根据两列之间的匹配值（精确）过滤数据框

Question

我有一个包含两列的数据框。一列包含句子列表，另一列包含单词。例如：

words   sentences
loose   Loose connection several times a day on my tablet.  
loud    People don't speak loud or clear enough to hear voicemails
vice    I strongly advice you to fix this issue
advice  I strongly advice you to fix this issue

现在我想过滤这个数据框，只得到那些单词与句子中的单词完全匹配的行：

words   sentences
loose   Loose connection several times a day on my tablet.  
loud    People don't speak loud or clear enough to hear voicemails
advice  I strongly advice you to fix this issue   

单词“vice”不完全匹配，因此必须将其删除。我在数据框中有近 20k 行。有人可以建议我使用哪种方法来完成这项任务，这样我就不会损失太多性能。

Answer 1

使用：

library(stringi)
df[stri_detect_regex(tolower(df$sentences), paste0('\\b',df$words,'\\b')),]

你得到：

   words                                                  sentences
1  loose         Loose connection several times a day on my tablet.
2   loud People don't speak loud or clear enough to hear voicemails
4 advice                    I strongly advice you to fix this issue

解释：

• 用tolower将句子中的大写字母转换为小写字母。
• 通过将words 中的单词包装在wordboundaries (\\b) 中，使用paste0 创建一个正则表达式向量。
• 使用 stringi-package 中的 stri_detect_regex 查看在哪些行中没有匹配项，从而生成具有 TRUE 和 FALSE 值的逻辑向量。
• 带有逻辑向量的子集。

您也可以使用stringr 包中的str_detect（实际上是stringi 包的包装）：

library(stringr)
df[str_detect(tolower(df$sentences), paste0('\\b',df$words,'\\b')),]

---

使用过的数据：

df <- structure(list(words = c("loose", "loud", "vice", "advice"), 
                     sentences = c("Loose connection several times a day on my tablet.", 
                                   "People don't speak loud or clear enough to hear voicemails", 
                                   "I strongly advice you to fix this issue", "I strongly advice you to fix this issue")), 
                .Names = c("words", "sentences"), class = "data.frame", row.names = c(NA, -4L))

Answer 2

您可以尝试以下方法：

df[apply(df, 1, function(x) tolower(x[1]) %in% tolower(unlist(strsplit(x[2], split='\\s+')))),]

df
   words                                                sentences
1  loose       Loose connection several times a day on my tablet.
2   loud People dont speak loud or clear enough to hear voicemail
4 advice          advice  I strongly advice you to fix this issue

Answer 3

最简单的解决方案是使用 stringr 包：

df<- data.frame(words=c("went","zero", "vice"), sent=c("a man went to the park","one minus one is 0","any advice?"))

df$words <- paste0(" ",df$words," ")
df$sent <- paste0(" ",df$sent," ")


df$match <- str_detect(df$sent,df$words)

df.res <- df[df$match > 0,]
df.res$match<-NULL
df.res