public IndexWord(string docId, string word, IndexAtom indexAtom) : base(word)
{
this.WordIndexAtom = indexAtom;
this.DoctId = docId;
}
public IndexWord(string docId,string word, IndexAtom indexAtom)
: base(word)
{
this.WordIndexAtom = indexAtom;
this.DoctId = docId;
}
/// <summary>
/// 把内容分词结果整理并返回结果
/// </summary>
/// <param name="docWords"></param>
/// <param name="docId"></param>
/// <returns></returns>
public List<IndexWord> Extraction(List<DocWord> docWords, string docId)
{
List<IndexWord> result = new List<IndexWord>();
var groups = docWords.GroupBy(o => o.Word);
foreach (var group in groups)
{
var atom = new IndexAtom();
atom.Position = group.Min(d => d.Position);
atom.Rank = group.Max(d => d.Rank);
atom.Frequency = Convert.ToUInt16((group.Count() * 1.0 / docWords.Count()) * 100);
result.Add(new IndexWord(docId, group.Key, atom));
}
return result;
}
