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 static void Main(System.String[] a)
{
System.String indexName = "localhost_index";
System.String fn = "c:/Program Files/Apache Group/Apache/htdocs/manual/vhosts/index.html.en";
System.Uri url = null;
for (int i = 0; i < a.Length; i++)
{
if (a[i].Equals("-i"))
{
indexName = a[++i];
}
else if (a[i].Equals("-f"))
{
fn = a[++i];
}
else if (a[i].Equals("-url"))
{
url = new System.Uri(a[++i]);
}
}
System.IO.StreamWriter temp_writer;
temp_writer = new System.IO.StreamWriter(System.Console.OpenStandardOutput(), System.Console.Out.Encoding);
temp_writer.AutoFlush = true;
System.IO.StreamWriter o = temp_writer;
IndexReader r = IndexReader.Open(indexName);
o.WriteLine("Open index " + indexName + " which has " + r.NumDocs() + " docs");
MoreLikeThis mlt = new MoreLikeThis(r);
o.WriteLine("Query generation parameters:");
o.WriteLine(mlt.DescribeParams());
o.WriteLine();
Query query = null;
if (url != null)
{
o.WriteLine("Parsing URL: " + url);
query = mlt.Like(url);
}
else if (fn != null)
{
o.WriteLine("Parsing file: " + fn);
query = mlt.Like(new System.IO.FileInfo(fn));
}
o.WriteLine("q: " + query);
o.WriteLine();
IndexSearcher searcher = new IndexSearcher(indexName);
Hits hits = searcher.Search(query);
int len = hits.Length();
o.WriteLine("found: " + len + " documents matching");
o.WriteLine();
for (int i = 0; i < System.Math.Min(25, len); i++)
{
Document d = hits.Doc(i);
System.String summary = d.Get("summary");
o.WriteLine("score : " + hits.Score(i));
o.WriteLine("url : " + d.Get("url"));
o.WriteLine("\ttitle : " + d.Get("title"));
if (summary != null)
o.WriteLine("\tsummary: " + d.Get("summary"));
o.WriteLine();
}
}
public static void Main(System.String[] a)
{
System.String indexName = "localhost_index";
System.String fn = "c:/Program Files/Apache Group/Apache/htdocs/manual/vhosts/index.html.en";
System.Uri url = null;
for (int i = 0; i < a.Length; i++)
{
if (a[i].Equals("-i"))
{
indexName = a[++i];
}
else if (a[i].Equals("-f"))
{
fn = a[++i];
}
else if (a[i].Equals("-url"))
{
url = new System.Uri(a[++i]);
}
}
System.IO.StreamWriter temp_writer;
temp_writer = new System.IO.StreamWriter(System.Console.OpenStandardOutput(), System.Console.Out.Encoding);
temp_writer.AutoFlush = true;
System.IO.StreamWriter o = temp_writer;
IndexReader r = IndexReader.Open(indexName);
o.WriteLine("Open index " + indexName + " which has " + r.NumDocs() + " docs");
MoreLikeThis mlt = new MoreLikeThis(r);
o.WriteLine("Query generation parameters:");
o.WriteLine(mlt.DescribeParams());
o.WriteLine();
Query query = null;
if (url != null)
{
o.WriteLine("Parsing URL: " + url);
query = mlt.Like(url);
}
else if (fn != null)
{
o.WriteLine("Parsing file: " + fn);
query = mlt.Like(new System.IO.FileInfo(fn));
}
o.WriteLine("q: " + query);
o.WriteLine();
IndexSearcher searcher = new IndexSearcher(indexName);
Hits hits = searcher.Search(query);
int len = hits.Length();
o.WriteLine("found: " + len + " documents matching");
o.WriteLine();
for (int i = 0; i < System.Math.Min(25, len); i++)
{
Document d = hits.Doc(i);
System.String summary = d.Get("summary");
o.WriteLine("score : " + hits.Score(i));
o.WriteLine("url : " + d.Get("url"));
o.WriteLine("\ttitle : " + d.Get("title"));
if (summary != null)
{
o.WriteLine("\tsummary: " + d.Get("summary"));
}
o.WriteLine();
}
}
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;
}
static void Main(string[] args)
{
var stopWords = new Hashtable(StringComparer.InvariantCultureIgnoreCase);
foreach (var s in StopWords)
{
stopWords[s] = s;
}
var connectionString = "Data Source=(local);Initial Catalog=AllgressDB;Integrated Security=true";
Repo = new Repository(connectionString);
var directory = PopulateIndex();
//SearchForTerm(directory);
var policy = Repo.GetPolicySection(35);
string text = Convert(policy.Text);
using (var reader = DirectoryReader.Open(directory, true))
using (var indexSearcher = new IndexSearcher(reader))
{
var moreLikeThis = new MoreLikeThis(reader);
moreLikeThis.SetStopWords(stopWords);
moreLikeThis.SetFieldNames(new[] { "description", "procedures", "objectives", "references" });
moreLikeThis.SetBoost(true);
moreLikeThis.SetMinDocFreq(2);
moreLikeThis.SetMinTermFreq(1);
var query = moreLikeThis.Like(text);
(query as BooleanQuery).Add(new TermQuery(new Term("type", "policy")), Occur.MUST_NOT);
moreLikeThis.SetAnalyzer(new StandardAnalyzer(Lucene.Net.Util.Version.LUCENE_29));
var docs = from scoreDoc in indexSearcher.Search(query, null, 1000).ScoreDocs
select new ScoredDocument { Score = scoreDoc.Score, Document = indexSearcher.Doc(scoreDoc.Doc) };
var results = docs.ToArray();
}
Console.ReadKey();
}