static void TryAddRelation(Dictionary <string, WordNode> dict, string a, string b)
{
WordNode x = GetOrAddNode(dict, a), y = GetOrAddNode(dict, b);
x.TryAddNeighbour(b);
y.TryAddNeighbour(a);
}
static WordNode GetOrAddNode(Dictionary <string, WordNode> dict, string word)
{
WordNode node = null;
if (dict.ContainsKey(word))
{
return(dict[word]);
}
else
{
node = new WordNode(word);
dict.Add(word, node);
}
return(node);
}
public static Dictionary <string, int> skuMaxWordCnt; //每个产品keyword中出现次数最多的单词的出现次数
#region Spell check
static void OutputConnections(Dictionary <string, WordNode> dict, string filename, string sku)
{
string[] words = dict.Keys.ToArray();
for (int i = 0; i < words.Length - 1; ++i)
{
if (words[i].Length >= Tools.MIN_WORD_LENGTH)
{
for (int j = i + 1; j < words.Length; ++j)
{
if (words[j].Length < Tools.MIN_WORD_LENGTH)
{
continue;
}
if (Tools.ISEditDistOK(words[i], words[j]))
{
WordNode x = GetOrAddNode(dict, words[i]), y = GetOrAddNode(dict, words[j]);
int t = Tools.GetWordSim(x, y);
File.AppendAllLines(filename, new string[] { sku + " " + words[i] + " " + words[j] + " " + t + " " + x.cnt + " " + y.cnt });
}
}
}
}
}
/// <summary>
/// 返回单词相似度
/// </summary>
/// <param name="x">单词x对应节点</param>
/// <param name="y">单词y对应节点</param>
/// <returns></returns>
public static int GetWordSim(WordNode x, WordNode y)
{
return WordNode.CountOverlap(x.pre, y.pre) + WordNode.CountOverlap(x.next, y.next);
}
static WordNode GetOrAddNode(Dictionary<string, WordNode> dict, string word)
{
WordNode node = null;
if (dict.ContainsKey(word))
return dict[word];
else
{
node = new WordNode(word);
dict.Add(word, node);
}
return node;
}
/// <summary>
/// 返回单词相似度
/// </summary>
/// <param name="x">单词x对应节点</param>
/// <param name="y">单词y对应节点</param>
/// <returns></returns>
public static int GetWordSim(WordNode x, WordNode y)
{
return(WordNode.CountOverlap(x.pre, y.pre) + WordNode.CountOverlap(x.next, y.next));
}