中文地址分词及匹配项目

处理中文地址的分词和匹配 采用混合分词算法进行中文地址分词 在中文地址分词基础上采用Double Levenshetin算法进行中文地址相似度进行地址匹配

package experiment;

import java.io.BufferedReader;
import java.io.FileInputStream;
import java.io.FileOutputStream;
import java.io.IOException;
import java.io.InputStreamReader;
import java.io.PrintStream;
import java.net.URI;
import java.net.URISyntaxException;
import java.util.ArrayList;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.hbase.HbaseConfiguration;

import com.AddressSegment.data.dao.impl.AddressQueryImpl;
import com.AddressSegment.logic.AddressSplitImpl;
import com.AddressSegment.logic.UndefinedWordRecognize;
import com.AddressSegment.metadata.model.CharDictionary;
import com.AddressSegment.metadata.model.WordDictionary;
import com.AddressSegment.tool.dao.impl.DictionaryFileOperationDAOImpl;
import com.AddressSegment.util.Config;

public class CountAddress {
	public static Configuration config = null;
	public static FileSystem fs = null;
	public static DictionaryFileOperationDAOImpl DF = null;
	public static WordDictionary wordDict = null;
	public static CharDictionary charDict = null;
	public static int rowkey = 0;

	// public static HTablePool pool = new HTablePool（config 1000）;

	static {
		config = HbaseConfiguration.create（）;
		wordDict = new WordDictionary（）;
		charDict = new CharDictionary（）;
		Configuration conf = new Configuration（）;
		try {
			fs = FileSystem.get（URI.create（“hdfs://192.168.31.172:9000“） conf）;
		} catch （IOException e1） {
			e1.printStackTrace（）;
		}
		try {
			DF = new DictionaryFileOperationDAOImpl（Config.getDefaultDictionaryHDFSURL（）
					Config.getCharDictionaryHDFSURL（） fs）;
		} catch （URISyntaxException e） {
			e.printStackTrace（）;
		} catch （IOException e） {
			e.printStackTrace（）;
		}
		DF.putFileToDict（wordDict charDict）;
	}
	
	public static void ComputeAddressCount（String fileInputPath String fileOutputPath） throws IOException URISyntaxException{
		BufferedReader br = new BufferedReader（new InputStreamReader（new FileInputStream（fileInputPath）“GBK“））;
		String line = ““;
		FileOutputStream out=new FileOutputStream（fileOutputPath）;
		PrintStream p=new PrintStream（out）;
		while （（line = br.readLine（）） != null） {
			Integer count = 0;
			
			AddressSplitImpl asi = new AddressSplitImpl（）;
			ArrayList strArray = asi.Split（line fs）;
			UndefinedWordRecognize uwr = new UndefinedWordRecognize（）;
			ArrayList wordArray1 = uwr.getUndefinedWord（strArray）;
			AddressQueryImpl aqi = new AddressQueryImpl（）;
			count = aqi.queryAddressCount（wordArray1）;
			
			System.out.println（line）;
			System.out.println（count）;
			p.println（line+“\t“+count）;
		}
		p.close（）;
		br.close（）;
	}

	public static void main（String[] args） throws IOException URISyntaxException {
		ComputeAddressCount（“C:/Users/HYFrank/Desktop/Noname1.txt“ “C:/Users/HYFrank/Desktop/countAddress.txt“）;
	}

}

 属性            大小     日期    时间   名称
----------- ---------  ---------- -----  ----

     文件       5930  2016-08-09 22:51  src\com\AddressSegment\data\dao\baseJdbcTemplate.java

     文件        411  2016-08-01 11:36  src\com\AddressSegment\data\dao\declare\AddressQuery.java

     文件        238  2016-01-31 15:22  src\com\AddressSegment\data\dao\declare\SegmentInsert.java

     文件       5207  2016-08-09 22:24  src\com\AddressSegment\data\dao\impl\AddressQueryImpl.java

     文件       1480  2016-02-25 14:56  src\com\AddressSegment\data\dao\impl\SegmentInsertImpl.java

     文件        212  2016-01-29 23:28  src\com\AddressSegment\data\dao\ModelRowMapper.java

     文件        814  2016-02-25 23:13  src\com\AddressSegment\logic\AddressEncodingService.java

     文件       1020  2016-04-05 00:21  src\com\AddressSegment\logic\AddressSplitImpl.java

     文件       3924  2016-08-30 00:37  src\com\AddressSegment\logic\AlgorithmDaoImpl.java

     文件       3620  2016-02-26 00:36  src\com\AddressSegment\logic\GaodeEncodingServiceInvoker.java

     文件        283  2016-02-25 17:00  src\com\AddressSegment\logic\service\AddressEncoding.java

     文件        621  2016-04-05 00:20  src\com\AddressSegment\logic\service\AddressSplit.java

     文件       1201  2016-04-12 21:53  src\com\AddressSegment\logic\service\AddressTageMaking.java

     文件        219  2016-01-25 21:49  src\com\AddressSegment\logic\service\AlgorithmInterface.java

     文件       3418  2016-03-05 14:01  src\com\AddressSegment\logic\service\HttpRequestTemplate.java

     文件        258  2016-07-24 23:35  src\com\AddressSegment\logic\service\IHttpResponseHandler.java

     文件        271  2016-07-24 23:35  src\com\AddressSegment\logic\service\UndefinedWordRecognizeInterface.java

     文件       4771  2016-07-22 09:49  src\com\AddressSegment\logic\UndefinedWordRecognize.java

     文件       5195  2016-04-12 23:04  src\com\AddressSegment\main\AddressRegexTage.java

     文件       4395  2016-04-09 01:07  src\com\AddressSegment\main\AddressSegment.java

     文件       5349  2016-07-07 21:17  src\com\AddressSegment\main\AddressSegmentTage.java

     文件       6496  2016-07-20 14:06  src\com\AddressSegment\main\AddressSegmentToHbase.java

     文件       5976  2016-07-24 23:26  src\com\AddressSegment\main\AddressSegmentToHDFS.java

     文件       6047  2016-07-27 00:45  src\com\AddressSegment\main\AddressSegmentToHDFSToHbase.java

     文件       2661  2016-04-07 23:20  src\com\AddressSegment\main\WordCount.java

     文件        719  2016-01-24 00:56  src\com\AddressSegment\metadata\model\Algorithm.java

     文件       1985  2016-01-23 16:59  src\com\AddressSegment\metadata\model\baseModel.java

     文件        836  2016-01-23 16:59  src\com\AddressSegment\metadata\model\CharDictionary.java

     文件        407  2016-02-25 16:59  src\com\AddressSegment\metadata\model\CoordinateCode.java

     文件       1179  2016-01-30 22:38  src\com\AddressSegment\metadata\model\Segment.java

............此处省略226个文件信息