public void AddNodesTest()
{
IBaseTree tree = new BaseTree();
Assert.IsNull(tree.Root);
Assert.IsFalse(tree.Contains("all"));
tree.AddNode(null, new Node("All"));
Assert.IsNotNull(tree.Root);
Assert.AreEqual("all", tree.Root.KeyWord);
Assert.IsTrue(tree.Contains("all"));
Assert.AreEqual("all", tree.GetNode("all").KeyWord);
}
public IBaseTree BuildTreeGoodMatches()
{
//change the .1?
var orderedByIDF = WordDocumentAppearances.Where(x => x.Value.IDF > .2).OrderBy(x => x.Value.IDF).ToList();
//int b = 0;
//var orderedByIDF = WordDocumentAppearances.Where(x => !Int32.TryParse(x.Key, out b)).OrderByDescending(x => x.Value.IDF).Take((int)(WordDocumentAppearances.Count * .7)).ToList();
IBaseTree tree = new BaseTree();
tree.AddWord((string)null, "the");
orderedByIDF = orderedByIDF.OrderBy(x => x.Value.IDF).ToList();
//Make the cutoff greater than 0, so not quite all words are added? - taken care of above?
while (orderedByIDF.Count > 0)
{
//decide next word to add
string word = orderedByIDF.First().Key;
Console.WriteLine("Adding word '" + word + "' to tree");
//find lowest branch in which all significant occurences of the word/document happen
int numDif = Int32.MaxValue;
Node parent = null;
foreach (Node n in tree) //for every branch in the tree
//Console.WriteLine("Checking branch");
//first check to make sure it is completely contained
{
bool contained = true;
foreach (var element in WordDocumentAppearances[word].Docs)
{
//TODO: Make this check to make sure appearances are significant
if (!WordDocumentAppearances[n.KeyWord].Docs.Contains(element) && PossibleContent[new KeyValuePair <string, Document>(word, new Document()
{
Name = element
})].Frequency > 1)
{
contained = false;
break;
}
}
//if it is,
if (contained)
{
//Console.WriteLine("Determining fit");
int numDifTemp = 0;
foreach (string s2 in WordDocumentAppearances[n.KeyWord].Docs) //for each document the branch's word appears in
{
if ((!WordDocumentAppearances[word].Docs.Contains(s2) && PossibleContent[new KeyValuePair <string, Document>(n.KeyWord, new Document()
{
Name = s2
})].Frequency > 1)) // if this word doesn't appear in a document which the other does
{
numDifTemp++;
}
}
//if the number of different branches is less than the last node's,
if (numDifTemp < numDif)
{
Console.WriteLine("New parent: " + n.KeyWord);
//make this the new parent of the new word
numDif = numDifTemp;
parent = n;
}
}
}
if (parent != null)
{
Console.WriteLine("Adding " + word + " to tree");
//add to that branch
tree.AddNode(parent, new Node(word));
}
orderedByIDF.Remove(orderedByIDF.First());
Console.WriteLine("Remaining words: " + orderedByIDF.Count);
}
return(tree);
}
