多国语言处理

开发平台：
C#

KeywordsSort.cs：源码内容
							﻿using System;
using System.Collections.Generic;
using System.Text;
using System.Text.RegularExpressions;
using CommonLibrary;
namespace ChineseSplitter
{
    /// <summary>
    /// 关键字排序
    /// </summary>
    public class KeywordsSort
    {
        static WordsType myWordsType;
        const string SPLIT_STRING = "  ";
        const char SPLIT_CHAR = '/';
        public List<KeywordsInfo> myKeyList;
        /// <summary>
        /// 构造函数
        /// </summary>
        public KeywordsSort()
        {
            if (myWordsType == null)
            {
                myWordsType = new WordsType();
            }
            this.myKeyList = new List<KeywordsInfo>();
        }
        /// <summary>
        /// 添加关键字
        /// </summary>
        /// <param name="inputStr">字符串</param>
        public void Append(string inputStr)
        {
            string[] keywords = Regex.Split(inputStr, SPLIT_STRING);
            KeywordsInfo myTempInfo = new KeywordsInfo("");
            int indexCount = 0;
            for (int i = 0; i < keywords.Length; i++)
            {
                string[] keys = keywords[i].Split(SPLIT_CHAR);
                string keyword = keys[0].Replace("rn","");
                string keywordType = "";
                if (keys.Length >= 2) keywordType = keys[1];
                if (keywords[i] != String.Empty && myWordsType.GetMark(keywordType) > 0)
                {
                    myTempInfo.Keywords = keyword;
                    int place = Finder.FindPlace<KeywordsInfo>(this.myKeyList, myTempInfo, KeywordsInfoHashCompare);
                    if (place == -1)
                    {
                        place = Finder.FindInsertPlace<KeywordsInfo>(this.myKeyList, myTempInfo, KeywordsInfoHashCompare);
                        KeywordsInfo myNewInfo = new KeywordsInfo(keyword);
                        this.myKeyList.Insert(place, myNewInfo);
                      
                    }
                    KeywordsInfo myKeyInfo = this.myKeyList[place];
                    myKeyInfo.Count += 1;
                    myKeyInfo.WordsMark += myWordsType.GetMark(keywordType);
                    myKeyInfo.PositionList.Add(indexCount);
                    this.myKeyList[place] = myKeyInfo;
                }
                indexCount += keyword.Length;
            }
        }
        /// <summary>
        /// 获得积分最前的关键字
        /// </summary>
        /// <returns>返回值</returns>
        public string GetTopKeywords()
        {
            return GetTopKeywords(10);
        }
        /// <summary>
        /// 获得积分最前的关键字
        /// </summary>
        /// <param name="topCount">关键字数</param>
        /// <returns>返回值</returns>
        public string GetTopKeywords(int topCount)
        {
            
            Sorter.DimidiateSort<KeywordsInfo>(this.myKeyList, KeywordsInfoAllMarkCompare);
            if (topCount > this.myKeyList.Count) topCount = this.myKeyList.Count;
            StringBuilder myStr = new StringBuilder();
            for (int i = 0; i < topCount; i++)
            {
                myStr.Append(this.myKeyList[i].Keywords + "  ");
            }
            Sorter.DimidiateSort<KeywordsInfo>(this.myKeyList, KeywordsInfoHashCompare);
            return myStr.ToString();
        }
        /// <summary>
        /// KeywordsInfo比较方法,根据Count
        /// </summary>
        /// <param name="value1">值1</param>
        /// <param name="value2">值2</param>
        /// <returns>返回值</returns>
        private int KeywordsInfoCountCompare(KeywordsInfo value1,KeywordsInfo value2)
        {
            if (value1.Count > value2.Count)
            {
                return 1;
            }
            else if (value1.Count < value2.Count)
            {
                return -1;
            }
            else 
            {
                return 0;
            }
        }
        /// <summary>
        /// KeywordsInfo比较方法,根据Hash
        /// </summary>
        /// <param name="value1">值1</param>
        /// <param name="value2">值2</param>
        /// <returns>返回值</returns>
        private int KeywordsInfoHashCompare(KeywordsInfo value1, KeywordsInfo value2)
        {
            if (value1.Keywords.GetHashCode() > value2.Keywords.GetHashCode())
            {
                return 1;
            }
            else if (value1.Keywords.GetHashCode() < value2.Keywords.GetHashCode())
            {
                return -1;
            }
            else
            {
                return 0;
            }
        }
        /// <summary>
        /// KeywordsInfo比较方法,根据AllMark
        /// </summary>
        /// <param name="value1">值1</param>
        /// <param name="value2">值2</param>
        /// <returns>返回值</returns>
        private int KeywordsInfoAllMarkCompare(KeywordsInfo value1, KeywordsInfo value2)
        {
            if (value1.AllMark > value2.AllMark)
            {
                return 1;
            }
            else if (value1.AllMark < value2.AllMark)
            {
                return -1;
            }
            else
            {
                return 0;
            }
        }
    }
    /// <summary>
    /// 关键字信息
    /// </summary>
    public struct KeywordsInfo
    {
        /// <summary>
        /// 关键字
        /// </summary>
        public string Keywords;
        /// <summary>
        /// 出现次数
        /// </summary>
        public short Count;
        /// <summary>
        /// 关键字积分
        /// </summary>
        public float WordsMark;
        /// <summary>
        /// 关键字出现的位置类列
        /// </summary>
        public List<int> PositionList;
        public KeywordsInfo(string _Keywords)
        {
            this.Keywords = _Keywords;
            this.Count = 0;
            this.WordsMark = 0;
            this.PositionList = new List<int>();
        }
        /// <summary>
        /// 总积分
        /// </summary>
        public float AllMark
        {
            get
            {
                return this.Count * this.WordsMark;
            }
        }
        public byte[] SaveBytes
        {
            get
            {
                return null;
            }
        }
    }
    /// <summary>
    /// 关键字类别积分
    /// </summary>
    public class WordsType
    {
        Dictionary<string, float> typeList;
        const float DEFAULT_MARK = 0.8f;
        /// <summary>
        /// 构造函数
        /// </summary>
        public WordsType()
        {
            this.typeList = new Dictionary<string, float>();
            this.typeList.Add("", (float)(DEFAULT_MARK));
            this.typeList.Add("Ag", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("a", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("ad", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("an", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("b", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("c", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("Dg", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("d", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("e", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("f", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("g", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("h", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("i", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("j", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("k", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("l", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("m", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("Ng", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("n", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("nr", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("ns", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("nt", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("nz", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("o", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("p", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("q", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("r", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("s", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("Tg", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("t", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("u", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("Vg", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("v", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("vd", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("vn", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("w", (float)(DEFAULT_MARK * 0));
            this.typeList.Add("x", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("y", (float)(DEFAULT_MARK * 0.50));
            this.typeList.Add("z", (float)(DEFAULT_MARK * 0.50));
        }
        /// <summary>
        /// 获取关键字的积分
        /// </summary>
        /// <param name="wordsType">词性类别</param>
        /// <returns>返回积分</returns>
        public float GetMark(string wordsType)
        {
            if (wordsType != null && this.typeList.ContainsKey(wordsType))
            {
                return this.typeList[wordsType];
            }
            else
            {
                return DEFAULT_MARK;
            }
        }
    }
    /*
    词性表
     
    Ag 	形语素 
    a 	形容词 
    ad 	副形词 
    an 	名形词 
    b 	区别词 
    c 	连词 
    Dg 	副语素 
    d 	副词 
    e 	叹词 
    f 	方位词 
    g 	语素 
    h 	前接成分 
    i 	成语 
    j 	简称略语 
    k 	后接成分 
    l 	习用语 
    m 	数词 
    Ng 	名语素 
    n 	名词 
    nr 	人名 
    ns 	地名 
    nt 	机构团体 
    nz 	其他专名 
    o 	拟声词 
    p 	介词 
    q 	量词 
    r 	代词 
    s 	处所词 
    Tg 	时语素 
    t 	时间词 
    u 	助词 
    Vg 	动语素 
    v 	动词 
    vd 	副动词 
    vn 	名动词 
    w 	标点符号 
    x 	非语素字 
    y 	语气词 
    z 	状态词
     */
    /*
   public struct WordsTypeMark
   {
       public string WordsType;
       public byte Mark;
       public WordsTypeMark(string _WordsType, byte _Mark)
       {
           this.WordsType = _WordsType;
           this.Mark = _Mark;
       }
   }*/
}

						