将排名最高的单词与数据框列 R 中的文本匹配

Question

我有两个数据框， df1:

df1 <- c("A large bunch of purple grapes", "large green potato sack", "small red tomatoes", "yellow and black bananas")
df1 <- data.frame(df1)

df2:

Word <- c("green", "purple", "grapes", "small", "sack", "yellow", "bananas", "large)

Rank <- c(20,18,22,16,15,17,6,12)

df2 <- data.frame(Word,Rank)

df1:

ID      Sentence  
 1      A large bunch of purple grapes  
 2      large green potato sack 
 3      small red tomatoes  
 4      yellow and black bananas

df2:

ID      Word      Rank
 1      green      20
 2      purple     18
 3      grapes     22
 4      small      16
 5      Sack       15
 6      yellow     17
 7      bananas    6
 8      large      12

我想做的是；将 df2 中的词与 "Sentence" 列中包含的词匹配，并在 df1 中插入一个新列，其中包含 df2 中排名最高的匹配词。所以像这样：

df1:

ID     Sentence                         Word
 1     A large bunch of purple grapes   grapes
 2     large green potato sack          green
 3     small red tomatoes               small
 4     yellow and black bananas         yellow

我最初习惯用下面的代码来匹配单词，但这当然会创建一个包含所有匹配单词的列：

x <- sapply(df2$Word, function(x) grepl(tolower(x), tolower(df1$Sentence)))

df1$top_match <- apply(x, 1, function(i) paste0(names(i)[i], collapse = " "))

Answer 1

我写了一个小片段（但变量名不同）

> inp1 
  ID                           Word new_word
1  1        large green potato sack    green
2  2 A large bunch of purple grapes   grapes
3  3       yellow and black bananas   yellow
> 
> inp2
  ID    Word Rank
1  1   green   20
2  2  purple   18
3  3  grapes   22
4  4   small   16
5  5    Sack   15
6  6  yellow   17
7  7 bananas    6
8  8   large   12
> 
> inp1$new_word <- lapply(inp1$Word, function(text){ inp2$Word[inp2$Rank == max(inp2$Rank[inp2$Word %in% unique(as.vector(str_match(text,inp2$Word)))])]})
> 
> inp1
  ID                           Word new_word
1  1        large green potato sack    green
2  2 A large bunch of purple grapes   grapes
3  3       yellow and black bananas   yellow
>

Answer 2

这是一个 tidyverse + stringr 解决方案：

library(tidyverse)
library(stringr)

df1$Sentence %>%
  str_split_fixed(" ", Inf) %>%
  as.data.frame(stringsAsFactors = FALSE) %>%
  cbind(ID = rownames(df1), .) %>%
  gather(word_count, Word, -ID) %>%
  inner_join(df2, by = "Word") %>%
  group_by(ID) %>%
  filter(Rank == max(Rank)) %>%
  select(ID, Word) %>%
  right_join(rownames_to_column(df1, "ID"), by = "ID") %>%
  select(ID, Sentence, Word)

结果：

# A tibble: 4 x 3
# Groups:   ID [4]
     ID                       Sentence   Word
  <chr>                          <chr>  <chr>
1     1 A large bunch of purple grapes grapes
2     2        large green potato sack  green
3     3             small red tomatoes  small
4     4       yellow and black bananas yellow

注：

您可以忽略将 ID 从因子强制转换为字符的警告。我还修改了您的数据集以包含 df1 的正确列名并禁止自动将字符强制转换为因子。

数据：

df1 <- c("A large bunch of purple grapes", "large green potato sack", "small red tomatoes", "yellow and black bananas")
df1 <- data.frame(Sentence = df1, stringsAsFactors = FALSE)

Word <- c("green", "purple", "grapes", "small", "sack", "yellow", "bananas", "large")
Rank <- c(20,18,22,16,15,17,6,12)
df2 <- data.frame(Word,Rank, stringsAsFactors = FALSE)

将排名最高的单词与数据框列 R 中的文本匹配

matching highest ranking word with text in dataframe column R

r

text-mining

dataframe