管道操作符为一条评论返回两行

如何解决管道操作符为一条评论返回两行

我试图在具有两列作者和评论的数据框中获取评论的情绪分数。我使用了命令

data %>%
  get_sentences() %>%
  sentiment() -> data_senti

将结果放入变量中。结果增加了一倍多，因为每个评论被分解为 n 行，用于评论中的 n 个句子。如何保持评论完整无缺，例如，对于每个评论只返回一行，而不是将评论中的 n 个句子分解为 n 行？

=> 输入示例：

=> 首选输出示例

我实际得到的：

到目前为止的完整代码：

# load proper libraries; ensure they are already installed
install.packages(c('syuzhet','sentimentr','tidyverse','magrittr','dplyr'))
library(syuzhet)
library(sentimentr)
library(tidyverse)

# set working dir
setwd("C:/Users/user3/OneDrive/Desktop/Sentiment_Analysis/Data/Output/Clean Sets")

# import file
data <- read.csv("C:/Users/user3/OneDrive/Desktop/Sentiment_Analysis/Data/Output/Clean Sets/all_comments.csv")

data %>%
  get_sentences() %>%
  sentiment() -> data_senti

更新：dput(head(data))

结构（列表（Primary.Key = c（“Google_e1”，“Google_e3”，“Google_e98”） ),Original.Text = c("我觉得奖励是公平的，但奖励的时间框架有时会因管理员而明显延迟。不及时向人员提供奖励会让人感觉该命令不关心成员",“我们全白人、全男性、全 40 或 50 多岁的领导层对女性和少数族裔来说是可怕的。在工作场所遇到困难的每个人都是女性、少数族裔或两者兼而有之。另一方面，白人男性（米奇、唐纳德等）被允许辱骂、不屑一顾和不专业。白人女性（米妮、卡迪 B 和可爱君）被允许将八卦作为一项运动，并且本质上像翻拍一样运行魔多电影中的贱女孩。老实说，我很难相信指挥系统从来没有注意到他们强迫失业的人是拉丁裔女性、黑人男性和其他少数民族，同样也没有注意到这些人他们给白人男性 (Lebron) 和白人女性 (Meghan The Horse) 提供无争议的晋升。老实说，对于一个分析组织来说，很难相信没有人通过少数人的眼光看过它。这就是为什么代表很重要！ !!”,“我没有观察到基于种族、宗教、性别或性偏好的‘更好治疗’的明确迹象。” )),row.names = c(NA,3L),class = "data.frame")

解决方法

欢迎来到 SO，Père Noël。 顾名思义，Pacakge {sentimenter} 的 get_sentences() 默认将文本输入分成句子。要将原始文本输入重构为最终数据框中的定义键，您需要对 sentiment() 生成的基于句子的输出进行分组和汇总。 在这个例子中，我将简单地平均情绪分数，并通过它们的 element_id 附加句子。

library(sentimentr)
library(tidyverse)

df <- tibble(Primary.Key   = c("Google_e1","Google_e3","Google_e98" ),Original.Text = c("I feel as though awards are fairly awarded,but the time frame in which awards are given is sometimes significantly late due at times to admin. Not providing awards to personnel in a timely fashion can give the perception that the command does not care about members","Our all white,all male,all 40 or 50-something leadership is horrible to women and minorities. Every single person that has had difficulties in the workplace has been a woman,a minority,or both. On the other hand,the white men (mickey,donald,and others) are allowed to be abusive,dismissive,and unprofessional. white women (Minnie,and Cardi B and Lovely Jun) are allowed to make a sport out of gossiping about people and essentially run Mordor like a remake of the Mean Girls movie. I honestly find it hard to believe that the chain of command has NEVER NOTICED that the people they force out of their jobs are Latina women,black men,and other minorities,and likewise don't notice that the people they give uncontested promotions to are white men (Lebron) and white women (Meghan The Horse). Honestly,for an analytical organization,it's hard to believe that nobody has ever seen it through the eyes of a minority. This is why representation matters!!!","I have not observed clear indications of 'better treatment' based off of Race,Religion,Sex or sexual prefrence." ))


df %>% 
  get_sentences() %>% 
  sentiment() %>% 
  group_by(Primary.Key,element_id) %>% 
  summarise(comment = paste(Original.Text,collapse = " "),sentiment = mean(sentiment)) %>% 
  ungroup()

# A tibble: 3 x 4
  Primary.Key element_id comment                                                sentiment
  <chr>            <int> <chr>                                                      <dbl>
1 Google_e1            1 I feel as though awards are fairly awarded,but the t~   -0.0410
2 Google_e3            2 Our all white,all 40 or 50-something leade~   -0.0735
3 Google_e98           3 I have not observed clear indications of 'better trea~   -0.0943

,

看起来 #forward() 函数旨在解析为行。对于许多分析来说，这是有道理的。无论如何，这是我解决这个问题的方法：

get_sentences

所以，如果你想重新组合成一个人，你可以这样做：

data<-as.data.frame(c("I love it. I do not hate it"))
data$who<-"Bob"  # since you didn't share data i added this
colnames(data)[1]<-"mysent"

data %>%
  get_sentences() %>%
  sentiment() -> data_senti

data_senti  # examine