向量在數(shù)學(xué)上余弦定義如下：

因此我們可以將句子向量話計算句子的余弦相似度。

Public static void Main()

{

var segmenter = new  JiebaSegmenter();

var douba = ClassSim.MatchKeywordSim("123", "145");

Console.WriteLine("【相似度1】：{0}", douba);

var douba1 = ClassSim.MatchKeywordSim("包皮手術(shù)治療費用怎么治療？", "包皮手術(shù)費用");

Console.WriteLine("【相似度1】：{0}", douba1);

var douba2= ClassSim.MatchKeywordSim("包皮手術(shù)治療費用怎么治療？", "包皮醫(yī)院");

Console.WriteLine("【相似度2】：{0}", douba2);  

Console.ReadKey();

}

然后我們看下使用的匹配方法：

public class ClassSim

{

 public static double MatchKeywordSim(string keyword, string matchkeyword)

        {

            List<char> keywordList = keyword.ToCharArray().ToList();

            List<char> matchkeywordList = matchkeyword.ToCharArray().ToList();

            List<char> unionKeyword = keywordList.union(matchkeywordList).ToList<char>();

            List<int> arrA = new List<int>();

            List<int> arrB = new List<int>();

            foreach (var str in unionKeyword)

            {

                arrA.Add(keywordList.where(x => x == str).Count());

                arrB.Add(matchkeywordList.where(x => x == str).Count());

            }

            double num = 0;

            double numA=0;

            double numB=0;

            for (int i = 0; i < unionKeyword.Count; i++)

            {

                num+=arrA[i]*arrB[i];

                numA+=Math.Pow(arrA[i], 2);

                numB+=Math.Pow(arrB[i], 2);

            }

            double cos = num / (Math.Sqrt(numA) * Math.Sqrt(numB));

            return cos;

        }

}

運行結(jié)果：