<int>();
for (int i = 0; i < hits.Length(); i++)
{
Document d = hits.Doc(i);
int hitStoryId = int.Parse(d.GetField("id").StringValue());
if (hitStoryId != storyId)
{
results.Add(hitStoryId);
if (results.Count == NUMBER_OF_RELATED_STORIES_TO_RETURN)
break;
}
}
return SearchQuery.LoadStorySearchResults(results);
}
else
return null;
}
public 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 searcher = GetSearcher(indexName);
var mlt = new MoreLikeThis(searcher.GetIndexReader());
mlt.SetAnalyzer(GetAnalyzer(SearchType.Morphologic));
mlt.SetFieldNames(new string[] {"Title", "Content"});
mlt.SetMinWordLen(4); // to avoid most Hebrew ambigous stop-words
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;
}