第五周：以情緒分析ptt福原愛事件
tina
2021/3/25
系統參數設定
Sys.setlocale(category = "LC_ALL", locale = "zh_TW.UTF-8") # 避免中文亂碼## [1] ""安裝需要的packages
# echo = T,results = 'hide'
packages = c("dplyr", "tidytext", "stringr", "wordcloud2", "ggplot2",'readr','data.table','reshape2','wordcloud','tidyr','scales')
existing = as.character(installed.packages()[,1])
for(pkg in packages[!(packages %in% existing)]) install.packages(pkg)讀進library
library(dplyr)
library(stringr)
library(tidytext)
library(wordcloud2)
library(data.table)
library(ggplot2)
library(reshape2)
library(wordcloud)
library(tidyr)
library(readr)
library(scales)
require(jiebaR)1. 資料基本介紹
- 資料來源: 文字平台收集PTT Gossip和Woman版2021-03-02 ~ 2021-03-25 所有文章
- 資料集： ptt福原愛事件.csv
- 關鍵字：福原愛、江宏傑、婚變、外遇、離婚、出軌
- 資料時間：2021-03-02 ~ 2021-03-25
# 把文章和留言讀進來
MetaData = fread('ptt_AIarticleMetaData.csv',encoding = 'UTF-8')
Reviews = fread('ptt_AIarticleReviews.csv',encoding = 'UTF-8')
# 再篩一次文章 1005 篇
keywords = c('福原愛','江宏傑','外遇')
toMatch = paste(keywords,collapse="|")
MetaData = with(MetaData, MetaData[grepl(toMatch,sentence)|grepl(toMatch,artTitle),])
# 挑選文章對應的留言
Reviews = left_join(MetaData, Reviews[,c("artUrl", "cmtContent")], by = "artUrl")2. 資料處理
(1). 文章斷詞
- 設定斷詞引擎
# 加入自定義的字典
jieba_tokenizer <- worker(user="dict/user_dict.txt", stop_word = "stop_words.txt")
# 設定斷詞function
customized_tokenizer <- function(t) {
lapply(t, function(x) {
tokens <- segment(x, jieba_tokenizer)
return(tokens)
})
}# 把文章和留言的斷詞結果併在一起
MToken <- MetaData %>% unnest_tokens(word, sentence, token=customized_tokenizer)
RToken <- Reviews %>% unnest_tokens(word, cmtContent, token=customized_tokenizer)
# 把資料併在一起
data <- rbind(MToken[,c("artDate","artUrl", "word")],RToken[,c("artDate","artUrl", "word")]) (2). 資料基本清理
- 日期格式化
- 去除特殊字元、詞頻太低的字
# 格式化日期欄位
data$artDate= data$artDate %>% as.Date("%Y/%m/%d")
# 過濾特殊字元
data_select = data %>%
filter(!grepl('[[:punct:]]',word)) %>% # 去標點符號
filter(!grepl("['^0-9a-z']",word)) %>% # 去英文、數字
filter(nchar(.$word)>1)
# 算每天不同字的詞頻
# word_count:artDate,word,count
word_count <- data_select %>%
select(artDate,word) %>%
group_by(artDate,word) %>%
summarise(count=n()) %>% # 算字詞單篇總數用summarise
filter(count>3) %>% # 過濾出現太少次的字
arrange(desc(count))## `summarise()` has grouped output by 'artDate'. You can override using the `.groups` argument.word_count## # A tibble: 12,151 x 3
## # Groups: artDate [23]
## artDate word count
## <date> <chr> <int>
## 1 2021-03-03 日本 1274
## 2 2021-03-04 日本 1129
## 3 2021-03-03 台灣 822
## 4 2021-03-03 福原 815
## 5 2021-03-10 日本 756
## 6 2021-03-04 台灣 730
## 7 2021-03-04 福原 640
## 8 2021-03-03 外遇 602
## 9 2021-03-04 外遇 592
## 10 2021-03-06 日本 567
## # ... with 12,141 more rows3. 準備LIWC字典
讀檔，字詞間以“,”將字分隔
P <- read_file("positive.txt") # 正向字典txt檔
N <- read_file("negative.txt") # 負向字典txt檔
#字典txt檔讀進來是一整個字串
typeof(P)## [1] "character"分割字詞，並將兩個情緒字典併在一起
# 將字串依,分割
# strsplit回傳list , 我們取出list中的第一個元素
P = strsplit(P, ",")[[1]]
N = strsplit(N, ",")[[1]]
# 建立dataframe 有兩個欄位word,sentiments，word欄位內容是字典向量
P = data.frame(word = P, sentiment = "positive") #664
N = data.frame(word = N, sentiment = "negative") #1047
# 把兩個字典拼在一起
LIWC = rbind(P, N)
# 檢視字典
head(LIWC)## word sentiment
## 1 一流 positive
## 2 下定決心 positive
## 3 不拘小節 positive
## 4 不費力 positive
## 5 不錯 positive
## 6 主動 positive4. 將文章和與LIWC情緒字典做合併
每日發文數量
MetaData$artDate= MetaData$artDate %>% as.Date("%Y/%m/%d")
MetaData %>%
group_by(artDate) %>%
summarise(count = n()) %>%
ggplot()+
geom_line(aes(x=artDate,y=count))+
scale_x_date(labels = date_format("%m/%d"))
算出每天情緒總和(sentiment_count)
# sentiment_count:artDate,sentiment,count
sentiment_count = data_select %>%
select(artDate,word) %>%
inner_join(LIWC) %>%
group_by(artDate,sentiment) %>%
summarise(count=n()) ## Joining, by = "word"## `summarise()` has grouped output by 'artDate'. You can override using the `.groups` argument.正負情緒分數折線圖
# 檢視資料的日期區間
range(sentiment_count$artDate) #"2021-03-02" "2021-03-25"## [1] "2021-03-02" "2021-03-24"sentiment_count %>%
ggplot()+
geom_line(aes(x=artDate,y=count,colour=sentiment))+
scale_x_date(labels = date_format("%m/%d"),
limits = as.Date(c('2021-03-01','2021-03-20')),
date_breaks = "2 days")## Warning: Removed 8 row(s) containing missing values (geom_path).
正負情緒比例折線圖
sentiment_count %>%
# 標準化的部分
group_by(artDate) %>%
mutate(ratio = count/sum(count)) %>%
# 畫圖的部分
ggplot()+
geom_line(aes(x=artDate,y=ratio,colour=sentiment))+
scale_x_date(labels = date_format("%m/%d"),
limits = as.Date(c('2021-03-01','2021-03-20')),
date_breaks = "2 days")+
# 加上標示日期的線
geom_vline(aes(xintercept = as.numeric(artDate[which(sentiment_count$artDate == as.Date('2021-03-18'))
[1]])),colour = "red")## Warning: Removed 8 row(s) containing missing values (geom_path).
# 查看每天的情緒分數排名
sentiment_count %>%
select(count,artDate) %>%
group_by(artDate) %>%
summarise(sum = sum(count)) %>%
arrange(desc(sum))## # A tibble: 23 x 2
## artDate sum
## <date> <int>
## 1 2021-03-04 5694
## 2 2021-03-03 5613
## 3 2021-03-10 3416
## 4 2021-03-05 1326
## 5 2021-03-11 1313
## 6 2021-03-06 1148
## 7 2021-03-09 1140
## 8 2021-03-12 760
## 9 2021-03-20 696
## 10 2021-03-24 386
## # ... with 13 more rows5. 畫出文字雲
3/3為情緒折線圖高點，因此進一步分析當天討論熱度較高的文字雲。
word_count %>%
filter(!(word %in% c("福原愛", "江宏傑", "福原", "外遇", "還要", "八卦"))) %>%
filter(artDate == as.Date('2021-03-03')) %>%
select(word,count) %>%
group_by(word) %>%
summarise(count = sum(count)) %>%
arrange(desc(count)) %>%
filter(count>50) %>% # 過濾出現太少次的字
wordcloud2()## Adding missing grouping variables: `artDate`3/10出現另一波情緒高峰
word_count %>%
filter(!(word %in% c("福原愛", "江宏傑", "福原", "外遇", "還要", "八卦"))) %>%
filter(artDate == as.Date('2021-03-10')) %>%
select(word,count) %>%
group_by(word) %>%
summarise(count = sum(count)) %>%
arrange(desc(count)) %>%
filter(count>30) %>% # 過濾出現太少次的字
wordcloud2()## Adding missing grouping variables: `artDate`3/18當天正負情緒出現逆轉
word_count %>%
filter(!(word %in% c("福原愛", "江宏傑", "福原", "外遇", "還要", "八卦"))) %>%
filter(artDate == as.Date('2021-03-18')) %>%
select(word,count) %>%
group_by(word) %>%
summarise(count = sum(count)) %>%
arrange(desc(count)) %>%
filter(count>5) %>%
wordcloud2()## Adding missing grouping variables: `artDate`正負情緒代表字
# sentiment_sum:word,sentiment,sum
sentiment_sum <-
word_count %>%
inner_join(LIWC) %>%
group_by(word,sentiment) %>%
summarise(
sum = sum(count)
) %>%
arrange(desc(sum)) %>%
data.frame() ## Joining, by = "word"## `summarise()` has grouped output by 'word'. You can override using the `.groups` argument.sentiment_sum %>%
top_n(30,wt = sum) %>%
mutate(word = reorder(word, sum)) %>%
ggplot(aes(word, sum, fill = sentiment)) +
geom_col(show.legend = FALSE) +
facet_wrap(~sentiment, scales = "free_y") +
labs(y = "Contribution to sentiment",
x = NULL) +
theme(text=element_text(size=14))+
coord_flip()
6.歸類正負面文章
之前的情緒分析大部分是全部的詞彙加總，接下來將正負面情緒的文章分開，看看能不能發現一些新的東西。接下來歸類文章，將每一篇文章正負面情緒的分數算出來，然後大概分類文章屬於正面還是負面。
# 依據情緒值的正負比例歸類文章
article_type =
data_select %>%
inner_join(LIWC) %>%
group_by(artUrl,sentiment) %>%
summarise(count=n()) %>%
spread(sentiment,count,fill = 0) %>% #把正負面情緒展開，缺值補0
mutate(type = case_when(positive > negative ~ "positive",
TRUE ~ "negative")) %>%
data.frame() ## Joining, by = "word"## `summarise()` has grouped output by 'artUrl'. You can override using the `.groups` argument.# 看一下正負比例的文章各有幾篇
article_type %>%
group_by(type) %>%
summarise(count = n())## # A tibble: 2 x 2
## type count
## <chr> <int>
## 1 negative 645
## 2 positive 297從情緒文章數量統計圖可以看出，事件爆發第一天討論度是最高，接著到了3/10又有另一波討論。
正負情緒文章數量統計圖
#
article_type_date = left_join(article_type[,c("artUrl", "type")], MetaData[,c("artUrl", "artDate")], by = "artUrl")
article_type_date %>%
group_by(artDate,type) %>%
summarise(count = n()) %>%
ggplot(aes(x = artDate, y = count, fill = type)) +
geom_bar(stat = "identity", position = "dodge")+
scale_x_date(labels = date_format("%m/%d"),
limits = as.Date(c('2021-03-02','2021-03-25')),
date_breaks = "2 days")## `summarise()` has grouped output by 'artDate'. You can override using the `.groups` argument.## Warning: Removed 1 rows containing missing values (geom_bar).
把正面和負面的文章挑出來，並和斷詞結果合併。
# negative_article:artUrl,word
negative_article <-
article_type %>%
filter(type=="negative")%>%
select(artUrl) %>%
left_join(data_select[,c("artUrl", "word")], by = "artUrl")
# positive_article:artUrl,word
positive_article <-
article_type %>%
filter(type=="positive")%>%
select(artUrl) %>%
left_join(data_select[,c("artUrl", "word")], by = "artUrl")畫出正負面文章情緒貢獻度較高的關鍵字
情緒關鍵字:負面情緒文章
# 負面情緒關鍵字貢獻圖
negative_article %>%
inner_join(LIWC) %>%
group_by(word,sentiment) %>%
summarise(
sum = n()
)%>%
arrange(desc(sum)) %>%
data.frame() %>%
top_n(30,wt = sum) %>%
ungroup() %>%
mutate(word = reorder(word, sum)) %>%
ggplot(aes(word, sum, fill = sentiment)) +
geom_col(show.legend = FALSE) +
facet_wrap(~sentiment, scales = "free_y") +
labs(y = "Contribution to negative sentiment",
x = NULL) +
theme(text=element_text(size=14))+
coord_flip()## Joining, by = "word"## `summarise()` has grouped output by 'word'. You can override using the `.groups` argument.
情緒關鍵字:正面情緒文章
# 正面情緒關鍵字貢獻圖
positive_article %>%
inner_join(LIWC) %>%
group_by(word,sentiment) %>%
summarise(
sum = n()
)%>%
arrange(desc(sum)) %>%
data.frame() %>%
top_n(30,wt = sum) %>%
ungroup() %>%
mutate(word = reorder(word, sum)) %>%
ggplot(aes(word, sum, fill = sentiment)) +
geom_col(show.legend = FALSE) +
facet_wrap(~sentiment, scales = "free_y") +
labs(y = "Contribution to positive sentiment",
x = NULL) +
theme(text=element_text(size=14))+
coord_flip()## Joining, by = "word"## `summarise()` has grouped output by 'word'. You can override using the `.groups` argument.