public IEnumerable <WordWeight> ExtractTagsWithWeight(ForensicResult resource)
{
init();
var source = new List <object>();
foreach (var it in resource.GetItemWithMark(UserTextMark))
{
source.AddRange(it.GetColumnDataByMark(UserTextMark));
}
var t = string.Join(" ", source);
return(Extractor.ExtractTagsWithWeight(t, KeywordCount).Select(c => new WordWeight()
{
Weight = c.Weight,
Word = c.Word,
}));
}
public void Classify(ForensicResult resource, int Kcount = 3)
{
init();
var source = new List <object>();
Dictionary <string, int> WordIndexMap = new Dictionary <string, int>();
List <string> WordList = new List <string>();
var WordWeight = new Dictionary <ForensicResultItem, IEnumerable <WordWeight> >();
foreach (var it in resource.GetItemWithMark(UserTextMark))
{
var tTags = ExtractTagsWithWeight(it);
if (tTags.Count() > 0)
{
WordWeight.Add(it, tTags);
WordList.AddRange(tTags.Select(c => c.Word));
}
}
WordList = WordList.Distinct().ToList();
for (int i = 0; i < WordList.Count; ++i)
{
WordIndexMap.Add(WordList[i], i);
}
;
var tEngine = new EngineModel()
{
ClassCount = Kcount,
Samples = WordWeight.Values.Select(c =>
{
var tX = new double[WordList.Count];
foreach (var it in c)
{
tX[WordIndexMap[it.Word]] = it.Weight;
}
var t = new PointModel()
{
X = tX,
};
return(t);
}).ToList()
};
tEngine.Run();
var res = tEngine.Classes;
}
static public void SaveForensicResult(ForensicResult res, string path, bool includeAll = false)
{
if (!includeAll)
{
res.Items.ForEach(c =>
{
c.Children.RemoveAll(s => s.Desc == "数据集合");
});
}
JsonSerializerSettings jsonSerializerSettings = new JsonSerializerSettings();
jsonSerializerSettings.TypeNameHandling = TypeNameHandling.Auto;
jsonSerializerSettings.Formatting = Formatting.Indented;
jsonSerializerSettings.NullValueHandling = NullValueHandling.Ignore;
var st = JsonConvert.SerializeObject(res, jsonSerializerSettings);
var by = Encoding.Default.GetBytes(st);
using (var fs = File.Create(path))
{
fs.Write(by, 0, by.Length);
}
}
