`

基于机器学习(machine learning)的SEO实战日记5--分词与词频计算

阅读更多
经过运行抓取程序,抓到的数据网站数量为4305个,接下来,需要进行分词处理,分词后,再统计词出现的次数,词出现的次数一部分提现了本类网站中该次的竞争情况。分词使用的hanlp开源项目,关于该开源项目的引用与使用,此处不详细介绍,读者可以访问 https://github.com/hankcs/HanLP了解详情。本篇博客涵盖的内容包括:分词、统计词频、结果保存数据库。表结构和相关代码如下:
表名:relative_hotwords
表中文名:相关热词信息表
字段名称字段类型字段解释
keywordsvarchar(100)关键词
rh_timesint出现次数
rh_title_timesint在title中出现的次数
rh_keyword_tiemsint在keywords中出现的次数
rh_description_timesint在description中出现的次数
rh_other_timesint在网页其他地方出现的次数
rh_hot_scoreint词热度(百度指数)
rh_pc_scoreint词在PC端的热度(百度指数)
rh_wise_scoreint词在移动端的热度(百度指数)

建表语句:
Create table relative_hotwords(keywords varchar(100),rh_times int,rh_title_times int,
rh_keyword_tiems int,rh_description_times int, rh_other_times   int,rh_hot_score int, rh_pc_score int, rh_wise_score int) character set utf8mb4 collate utf8mb4_bin;


java代码:
/**
 * 分词
 */
public void segWordsAndSave(){
    Sort sort = new Sort();
    Sqlca sqlca = null;
    try {
        sqlca = sort.getSqlca();
        sqlca.execute("select web_html from web_detail limit 1220,5000");

        FileTool ft=new FileTool();
        int k=0;
        while (sqlca.next()){
            ft.saveRowToFile(sqlca.getString("web_html"),"C:\\temp\\hotwords\\t.txt");
            System.out.println(k++);
        }

        ComputeHis ch=new ComputeHis();
        ch.segFileWord("C:\\temp\\hotwords\\t.txt","C:\\temp\\hotwords\\t2.txt",0,0);
    }catch (Exception e){
        e.printStackTrace();
    } finally {
        if (sqlca != null) sqlca.closeAll();
    }

}

/**
     * 统计词频
     */
    public void wordCoutn(){
        List<String> ls=new ArrayList<>();
        NlpTool nt= new NlpTool();
        List<String> res=new ArrayList<>();
        FileTool ft=new FileTool();
        ComputeHis ch=new ComputeHis();
        try {
            BufferedReader reader = new BufferedReader(new InputStreamReader(new FileInputStream("C:\\temp\\hotwords\\t2.txt"),"utf-8"));
            int x=0;
            String line=null;
            String content="";


            Map<String ,Integer> mp=new HashMap <String ,Integer> ();
            while ((line = reader.readLine()) != null) {
                x++;
                if(x%100000==0) {
                    System.out.println(x+"       "+new Date());
                }
                String[] words=line.split(" ");
                for(String word:words){
                    if(mp.containsKey(word)){
                        mp.put(word,mp.get(word) +1);
                    }else{
                        mp.put(word, 1);
                    }
                }
             }
            reader.close();

            Set<String> set=mp.keySet();
            java.util.SortedMap<String, String> topN = new java.util.TreeMap<String, String>();
            for(String key:set){
                topN.put((10000000 - mp.get(key))+"-"+key,"");
            }

            set=topN.keySet();
            System.out.println(set.size());
            int i=0;
            for(String key:set){
                String[] words=key.split("-");
                if(words==null){
                    continue;
                }
                if(words.length<2){
                    continue;
                }
                words[1]=words[1].replace("\r","").replace("\t","").replace("\n","");

                if(words[1]!=null&&!words[1].isEmpty()&&!ch.isStop(words[1])&&words[1].length()>1){
                    res.add(key);
                    i++;
                }
//                if(i>=20) break;
            }
            ft.saveListToFile("C:\\temp\\hotwords\\wdcount.txt",res);
        } catch (Exception e) {
            // TODO Auto-generated catch block
            e.printStackTrace();
        }
    }

/**
     * 多线程解析html,抽取关键词信息
     * @param fileName
     */
    public void getKeyWordsFromHtml(String fileName){
        FileTool ft=new FileTool();
        Sort sort = new Sort();
        NlpTool nlpTool=new NlpTool();
        Sqlca sqlca = null;
        Sqlca sqlcah = null;
        try {
            sqlca = sort.getSqlca();
            sqlcah=sort.getSqlcaH();
            int webNums=0;
            //获得网页数量
            sqlca.execute("select count(*) ct from web_detail");
            if(sqlca.next()){
                webNums=sqlca.getInt("ct");
            }
            if(webNums==0){ //没有网页数据
                sqlca.closeAll();
                return;
            }
            System.out.println(webNums);
            for(int i=0;i<8;i++){
                String parm=fileName;
                int step=webNums/8;
                int begin=i*step;
                int end=(i+1)*step;
                if(end<webNums && end >(webNums - step)) end=webNums;
                parm+=","+begin+","+end;
//                System.out.println(parm);
                final ActorRef ta = system.actorOf(Props.create(AnalyseHtmlActor.class));
                ta.tell(parm, ActorRef.noSender());
                if(end>=webNums) break;
            }
        }catch (Exception e){
            e.printStackTrace();
        }finally {
            if(sqlca!=null) sqlca.closeAll();
            if(sqlcah!=null) sqlcah.closeAll();
        }
    }

0
1
分享到:
评论

相关推荐

Global site tag (gtag.js) - Google Analytics