C# 文章相似度算法 Levenshtein 编辑距离算法(转)

编辑距离的算法是首先由俄国科学家Levenshtein提出的，故又叫 Levenshtein Distance。一个字符串可以通过增加一个字符，删除一个字符，替换一个字符得到另外一个字符串，假设，我们把从字符串A转换成字符串B，前面3种操作所执行的最少次数称为AB相似度
如  abc adc  度为 1
      ababababa babababab 度为 2
      abcd acdb 度为2

下面是C#实现

using System;
using System.Text.RegularExpressions;
using System.Threading.Tasks;

namespace Levenshtein
{
    /// <summary>
    /// 分析完成事件委托
    /// </summary>
    /// <param name="sim">相似度</param>
    public delegate void AnalyzerCompletedHander(double sim);

    /// <summary>
    /// 文章相似度工具
    /// </summary>
    public class LevenshteinDistance:IDisposable
    {
        private string str1;
        private string str2;
        private int[,] index;
        int k;
        Task<double> task;

        /// <summary>
        /// 分析完成事件
        /// </summary>
        public event AnalyzerCompletedHander AnalyzerCompleted;

        /// <summary>
        /// 获取或设置文章1
        /// </summary>
        public string Str1
        {
            get { return str1; }
            set
            {
                str1 = Format(value);
                index = new int[str1.Length, str2.Length];
            }
        }

        /// <summary>
        /// 获取或设置文章2
        /// </summary>
        public string Str2
        {
            get { return str2; }
            set
            {
                str2 = Format(value);
                index = new int[str1.Length, str2.Length];
            }
        }

        /// <summary>
        /// 运算总次数
        /// </summary>
        public int TotalTimes
        {
            get { return str1.Length * str2.Length; }
        }

        /// <summary>
        /// 是否完成
        /// </summary>
        public bool IsCompleted
        {
            get { return task.IsCompleted; }
        }

        /// <summary>
        /// 实例化
        /// </summary>
        /// <param name="str1">文章1</param>
        /// <param name="str2">文章2</param>
        public LevenshteinDistance(string str1, string str2)
        {
            this.str1 = Format(str1);
            this.str2 = Format(str2);
            index = new int[str1.Length, str2.Length];
        }

        public LevenshteinDistance()
        {
        }

        /// <summary>
        /// 异步开始任务
        /// </summary>
        public void Start()
        {
            task = new Task<double>(Analyzer);
            task.Start();
            task.ContinueWith(o => Completed(o.Result));
        }

        /// <summary>
        /// 同步开始任务
        /// </summary>
        /// <returns>相似度</returns>
        public double StartAyns()
        {
            task = new Task<double>(Analyzer);
            task.Start();
            task.Wait();
            return task.Result;
        }

        private void Completed(double s)
        {
            if (AnalyzerCompleted != null)
            {
                AnalyzerCompleted(s);
            }
        }

        private double Analyzer()
        {
            if (str1.Length == 0 || str2.Length == 0)
                return 0;
            for (int i = 0; i < str1.Length; i++)
            {
                for (int j = 0; j < str2.Length; j++)
                {
                    k = str1[i] == str2[j] ? 0 : 1;
                    if (i == 0&&j==0)
                    {
                        continue;
                    }
                    else if (i == 0)
                    {
                        index[i, j] = k + index[i, j - 1];
                        continue;
                    }
                    else if (j == 0)
                    {
                        index[i, j] = k + index[i - 1, j];
                        continue;
                    }
                    int temp = Min(index[i, j - 1],
                        index[i - 1, j],
                        index[i - 1, j - 1]);
                    index[i, j] = temp + k;
                }
            }
            float similarty = 1 - (float)index[str1.Length - 1, str2.Length - 1]
                / (str1.Length > str2.Length ? str1.Length : str2.Length);
            return similarty;
        }

        private string Format(string str)
        {
            str = Regex.Replace(str, @"[^a-zA-Z0-9\u4e00-\u9fa5\s]", "");
            return str;
        }

        private int Min(int a, int b, int c)
        {
            int temp = a < b ? a : b;
            temp = temp < c ? temp : c;
            return temp;
        }

        public void Dispose()
        {
            task.Dispose();
        }
    }
}

如非注明 均属原创 如需转载 都请注明： Alien的博客

本文链接地址: C# 文章相似度算法 Levenshtein 编辑距离算法

转载于:https://www.cnblogs.com/haorensw/archive/2012/05/26/2518868.html