WebCollector-master
...................\.gitignore
...................\Lazy
...................\....\.idea
...................\....\.....\.name
...................\....\.....\compiler.xml
...................\....\.....\copyright
...................\....\.....\.........\profiles_settings.xml
...................\....\.....\encodings.xml
...................\....\.....\libraries
...................\....\.....\.........\Maven__com_googlecode_juniversalchardet_juniversalchardet_1_0_3.xml
...................\....\.....\.........\Maven__com_sleepycat_je_5_0_73.xml
...................\....\.....\.........\Maven__junit_junit_4_11.xml
...................\....\.....\.........\Maven__log4j_log4j_1_2_17.xml
...................\....\.....\.........\Maven__org_hamcrest_hamcrest_core_1_3.xml
...................\....\.....\.........\Maven__org_json_json_20140107.xml
...................\....\.....\.........\Maven__org_jsoup_jsoup_1_8_3.xml
...................\....\.....\.........\Maven__org_mongodb_mongo_java_driver_3_2_0_rc0.xml
...................\....\.....\.........\Maven__org_slf4j_slf4j_api_1_7_9.xml
...................\....\.....\.........\Maven__org_slf4j_slf4j_log4j12_1_7_9.xml
...................\....\.....\misc.xml
...................\....\.....\modules.xml
...................\....\.....\uiDesigner.xml
...................\....\.....\workspace.xml
...................\....\Lazy.iml
...................\....\README.md
...................\....\demo_task.json
...................\....\demo_task1.json
...................\....\pom.xml
...................\....\src
...................\....\...\main
...................\....\...\....\java
...................\....\...\....\....\cn
...................\....\...\....\....\..\edu
...................\....\...\....\....\..\...\hfut
...................\....\...\....\....\..\...\....\dmic
...................\....\...\....\....\..\...\....\....\dm
...................\....\...\....\....\..\...\....\....\..\KMeans.java
...................\....\...\....\....\..\...\....\....\..\example
...................\....\...\....\....\..\...\....\....\..\.......\StopWords.java
...................\....\...\....\....\..\...\....\....\..\.......\WebpageKmeans.java
...................\....\...\....\....\..\...\....\....\..\.......\WordsBag.java
...................\....\...\....\....\..\...\....\....\webcollector
...................\....\...\....\....\..\...\....\....\............\lazy
...................\....\...\....\....\..\...\....\....\............\....\LazyConfig.java
...................\....\...\....\....\..\...\....\....\............\....\LazyCrawler.java
...................\....\...\....\....\..\...\....\....\............\....\Main.java
...................\....\...\....\....\..\...\....\....\............\....\util
...................\....\...\....\....\..\...\....\....\............\....\....\MongoHelper.java
...................\....\...\....\resources
...................\....\...\....\.........\stopwords.txt
...................\NewsCrawler.java
...................\README.md
...................\README.zh-cn.md
...................\WebCollector-Hadoop
...................\...................\README.md
...................\...................\build.sh
...................\...................\conf
...................\...................\....\crawler-default.xml
...................\...................\....\hadoop
...................\...................\....\......\core-site.xml
...................\...................\....\......\hdfs-site.xml
...................\...................\....\......\mapred-site.xml
...................\...................\....\regex
...................\...................\pom.xml
...................\...................\src
...................\...................\...\main
...................\...................\...\....\java
...................\...................\...\....\....\cn
...................\...................\...\....\....\..\edu
...................\...................\...\....\....\..\...\hfut
...................\...................\...\....\....\..\...\....\dmic
...............