private void CreateMLTQuery(MoreLikeThis query)
{
query.Analyzer = new StandardAnalyzer(Lucene.Net.Util.Version.LUCENE_30);
query.MinTermFreq = 1;
query.MinDocFreq = 1;
query.MaxQueryTerms = 15;
query.SetFieldNames(new string[] { "title", "description" });
query.SetStopWords(StopAnalyzer.ENGLISH_STOP_WORDS_SET);
}
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);
}
}
//Realiza la búsqueda de un texto "original" contra los documentos indexados utilizando TFIDF
public static List <Document> moreLikeThisAnalyzer(String original, ISet <string> stopWords, Lucene.Net.Analysis.Analyzer analyzer)
{
Trace.WriteLine("Realizando la Búsqueda");
List <Document> DocumenResult = new List <Document>();
IndexReader indexReader = IndexReader.Open(_directory, true);
IndexSearcher indexSearcher = new IndexSearcher(indexReader);
MoreLikeThis mlt = new MoreLikeThis(indexReader);
mlt.SetFieldNames(DEFAULT_FIELD_NAMES);
mlt.MinDocFreq = DEFALT_MIN_DOC_FREQ;
mlt.MinTermFreq = DEFAULT_MIN_TERM_FREQ;
mlt.MaxQueryTerms = MAX_QUERY_TERMS;
mlt.MinWordLen = DEFAULT_MIN_WORD_LENGTH;
mlt.Analyzer = analyzer;
mlt.SetStopWords(stopWords);
Query query = mlt.Like(new System.IO.StringReader(original));
int topCount = DEFAULT_DOCUMENT_TO_SEARCH;
TopScoreDocCollector collector = TopScoreDocCollector.Create(topCount, true);
indexSearcher.Search(query, collector);
ScoreDoc[] hits = collector.TopDocs().ScoreDocs;
var result = new List <string>();
//Hits hits = indexSearcher.Search(query);
int len = hits.Length;
Trace.WriteLine("Entering");
Trace.WriteLine("-------------------------------------------");
Trace.WriteLine("original :" + original);
Trace.WriteLine("query: " + query);
Trace.WriteLine("found: " + len + " documents");
for (int i = 0; i < Math.Min(25, len); i++)
{
int d = hits[i].Doc;
Trace.WriteLine("score : " + hits[i].Score);
Trace.WriteLine("name : " + d.ToString());
//Colocar los datos en el arreglo de resultados
Document doc = indexSearcher.Doc(hits[i].Doc);
DocumenResult.Add(doc);
}
Trace.WriteLine("-------------------------------------------");
Trace.WriteLine("Exiting");
return(DocumenResult);
}
public SearchResult FindSimular(string key, int resultOffset, int resultLength, bool matchCategory) {
var pageQuery = new TermQuery(new Term("key", key));
var topDocs = _searcher.Search(pageQuery, 1);
if (topDocs.TotalHits == 0) {
return new SearchResult();
}
var doc = topDocs.ScoreDocs[0].Doc;
var moreLikeThis = new MoreLikeThis(_reader) {
Analyzer = _analyzer,
MinWordLen = 3
};
moreLikeThis.SetFieldNames(new[] { "title", "summary", "content", "tags" });
moreLikeThis.SetStopWords(StopWords.DefaultEnglish);
moreLikeThis.MinDocFreq = 2;
var query = moreLikeThis.Like(doc);
var startTime = DateTime.Now;
var ticks = DateTime.Now.ToUniversalTime().Ticks;
Query publishStartQuery = NumericRangeQuery.NewLongRange("publishStart", null, ticks, true, false);
Query publishStopQuery = NumericRangeQuery.NewLongRange("publishStop", ticks, null, false, true);
var booleanQuery = new BooleanQuery {
{query, Occur.MUST},
{pageQuery, Occur.MUST_NOT},
{publishStartQuery, Occur.MUST},
{publishStopQuery, Occur.MUST}
};
if (matchCategory) {
var document = _searcher.Doc(doc);
var field = document.GetField("category");
if (field != null && !string.IsNullOrEmpty(field.StringValue)) {
var categoryQuery = new TermQuery(new Term("category", field.StringValue.ToLowerInvariant()));
booleanQuery.Add(categoryQuery, Occur.MUST);
}
}
var scoreDocs = _searcher.Search(booleanQuery, null, MaxHits, Sort.RELEVANCE).ScoreDocs;
var result = new SearchResult { NumberOfHits = scoreDocs.Length };
if (resultOffset < scoreDocs.Length) {
var resultUpperOffset = resultOffset + resultLength;
if (resultUpperOffset > scoreDocs.Length) {
resultUpperOffset = scoreDocs.Length;
}
for (int i = resultOffset; i < resultUpperOffset; i++) {
Document document = _searcher.Doc(scoreDocs[i].Doc);
Guid pageId;
(document.Get("pageId") ?? string.Empty).TryParseGuid(out pageId);
var hit = new SearchHit {
PageId = pageId,
Path = document.Get("path"),
Title = document.Get("title"),
Excerpt = document.Get("summary")
};
//foreach (string key in metaData) {
// hit.MetaData.Add(key, document.Get(key));
//}
result.Hits.Add(hit);
}
}
var timeTaken = DateTime.Now - startTime;
result.SecondsTaken = timeTaken.TotalSeconds;
return result;
}