views:

91

answers:

1

Hi all, I am using Lucene.net V2.9 to create some concept clusters i.e. list of concepts and for each of those concepts associated concepts. But I am getting this exception in the MoreLikeThis.cs class:

AddTermFrequencies(new System.IO.StreamReader(text[j]), termFreqMap, fieldName);

Could not find file 'C:\DEV.....\Debug\signing technology'.

This exception is thrown in: private PriorityQueue RetrieveTerms(int docNum) method

I assume this is because the MLT class is looking for the file to load the text from. But it seems to take an argument of doc_id integer, which I am passing in properly. What am I doing wrong?

Lucene.Net.Store.Directory dir = Lucene.Net.Store.FSDirectory.GetDirectory("ConceptsIndex", false);
            IndexSearcher obj = new IndexSearcher(dir, true); 
            QueryParser qp = new QueryParser(Lucene.Net.Util.Version.LUCENE_29,"Descriptors", new StandardAnalyzer());
            Query q = qp.Parse(solrQuery);
            TopDocs hits = obj.Search(q,null,10);
            List<ConceptCluster> list = new List<ConceptCluster>();
            for (int n = 0; n < hits.scoreDocs.GetLength(0); n++)
            {
                Document doc = obj.Doc(hits.scoreDocs[n].doc);
                string concept = doc.Get("ConceptID");
                MoreLikeThis mlt = new MoreLikeThis(obj.GetIndexReader());
                mlt.SetFieldNames(new string[] { "Descriptors","Concept" });
                mlt.SetMinDocFreq(5);
                mlt.SetMinWordLen(3);

                int like = hits.scoreDocs[n].doc;


                Query mltQuery = mlt.Like(like);
                TopDocs mltHits = obj.Search(mltQuery, null, 10);
                ConceptCluster cc = new ConceptCluster();
                cc.Concept = concept;
                for (int m = 0; m < mltHits.scoreDocs.GetLength(0); m++)
                {
                    cc.RelatedConcepts.Add(obj.Doc(mltHits.scoreDocs[m].doc).Get("ConceptID"));
                }
                list.Add(cc);
            }
A: 

Closing down due to lack of responses.

Mikos