多国语言处理

开发平台：

C#

DefaultSplitter.cs：源码内容

							﻿using System;
using System.Collections.Generic;
using System.Text;
using System.IO;
////该源码首发自win.51aspx.com(５１ａsｐｘ．ｃｏｍ)
 
namespace ChineseSplitter
{
    /// <summary>
    /// 词义切分,即中文分词
    /// </summary>
    public class DefaultSplitter : BaseSplitter
    {
        static NICTCLAS nictclas;
        /// <summary>
        /// 构造函数
        /// </summary>
        /// <param name="wordFilePath">中文词库目录路径</param>
        public DefaultSplitter(DirectoryInfo wordFilePath)
        {
            try
            {
                if (nictclas == null)
                {
                    nictclas = new NICTCLAS(wordFilePath.FullName);
                    nictclas.OperateType = (global::eOperateType)eOperateType.FirstTag;
                    nictclas.OutputFormat = (global::eOutputFormat)eOutputFormat.PKU;
                }
            }
            catch(Exception ex)
            {
                throw ex;
            }
        }
        public override string Split(string inputStr)
        {
            string result = "";
            nictclas.ParagraphProcessing(inputStr, ref result);
            return result;
        }
        /// <summary>
        /// 释放内存
        /// </summary>
        public override void Dispose()
        {
            nictclas.Dispose();
            
        }
    }
    /// <summary>
    /// 标注类型
    /// </summary>
    public enum eOperateType
    {
        /// <summary>
        /// 词语切分
        /// </summary>
        OnlySegment = 0,
        /// <summary>
        /// 一级标注
        /// </summary>
        FirstTag = 1,
        /// <summary>
        /// 二级标注
        /// </summary>
        SecondTag = 2,
    }
    /// <summary>
    /// 输出格式
    /// </summary>
    public enum eOutputFormat
    {
        /// <summary>
        /// 北大标注
        /// </summary>
        PKU = 0,
        /// <summary>
        /// 973标注
        /// </summary>
        _973 = 1,
        /// <summary>
        /// XML格式
        /// </summary>
        XML = 2,
    }
}

						