如何使用自定义分隔符将语料库分成段落

问题描述

我正在抓取纽约时报的网页以对其进行一些自然语言处理，我想在使用语料库时将网页拆分为多个段落，以便对出现在段落中且包含关键词或短语的单词进行频率计数.

以下适用于句子，但段落是由纽约时报的 • 捐赠的，所以我需要将其替换为语料库阅读段落的方式 - 有人知道吗？我试过 gsub("•","/n",...) 和 gsub("•","/r/n") 但这没有用。

如果有人知道如何在 tm 语料库中完成这一切，而不必在 quanteda 和 TM 之间切换，这样可以节省一些代码。

 website<-read_html("https://www.nytimes.com/2017/01/03/briefing/asia-australia-briefing.html") #Read URL
     


  #Obtain any text with the paragraph Html deliminator 
  text<-website%>%
    html_nodes("p") %>%
    html_text() %>% as.character()
  
  #Collapse the string as it is currently text[1]=para1 and text[2]= para 2
  text<- str_c(text,collapse=" ")


data_corpus_para <- 
  corpus_reshape(corpus((text),to="paragraphs"))


data_corpus_para <-tolower(data_corpus_para )


containstarget <- 
  stringr::str_detect(texts(data_corpus_para ),"pull out of peace talks") #Random string in only one of the paragraphs to proof concept

#Filter for the para's that only contain the sentence above
data_corpus_para <- 
  corpus_subset(data_corpus_para,containstarget)                 

data_corpus_para <-corpus_reshape(data_corpus_para,to = "documents")


#There are quanteda corpus and TM Corpuses. And so I have to convert to a dataframe and then make back into a vcorupus.. this is very messy

data_corpus_para <-quanteda::convert(data_corpus_para )
data_corpus_para_VCorpus<-tm::VCorpus(tm::VectorSource(data_corpus_para$text))

dt.dtm = tm::DocumentTermMatrix(data_corpus_para_VCorpus)
tm::findFreqTerms(dt.dtm,1)

解决方法

如果段落分隔符是“•”，则可以使用corpus_segment()：

library("quanteda")
## Package version: 3.0.0
## Unicode version: 10.0
## ICU version: 61.1
## Parallel computing: 12 of 12 threads used.
## See https://quanteda.io for tutorials and examples.

txt <- "
• This is the first paragraph.
This is still the first paragraph.
• Here is the third paragraph.  Last sentence"

corpus(txt) %>%
  corpus_segment(pattern = "•")
## Corpus consisting of 2 documents and 1 docvar.
## text1.1 :
## "This is the first paragraph. This is still the first paragra..."
## 
## text1.2 :
## "Here is the third paragraph.  Last sentence"

^{由 reprex package (v1.0.0) 于 2021 年 4 月 10 日创建}