public DocumentScore Score(DocumentPosting posting)
{
//var tf = 1 + Math.Log10(Math.Pow(posting.Count, 1 / 2));
var score = Math.Sqrt(posting.Count) * _idf;
return(new DocumentScore(posting.DocumentId, score));
}
public void Add(DocumentPosting data)
{
var node = this;
while (true)
{
if (data.DocumentId < node.Data.DocumentId)
{
if (node.Left == null)
{
node.Left = new DocumentPostingNode(data);
break;
}
else
{
node = node.Left;
}
}
else if (data.DocumentId > node.Data.DocumentId)
{
if (node.Right == null)
{
node.Right = new DocumentPostingNode(data);
break;
}
else
{
node = node.Right;
}
}
else
{
if (data.Data < node.Data.Data)
{
if (node.Left == null)
{
node.Left = new DocumentPostingNode(data);
break;
}
else
{
node = node.Left;
}
}
else
{
if (node.Right == null)
{
node.Right = new DocumentPostingNode(data);
break;
}
else
{
node = node.Right;
}
}
}
}
}
public IList <AnalyzedTerm> AnalyzeDocument(Document document)
{
var analyzedTerms = new List <AnalyzedTerm>();
foreach (var field in document.Fields.Values)
{
if (field.Analyze && field.Index)
{
var tokens = Analyze(field.Value);
var tokenDic = new Dictionary <string, int>();
foreach (var token in tokens)
{
if (tokenDic.ContainsKey(token))
{
tokenDic[token]++;
}
else
{
tokenDic[token] = 1;
}
}
foreach (var tokenGroup in tokenDic)
{
var word = new Word(tokenGroup.Key);
var term = new Term(field.Key, word);
var posting = new DocumentPosting(document.Id, tokenGroup.Value);
analyzedTerms.Add(new AnalyzedTerm(term, posting));
}
}
else if (field.Index)
{
var term = new Term(field.Key, new Word(field.Value));
var posting = new DocumentPosting(document.Id, 1);
analyzedTerms.Add(new AnalyzedTerm(term, posting));
}
}
return(analyzedTerms);
}
public virtual AnalyzedDocument AnalyzeDocument(Document document)
{
var words = new List <AnalyzedTerm>();
foreach (var field in document.Fields)
{
if (field.Key[0] == '_')
{
var term = new Term(field.Key, new Word(field.Value));
var posting = new DocumentPosting(document.Id, 1);
words.Add(new AnalyzedTerm(term, posting));
}
else
{
var tokenDic = new Dictionary <string, int>();
foreach (var token in Analyze(field.Value))
{
if (tokenDic.ContainsKey(token))
{
tokenDic[token]++;
}
else
{
tokenDic[token] = 1;
}
}
foreach (var tokenGroup in tokenDic)
{
var word = new Word(tokenGroup.Key);
var term = new Term(field.Key, word);
var posting = new DocumentPosting(document.Id, tokenGroup.Value);
words.Add(new AnalyzedTerm(term, posting));
}
}
}
return(new AnalyzedDocument(document.Id, words));
}
public IEnumerable <AnalyzedTerm> AnalyzeDocumentInternal(Document document)
{
foreach (var field in document.Fields.Values)
{
if (field.Analyze)
{
var tokenDic = new Dictionary <string, int>();
foreach (var token in Analyze(field.Value))
{
if (tokenDic.ContainsKey(token))
{
tokenDic[token]++;
}
else
{
tokenDic[token] = 1;
}
}
foreach (var tokenGroup in tokenDic)
{
var word = new Word(tokenGroup.Key);
var term = new Term(field.Key, word);
var posting = new DocumentPosting(document.Id, tokenGroup.Value);
yield return(new AnalyzedTerm(term, posting));
}
}
else
{
var term = new Term(field.Key, new Word(field.Value));
var posting = new DocumentPosting(document.Id, 1);
yield return(new AnalyzedTerm(term, posting));
}
}
}
public double Score(DocumentPosting posting)
{
return(Math.Sqrt(posting.Count) * _idf);
}
public double Score(DocumentPosting posting)
{
// log-normalized term frequency
return(1 + Math.Log10(posting.Count) * _idf);
}
public AnalyzedTerm(Term term, DocumentPosting posting)
{
Term = term;
Posting = posting;
}
public DocumentPostingNode(DocumentPosting data)
{
Data = data;
}
