R2-28 第二阶段第3次作业

探花1号 2018-03-30 00:28:55 阅读: 1129

library(RMySQL)

 killDbConnections <- function () {

 all_cons <- dbListConnections(MySQL())

print(all_cons)

for(con in all_cons)

 dbDisconnect(con)

print(paste(length(all_cons), " connections killed."))

 }

 killDbConnections()

con<-dbConnect(MySQL(),host="localhost",dbname="rdb",user="root",password="15521384339")

 dbSendQuery(con,'SET NAMES utf8')

rs <- dbSendQuery(con, "SELECT * FROM article WHERE isdone=1")

words = data.frame(word=c(), freq = c())

while (!dbHasCompleted(rs)) {

 chunk <- dbFetch(rs, 10)

 chunk$abstract<-iconv(chunk$abstract,"WINDOWS-1252","UTF-8")

 count=nrow(chunk)

 cnt=1

  while(cnt<=count){

str = gsub("[[:punct:]]", "", tolower(chunk[cnt,5]))

temp = as.vector(unlist(strsplit(str, split = " ")))

  temp_len = length(temp)

    cnt2 = 1

  while(cnt2 <= temp_len){

  if(temp[cnt2] %in% words$word){

 words[words$word == temp[cnt2],]$freq = words[words$word == temp[cnt2],]$freq+1

  }else{

   words = rbind(words,data.frame(word=c(temp[cnt2]),freq=c(1)))

 }

 cnt2 = cnt2+1

 }

 cnt = cnt +1

}

 }

head(words)


 

install.packages("wordcloud2")

library(wordcloud2)

wordcloud2(words[0:1000,])

new_words = words[order(words$freq,decreasing=T),]

 del_word = c('of','the','and','in','to','a','that','is','for','buy','with','we','are','an','this','these','as','from','which','at','their','have','or','our','its','but','how','be','as','here','on','can','into','data','between','both','also','by')

  words2 =words[which(!(words$word %in% del_word)),]

 wordcloud2(words2)


 
邀请讨论

附件

{{f.title}} 大小 {{f.file_size}} 下载 {{f.count_download}} 金币 {{f.count_gold}}
{{item.nick_name}} 受邀请回答 {{item.create_time}}
{{item.refer_comment.nick_name}} {{item.refer_comment.create_time}}

附件

{{f.title}} 大小 {{f.file_size}} 下载 {{f.count_download}} 金币 {{f.count_gold}}
切换到完整回复 发送回复
赞({{item.count_zan}}) 踩({{item.count_cai}}) 删除 回复 关闭
科研狗©2015-2024 科研好助手,京ICP备20005780号-1 建议意见

服务热线

178 0020 3020

微信服务号