为数据框的每一行提取情绪计算

Question

我有一个包含文本行的数据框。我想为每一行文本提取一个特定情感的向量，它将是一个二进制 0 不存在这种情感或 1 存在。
它们总共有 5 种情感，但我只想拥有 1对于似乎最多的情感。

我尝试过的示例：

library(tidytext)
text = data.frame(id = c(11,12,13), text=c("bad movie","good movie","I think it would benefit religious people to see things like this, not just to learn about our home, the Universe, in a fun and easy way, but also to understand that non- religious explanations don't leave people hopeless and",))
nrc_lexicon <- get_sentiments("nrc")

预期输出示例：

    id text sadness anger joy love neutral
11 "bad movie" 1 0 0 0 0
12 "good movie" 0 0 1 0 0

任何提示都会对我有所帮助。

为每一行制作下一步是什么的示例？
我如何用 nrc 词典分析调用每一行？

for (i in 1:nrow(text)) {
(text$text[i], nrc_lexicon)
}

Answer 1

这个怎么样：

library(tidytext)   # library for text
library(dplyr)

# your data
text <- data.frame(id = c(11,12,13),
 text=c("bad movie","good movie","I think it would benefit religious
 people to see things like this, not just to learn about our home, 
the Universe, in a fun and easy way, but also to understand that non- religious
 explanations don't leave people hopeless and"), stringsAsFactors = FALSE)  # here put this option, stringAsFactors = FALSE!

# the lexicon
nrc_lexicon <- get_sentiments("nrc")

# now the job
unnested <- text %>%
             unnest_tokens(word, text) %>%  # unnest the words
             left_join(nrc_lexicon) %>%     # join with the lexicon to have sentiments
             left_join(text)                # join with your data to have titles

这里是带id的输出，你也可以带标题，但由于第三个标题太长，我没有放，你可以很容易地把它放成unnested$text共 unnested$id 个：

table_sentiment <- table(unnested$id, unnested$sentiment)
table_sentiment
     anger anticipation disgust fear joy negative positive sadness surprise trust
  11     1            0       1    1   0        1        0       1        0     0
  12     0            1       0    0   1        0        1       0        1     1
  13     0            1       0    1   1        2        3       2        1     0

如果你想要它 data.frame:

 df_sentiment <- as.data.frame.matrix(table_sentiment)

现在你可以做任何你想做的事，例如，如果我没记错的话，你想要一个二进制输出是否存在一个情绪：

df_sentiment[df_sentiment>1]<-1
df_sentiment
   anger anticipation disgust fear joy negative positive sadness surprise trust
11     1            0       1    1   0        1        0       1        0     0
12     0            1       0    0   1        0        1       0        1     1
13     0            1       0    1   1        1        1       1        1     0

为数据框的每一行提取情绪计算

Extract emotions calculation for every row of a dataframe

r

text-mining

sentiment-analysis

tidyr