public void build(string fileName)
{
var text = File.ReadAllText(fileName);
if (String.IsNullOrEmpty(text))
{
return;
}
int i = 0;
foreach (var abz in Regex.Split(text, "\r\n\r\n"))
{
prepareData(abz.Replace("\r\n", " "), i++);
}
freq = clearObject().OrderByDescending(t => t.Value).Take(100).ToDictionary(t => t.Key, t => t.Value);
List <Clust> clust = new List <Clust>();
foreach (var s in freq)
{
Clust c = new Clust();
foreach (var t in source[s.Key])
{
c.sources.Add(t, 1);
}
c.words.Add(s.Key, 1);
clust.Add(c);
}
clust = clustering(clust);
Word root = new Word
{
key = "root",
parent = null,
lev = 0,
name = "root"
};
mapping(root, clust);
string res = "<map version=\"1.0.1\">\r\n" + printMM(root, "") + "\r\n</map>";
File.WriteAllText(Path.ChangeExtension(fileName, "mm"), res);
File.WriteAllText(Path.ChangeExtension(fileName, "xml"), res);
}
private static void mapping(Word parent, List <Clust> clust)
{
if (parent.lev > 3)
{
return;
}
foreach (var c in clust)
{
if (c.words.Count <= 2) //Если осталось меньше трех элементов
{
foreach (var a in c.words)
{
Word root = new Word
{
key = a.Key,
name = dicNames[a.Key].First(),
parent = parent,
lev = parent.lev + 1
};
words.Add(root);
}
}
else
{
var sel = from w in c.words
join f in freq on w.Key equals f.Key
orderby f.Value descending //w.Value * 10000 +
select w.Key;
Word root = new Word
{
key = sel.First(),
name = dicNames[sel.First()].First(),
parent = parent,
lev = parent.lev + 1
};
words.Add(root);
List <Clust> new_clust = new List <Clust>();
foreach (var a in c.words)
{
if (a.Key == root.key)
{
continue;
}
Clust n = new Clust();
foreach (var t in source[a.Key])
{
n.sources.Add(t, 1);
}
n.words.Add(a.Key, 1);
new_clust.Add(n);
}
new_clust = clustering(new_clust);
mapping(root, new_clust);
}
}
}
