in this mini project I will use quRan, quanteda.corpora, seededlda, quanteda, and lubridate packages. Those packages are helpful for text mining tasks
library(quRan)
library(quanteda.corpora)
library(seededlda)
library(quanteda)
library(lubridate)I select Yusuf Ali’s Quran translation
df <- quran_en_yusufali
df <- subset(df, select = c(text, surah_title_en, surah, juz, revelation_type))corpus_quran_en <- corpus(df, text_field = "text")
head(corpus_quran_en)Corpus consisting of 6 documents and 4 docvars.
text1 :
"In the name of Allah, Most Gracious, Most Merciful."
text2 :
"Praise be to Allah, the Cherisher and Sustainer of the world..."
text3 :
"Most Gracious, Most Merciful;"
text4 :
"Master of the Day of Judgment."
text5 :
"Thee do we worship, and Thine aid we seek."
text6 :
"Show us the straight way,"tokens_quran <- tokens(corpus_quran_en, remove_punct = T,
remove_numbers = T, remove_symbols = T)
head(tokens_quran)Tokens consisting of 6 documents and 4 docvars.
text1 :
[1] "In" "the" "name" "of" "Allah" "Most" "Gracious" "Most" "Merciful"
text2 :
[1] "Praise" "be" "to" "Allah" "the" "Cherisher" "and" "Sustainer" "of" "the" "worlds"
text3 :
[1] "Most" "Gracious" "Most" "Merciful"
text4 :
[1] "Master" "of" "the" "Day" "of" "Judgment"
text5 :
[1] "Thee" "do" "we" "worship" "and" "Thine" "aid" "we" "seek"
text6 :
[1] "Show" "us" "the" "straight" "way" I will remove stopwords, because they are not useful for analysis
tokens_quran <- tokens_remove(tokens_quran, pattern = c(stopwords("en")))
head(tokens_quran)Tokens consisting of 6 documents and 4 docvars.
text1 :
[1] "name" "Allah" "Gracious" "Merciful"
text2 :
[1] "Praise" "Allah" "Cherisher" "Sustainer" "worlds"
text3 :
[1] "Gracious" "Merciful"
text4 :
[1] "Master" "Day" "Judgment"
text5 :
[1] "Thee" "worship" "Thine" "aid" "seek"
text6 :
[1] "Show" "us" "straight" "way" Any uppercase will be transformed into lowercase
tokens_quran <- tokens_tolower(tokens_quran, keep_acronyms = F)
head(tokens_quran)Tokens consisting of 6 documents and 4 docvars.
text1 :
[1] "name" "allah" "gracious" "merciful"
text2 :
[1] "praise" "allah" "cherisher" "sustainer" "worlds"
text3 :
[1] "gracious" "merciful"
text4 :
[1] "master" "day" "judgment"
text5 :
[1] "thee" "worship" "thine" "aid" "seek"
text6 :
[1] "show" "us" "straight" "way" document feature matrix will be constructed for LDA
dfm_quran <- dfm(tokens_quran) %>%
dfm_trim(min_termfreq = 0.8, termfreq_type = "quantile",
max_docfreq = 0.1, docfreq_type = "prop")
head(dfm_quran)Document-feature matrix of: 6 documents, 1,304 features (99.76% sparse) and 4 docvars.
features
docs name gracious merciful praise cherisher worlds day judgment thee worship
text1 1 1 1 0 0 0 0 0 0 0
text2 0 0 0 1 1 1 0 0 0 0
text3 0 1 1 0 0 0 0 0 0 0
text4 0 0 0 0 0 0 1 1 0 0
text5 0 0 0 0 0 0 0 0 1 1
text6 0 0 0 0 0 0 0 0 0 0
[ reached max_nfeat ... 1,294 more features ]I only choose ten topics in this topic modelling
# start text modelling
text_model_lda <- textmodel_lda(dfm_quran, k = 10)
terms(text_model_lda, 10) topic1 topic2 topic3 topic4 topic5 topic6 topic7 topic8 topic9 topic10
[1,] "earth" "believe" "shall" "day" "people" "way" "know" "thou" "sent" "men"
[2,] "heavens" "good" "penalty" "shall" "o" "call" "things" "thee" "signs" "one"
[3,] "things" "reward" "fire" "judgment" "moses" "one" "full" "thy" "thee" "man"
[4,] "created" "give" "therein" "every" "behold" "follow" "knowledge" "us" "book" "let"
[5,] "made" "deeds" "evil" "soul" "pharaoh" "can" "verily" "turn" "hath" "two"
[6,] "verily" "shall" "reject" "can" "among" "even" "hearts" "away" "people" "make"
[7,] "night" "life" "faith" "us" "gave" "besides" "hath" "see" "truth" "women"
[8,] "see" "hereafter" "grievous" "one" "came" "worship" "well" "art" "clear" "among"
[9,] "power" "evil" "hell" "together" "indeed" "upon" "merciful" "o" "unto" "fear"
[10,] "forth" "messenger" "companions" "come" "remember" "men" "thy" "put" "messenger" "o" With Seeded LDA we can choose our own topic, let’s say we want to only want topics about war, politics, and economy. So we create the small dictionary contains related keywords.
dictionary_lda <- dictionary(file = "topics.yml")Warning in readLines(con, warn = readLines.warn) :
incomplete final line found on 'topics.yml'dictionary_ldaDictionary object with 5 key entries.
- [war]:
- war, sword, kill*, conflict, battle, fight*, attack*, crusade
- [politics]:
- parliament*, congress*, white house, party leader*, party member*, voter*, lawmaker*, politician*
- [economy]:
- money, trade*, market, wage, wealth
- [law]:
- rule, justice, court, regulation, concept, legal, right*
- [eschatology]:
- afterlife, destiny, hell, heaven, end, punish*, fireSo let’s see the result
text_model_slda <- textmodel_seededlda(dfm_quran, dictionary = dictionary_lda)
terms(text_model_slda, 10) war politics economy law eschatology
[1,] "fight" "may" "wealth" "thou" "fire"
[2,] "war" "men" "people" "right" "punishment"
[3,] "fighting" "made" "sent" "righteous" "hell"
[4,] "earth" "one" "thee" "righteousness" "end"
[5,] "things" "fear" "signs" "justice" "heaven"
[6,] "heavens" "two" "book" "thy" "punish"
[7,] "power" "good" "o" "thee" "punished"
[8,] "verily" "night" "among" "us" "shall"
[9,] "created" "let" "moses" "believe" "day"
[10,] "know" "women" "believe" "o" "penalty"