Lec07: Shiny

Lecture07: Shiny

感情分析の補足：

NRC Word-Emotion Association Lexicon

ライブラリの読み込み

library(syuzhet)

Sentiment analysis of an Aljazeera article

• テキストファイルの1行ずつ読み込み

text1_Lines <- readLines("G7/Aljazeera.txt")
text1_Lines[1]

[1] "Top diplomats from the Group of Seven (G7) have called for “humanitarian pauses” in Israel’s bombardment in order to deliver aid to desperate Palestinian civilians in the Gaza Strip."

• トークン化

(text1_1_words <- get_tokens(text1_Lines[1]))

 [1] "top"          "diplomats"    "from"         "the"         
 [5] "group"        "of"           "seven"        "g7"          
 [9] "have"         "called"       "for"          "humanitarian"
[13] "pauses"       "in"           "israel"       "s"           
[17] "bombardment"  "in"           "order"        "to"          
[21] "deliver"      "aid"          "to"           "desperate"   
[25] "palestinian"  "civilians"    "in"           "the"         
[29] "gaza"         "strip"       

• 感情得点の取得

text1_1_sentiment_scores <- get_nrc_sentiment(text1_1_words)

• 行名のインデックスを出現単語に変更

row.names(text1_1_sentiment_scores)

 [1] "1"  "2"  "3"  "4"  "5"  "6"  "7"  "8"  "9"  "10" "11" "12" "13" "14"
[15] "15" "16" "17" "18" "19" "20" "21" "22" "23" "24" "25" "26" "27" "28"
[29] "29" "30"

row.names(text1_1_sentiment_scores)<-NULL
row.names(text1_1_sentiment_scores)<-make.unique(text1_1_words)

• 感情得点の相対頻度の算出（総数１）

View(prop.table(text1_1_sentiment_scores[, 1:8]))

“top”のsentiment_score

(tmp<-text1_1_sentiment_scores[1,1:8])

text1_1_sentiment_scores[, 1:8]

sum(text1_1_sentiment_scores[, 1:8])

[1] 9

tmp/sum(text1_1_sentiment_scores[, 1:8])

全テキスト感情得点の取得

text1_string <- get_text_as_string("G7/Aljazeera.txt")
text1_words <- get_tokens(text1_string)
text1_sentiment_scores <- get_nrc_sentiment(text1_words)

感情カテゴリ別集計

SentTotal<-colSums(prop.table(text1_sentiment_scores[, 1:8]))
sort(SentTotal,decreasing=TRUE)

       trust         fear anticipation          joy      sadness 
  0.28571429   0.20000000   0.15714286   0.10000000   0.10000000 
       anger     surprise      disgust 
  0.08571429   0.05714286   0.01428571 

感情カテゴリ内単語情報:trust

id<-which(text1_sentiment_scores$trust>0)
sort(table(text1_words[id]),decreasing=TRUE)

        food humanitarian         bank      medical    statement 
           3            3            2            2            2 
      agreed    committed     continue      leading        peace 
           1            1            1            1            1 
     shelter          top       united 
           1            1            1 

感情カテゴリ内単語情報:fear

id<-which(text1_sentiment_scores$fear>0)
sort(table(text1_words[id]),decreasing=TRUE)

         war  bombardment      medical        broke condemnation 
           3            2            2            1            1 
   displaced         fire     military       urgent     violence 
           1            1            1            1            1 

視覚化: Rader chart

fmsbパッケージのインストール

install.packages("fmsb")

ライブラリの読み込み

library("fmsb")

radarchart関数で描画

参考サイトのコードを土台に作成

#最大、最小データの準備
index<-1:8
SentimentLabels <- names(text1_sentiment_scores)
maxmin <- data.frame(
  lapply(index,function(i) assign(SentimentLabels[i], c(0.3,0), envir = globalenv()) )
)
names(maxmin)<-SentimentLabels[index]

scores <-colSums(prop.table(text1_sentiment_scores[, 1:8]))
dat <- data.frame(
  lapply(index,function(i) assign(sentiments[i], scores[i], envir = globalenv()) )
  )
names(dat)<-SentimentLabels[index]
dat <- rbind(maxmin, dat) #データの結合

####radarchartの設定##### 
#centerzero = TRUEで中心が0
#axistype:軸基準設定,0:無し, 1:割合, 2:実数, 3:割合,実数, 4:最大を1, 5:最大を1,実数
#seg:分割数
#plty:線の種類
#vlcex:ラベルの大きさ
radarchart(dat, axistype = 2, seg = 8, plty = 1, pcol="blue", vlcex = 1.,
           centerzero = TRUE, vlabels = colnames(dat),cglcol ="gray",
           title = "NRC Emotions in AL JAZEERA Text (8 Nov 2023)")

# default: pcol=1:8 (1:black, 2:red, 3:green, 4:blue, 5:cyan, 6:magenta, 7:yellow, 8:gray)

Sentiment analysis of an BBC articleの感情得点

text2_string <- get_text_as_string("G7/BBC.txt")
text2_words <- get_tokens(text2_string)
text2_sentiment_scores <- get_nrc_sentiment(text2_words)

radarchart関数で描画

#最大、最小データの準備
index<-1:8
SentimentLabels <- names(text1_sentiment_scores)
maxmin <- data.frame(
  lapply(index,function(i) assign(SentimentLabels[i], c(0.3,0), envir = globalenv()) )
)
names(maxmin)<-SentimentLabels[index]

scores1<-colSums(prop.table(text1_sentiment_scores[, 1:8]))
scores2<-colSums(prop.table(text2_sentiment_scores[, 1:8]))
scores <-mapply(c,scores1,scores2)

RNGkind("Mersenne-Twister")
dat <- data.frame(
  lapply(index,function(i) assign(sentiments[i], scores[,i], envir = globalenv()) )
  )
names(dat)<-SentimentLabels[index]
dat <- rbind(maxmin, dat)

radarchart(dat, axistype = 2, seg = 8, plty = 1, vlcex = 1., pcol=c(2,3),
           centerzero = TRUE, vlabels = colnames(dat), cglcol ="gray",
           title = "NRC Emotions in AL JAZEERA & BBC (8 Nov 2023)")

shiny

Shiny apps Demo

shinyパッケージのインストール

install.packages("shiny", dependencies = TRUE)

shinyパッケージのロード

library(shiny)

Shiny app作成手順

1. Shiny appのフォルダを作成
2. ui.R, server.R, (global.R)のスクリプトファイルを作成
3. runApp(フォルダ名)で実行

• Tutorial

app_hoge

• server.R
• ui.R ### Shinyアプリケーション”app_hoge”の実行

runApp("app_hoge")

app_hoge2

• server.R
• ui.R
• global.R

Shinyアプリケーション”app_hoge2”の実行

runApp("app_hoge2")

プロットマーカーのアプリケーション1”app_pch1”の実行

• PCH Symbols

runApp("app_pch1")

プロットマーカーのアプリケーション2”app_pch2”の実行

runApp("app_pch2")

Shinyでアプリケーションを作成する際の注意点

• UI（外観）部分から作り始める
• 少しずつコードを書いて、こまめに動作確認