RDataMining slides-text-mining-with-r
1 like820 views
This document provides an overview of text mining techniques and processes for analyzing Twitter data with R. It discusses concepts like term-document matrices, text cleaning, frequent term analysis, word clouds, clustering, topic modeling, sentiment analysis and social network analysis. It then provides a step-by-step example of applying these techniques to Twitter data from an R Twitter account, including retrieving tweets, text preprocessing, building term-document matrices, and various analyses.
![Process†
1. Extract tweets and followers from the Twitter website with R
and the twitteR package
2. With the tm package, clean text by removing punctuations,
numbers, hyperlinks and stop words, followed by stemming
and stem completion
3. Build a term-document matrix
4. Cluster Tweets with text clustering
5. Analyse topics with the topicmodels package
6. Analyse sentiment with the sentiment140 package
7. Analyse following/followed and retweeting relationships with
the igraph package
†
More details in paper titled Analysing Twitter Data with Text Mining and
Social Network Analysis [Zhao, 2013].
19 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-22-320.jpg)
![(n.tweet <- tweets %>% length())
## [1] 448
# convert tweets to a data frame
tweets.df <- tweets %>% twListToDF()
# tweet #1
tweets.df[1, c("id", "created", "screenName", "replyToSN",
"favoriteCount", "retweetCount", "longitude", "latitude", "text")]
## id created screenName replyToSN
## 1 697031245503418368 2016-02-09 12:16:13 RDataMining <NA>
## favoriteCount retweetCount longitude latitude
## 1 13 14 NA NA
## ...
## 1 A Twitter dataset for text mining: @RDataMining Tweets ex...
# print tweet #1 and make text fit for slide width
tweets.df$text[1] %>% strwrap(60) %>% writeLines()
## A Twitter dataset for text mining: @RDataMining Tweets
## extracted on 3 February 2016. Download it at
## https://t.co/lQp94IvfPf
21 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-24-320.jpg)
![Text Cleaning Functions
Convert to lower case: tolower
Remove punctuation: removePunctuation
Remove numbers: removeNumbers
Remove URLs
Remove stop words (like ’a’, ’the’, ’in’): removeWords,
stopwords
Remove extra white space: stripWhitespace
## text cleaning
library(tm)
# function for removing URLs, i.e.,
# "http" followed by any non-space letters
removeURL <- function(x) gsub("http[^[:space:]]*", "", x)
# function for removing anything other than English letters or space
removeNumPunct <- function(x) gsub("[^[:alpha:][:space:]]*", "", x)
# customize stop words
myStopwords <- c(setdiff(stopwords('english'), c("r", "big")),
"use", "see", "used", "via", "amp")
See details of regular expressions by running ?regex in R console.
22 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-25-320.jpg)
![Stemming and Stem Completion ‡
## stem words
corpus.stemmed <- corpus.cleaned %>% tm_map(stemDocument)
## stem completion
stemCompletion2 <- function(x, dictionary) {
x <- unlist(strsplit(as.character(x), " "))
x <- x[x != ""]
x <- stemCompletion(x, dictionary=dictionary)
x <- paste(x, sep="", collapse=" ")
stripWhitespace(x)
}
corpus.completed <- corpus.stemmed %>%
lapply(stemCompletion2, dictionary=corpus.cleaned) %>%
VectorSource() %>% Corpus()
‡
http://stackoverflow.com/questions/25206049/stemcompletion-is-not-working
24 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-27-320.jpg)
![Before/After Text Cleaning and Stemming
## compare text before/after cleaning
# original text
corpus.raw[[1]]$content %>% strwrap(60) %>% writeLines()
## A Twitter dataset for text mining: @RDataMining Tweets
## extracted on 3 February 2016. Download it at
## https://t.co/lQp94IvfPf
# after basic cleaning
corpus.cleaned[[1]]$content %>% strwrap(60) %>% writeLines()
## twitter dataset text mining rdatamining tweets extracted
## february download
# stemmed text
corpus.stemmed[[1]]$content %>% strwrap(60) %>% writeLines()
## twitter dataset text mine rdatamin tweet extract februari
## download
# after stem completion
corpus.completed[[1]]$content %>% strwrap(60) %>% writeLines()
## twitter dataset text miner rdatamining tweet extract
## download 25 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-28-320.jpg)
![Build Term Document Matrix
## Build Term Document Matrix
tdm <- corpus.completed %>%
TermDocumentMatrix(control = list(wordLengths = c(1, Inf))) %>%
print
## <<TermDocumentMatrix (terms: 1073, documents: 448)>>
## Non-/sparse entries: 3594/477110
## Sparsity : 99%
## Maximal term length: 23
## Weighting : term frequency (tf)
idx <- which(dimnames(tdm)$Terms %in% c("r", "data", "mining"))
tdm[idx, 21:30] %>% as.matrix()
## Docs
## Terms 21 22 23 24 25 26 27 28 29 30
## mining 0 0 0 0 1 0 0 0 0 1
## data 0 1 0 0 1 0 0 0 0 1
## r 1 1 1 1 0 1 0 1 1 1
27 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-30-320.jpg)
![Top Frequent Terms
# inspect frequent words
freq.terms <- tdm %>% findFreqTerms(lowfreq = 20) %>% print
## [1] "mining" "rdatamining" "text" "analytics"
## [5] "australia" "data" "canberra" "group"
## [9] "university" "science" "slide" "tutorial"
## [13] "big" "learn" "package" "r"
## [17] "network" "course" "introduction" "talk"
## [21] "analysing" "research" "position" "example"
term.freq <- tdm %>% as.matrix() %>% rowSums()
term.freq <- term.freq %>% subset(term.freq >= 20)
df <- data.frame(term = names(term.freq), freq = term.freq)
28 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-31-320.jpg)
![Wordcloud
## word cloud
m <- tdm %>% as.matrix
# calculate the frequency of words and sort it by frequency
word.freq <- m %>% rowSums() %>% sort(decreasing = T)
# colors
library(RColorBrewer)
pal <- brewer.pal(9, "BuGn")[-(1:4)]
# plot word cloud
library(wordcloud)
wordcloud(words = names(word.freq), freq = word.freq, min.freq = 3,
random.order = F, colors = pal)
30 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-33-320.jpg)
![for (i in 1:k) {
cat(paste("cluster ", i, ": ", sep = ""))
s <- sort(kmeansResult$centers[i, ], decreasing = T)
cat(names(s)[1:5], "n")
# print the tweets of every cluster
# print(tweets[which(kmeansResult£cluster==i)])
}
## cluster 1: r example mining data analysing
## cluster 2: data mining r package slide
## cluster 3: r slide data package analysing
## cluster 4: analysing university slide package network
## cluster 5: data mining big analytics canberra
## cluster 6: research data position university mining
37 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-40-320.jpg)
![Topic Modelling
rdm.topics <- topics(lda) # 1st topic identified for every document (twe
rdm.topics <- data.frame(date=as.IDate(tweets.df$created),
topic=rdm.topics)
ggplot(rdm.topics, aes(date, fill = term[topic])) +
geom_density(position = "stack")
0.000
0.002
0.004
0.006
2012 2013 2014 2015 2016
date
density
term[topic]
analysing, network, r, canberra, data, social, present
australia, data, ausdm, submission, workshop, mining, august
data, big, mining, r, research, group, science
data, mining, r, application, book, dataset, associate
data, university, analytics, mining, position, research, scientist
r, data, course, introduction, free, online, mining
r, package, example, useful, program, slide, code
r, talk, slide, series, learn, rdatamining, time
Another way to plot steam graph:
http://menugget.blogspot.com.au/2013/12/data-mountains-and-streams-stacked-area.html 39 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-42-320.jpg)
![Sentiment Analysis
## sentiment analysis install package sentiment140
require(devtools)
install_github("sentiment140", "okugami79")
# sentiment analysis
library(sentiment)
sentiments <- sentiment(tweets.df$text)
table(sentiments$polarity)
# sentiment plot
sentiments$score <- 0
sentiments$score[sentiments$polarity == "positive"] <- 1
sentiments$score[sentiments$polarity == "negative"] <- -1
sentiments$date <- as.IDate(tweets.df$created)
result <- aggregate(score ~ date, data = sentiments, sum)
40 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-43-320.jpg)
![Retrieve User Info and Followers
## follower analysis
user <- getUser("RDataMining")
user$toDataFrame()
friends <- user$getFriends() # who this user follows
followers <- user$getFollowers() # this user's followers
followers2 <- followers[[1]]$getFollowers() # a follower's followers
## [,1] ...
## description "R and Data Mining. Group on LinkedIn: ht...
## statusesCount "583" ...
## followersCount "2376" ...
## favoritesCount "6" ...
## friendsCount "72" ...
## url "http://t.co/LwL50uRmPd" ...
## name "Yanchang Zhao" ...
## created "2011-04-04 09:15:43" ...
## protected "FALSE" ...
## verified "FALSE" ...
## screenName "RDataMining" ...
## location "Australia" ...
## lang "en" ... 41 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-44-320.jpg)
![Top Retweeted Tweets
## retweet analysis
## select top retweeted tweets
table(tweets.df$retweetCount)
selected <- which(tweets.df$retweetCount >= 9)
## plot them
dates <- strptime(tweets.df$created, format="%Y-%m-%d")
plot(x=dates, y=tweets.df$retweetCount, type="l", col="grey",
xlab="Date", ylab="Times retweeted")
colors <- rainbow(10)[1:length(selected)]
points(dates[selected], tweets.df$retweetCount[selected],
pch=19, col=colors)
text(dates[selected], tweets.df$retweetCount[selected],
tweets.df$text[selected], col=colors, cex=.9)
44 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-47-320.jpg)
![Tracking Message Propagation
tweets[[1]]
retweeters(tweets[[1]]$id)
retweets(tweets[[1]]$id)
## [1] "RDataMining: A Twitter dataset for text mining: @RData...
## [1] "197489286" "316875164" "229796464" "3316009302"
## [5] "244077734" "16900353" "2404767650" "222061895"
## [9] "11686382" "190569306" "49413866" "187048879"
## [13] "6146692" "2591996912"
## [[1]]
## [1] "bobaiKato: RT @RDataMining: A Twitter dataset for text...
##
## [[2]]
## [1] "VipulMathur: RT @RDataMining: A Twitter dataset for te...
##
## [[3]]
## [1] "tau_phoenix: RT @RDataMining: A Twitter dataset for te...
The tweet potentially reached around 120,000 users. 46 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-49-320.jpg)
![Online Resources
Book titled R and Data Mining: Examples and Case
Studies [Zhao, 2012]
http://www.rdatamining.com/docs/RDataMining-book.pdf
R Reference Card for Data Mining
http://www.rdatamining.com/docs/RDataMining-reference-card.pdf
Free online courses and documents
http://www.rdatamining.com/resources/
RDataMining Group on LinkedIn (27,000+ members)
http://group.rdatamining.com
Twitter (3,300+ followers)
@RDataMining
58 / 61](https://image.slidesharecdn.com/rdm-slides-text-mining-with-r-190703041151/85/RDataMining-slides-text-mining-with-r-61-320.jpg)
![[IJET V2I3P7] Authors: Muthe Sandhya, Shitole Sarika, Sinha Anukriti, Aghav S...](https://cdn.slidesharecdn.com/ss_thumbnails/ijet-v2i3p7-160609052352-thumbnail.jpg?width=560&fit=bounds)
Ad
More Related Content
Similar to RDataMining slides-text-mining-with-r (20)
More from Yanchang Zhao (17)
Ad
Recently uploaded (20)
Ad
RDataMining slides-text-mining-with-r
- 1. Text Mining with R ∗ Yanchang Zhao http://www.RDataMining.com R and Data Mining Course Beijing University of Posts and Telecommunications, Beijing, China July 2019 ∗ Chapter 10: Text Mining, in R and Data Mining: Examples and Case Studies. http://www.rdatamining.com/docs/RDataMining-book.pdf 1 / 61
- 2. Contents Text Mining Concept Tasks Twitter Data Analysis with R Twitter Extracting Tweets Text Cleaning Frequent Words and Word Cloud Word Associations Clustering Topic Modelling Sentiment Analysis Follower Analysis Retweeting Analysis R Packages Wrap Up Further Readings and Online Resources 2 / 61
- 3. Text Data Text documents in a natural language Unstructured Documents in plain text, Word or PDF format Emails, online chat logs and phone transcripts Online news and forums, blogs, micro-blogs and social media . . . 3 / 61
- 4. Typical Process of Text Mining 1. Transform text into structured data Term-Document Matrix (TDM) Entities and relations . . . 2. Apply traditional data mining techniques to the above structured data Clustering Classification Social Network Analysis (SNA) . . . 4 / 61
- 5. Typical Process of Text Mining (cont.) 5 / 61
- 6. Term-Document Matrix (TDM) Also known as Document-Term Matrix (DTM) A 2D matrix Rows: terms or words Columns: documents Entry mi,j : number of occurrences of term ti in document dj Term weighting schemes: Term Frequency, Binary Weight, TF-IDF, etc. 6 / 61
- 7. TF-IDF Term Frequency (TF) tfi,j : the number of occurrences of term ti in document dj Inverse Document Frequency (IDF) for term ti is: idfi = log2 |D| |{d | ti ∈ d}| (1) |D|: the total number of documents |{d | ti ∈ d}|: the number of documents where term ti appears Term Frequency - Inverse Document Frequency (TF-IDF) tfidf = tfi,j · idfi (2) IDF reduces the weight of terms that occur frequently in documents and increases the weight of terms that occur rarely. 7 / 61
- 8. An Example of TDM Doc1: I like R. Doc2: I like Python. Term Frequency TF-IDF IDF 8 / 61
- 9. An Example of TDM Doc1: I like R. Doc2: I like Python. Term Frequency TF-IDF IDF 8 / 61 Terms that can distinguish different documents are given greater weights.
- 10. An Example of TDM (cont.) Doc1: I like R. Doc2: I like Python. Term Frequency Normalized Term Frequency IDF Normalized TF-IDF 9 / 61
- 11. An Example of Term Weighting in R ## term weighting library(magrittr) library(tm) ## package for text mining a <- c("I like R", "I like Python") ## build corpus b <- a %>% VectorSource() %>% Corpus() ## build term document matrix m <- b %>% TermDocumentMatrix(control=list(wordLengths=c(1, Inf))) m %>% inspect() ## various term weighting schemes m %>% weightBin() %>% inspect() ## binary weighting m %>% weightTf() %>% inspect() ## term frequency m %>% weightTfIdf(normalize=F) %>% inspect() ## TF-IDF m %>% weightTfIdf(normalize=T) %>% inspect() ## normalized TF-IDF More options provided in package tm: weightSMART WeightFunction 10 / 61
- 12. Text Mining Tasks Text classification Text clustering and categorization Topic modelling Sentiment analysis Document summarization Entity and relation extraction . . . 11 / 61
- 13. Topic Modelling To identify topics in a set of documents It groups both documents that use similar words and words that occur in a similar set of documents. Intuition: Documents related to R would contain more words like R, ggplot2, plyr, stringr, knitr and other R packages, than Python related keywords like Python, NumPy, SciPy, Matplotlib, etc. A document can be of multiple topics in different proportions. For instance, a document can be 90% about R and 10% about Python. ⇒ soft/fuzzy clustering Latent Dirichlet Allocation (LDA): the most widely used topic model 12 / 61
- 14. Sentiment Analysis Also known as opinion mining To determine attitude, polarity or emotions from documents Polarity: positive, negative, netural Emotions: angry, sad, happy, bored, afraid, etc. Method: 1. identify invidual words and phrases and map them to different emotional scales 2. adjust the sentiment value of a concept based on modifications surrounding it 13 / 61
- 15. Document Summarization To create a summary with major points of the orignial document Approaches Extraction: select a subset of existing words, phrases or sentences to build a summary Abstraction: use natural language generation techniques to build a summary that is similar to natural language 14 / 61
- 16. Entity and Relationship Extraction Named Entity Recognition (NER): identify named entities in text into pre-defined categories, such as person names, organizations, locations, date and time, etc. Relationship Extraction: identify associations among entities Example: Ben lives at 5 Geroge St, Sydney. 15 / 61
- 17. Entity and Relationship Extraction Named Entity Recognition (NER): identify named entities in text into pre-defined categories, such as person names, organizations, locations, date and time, etc. Relationship Extraction: identify associations among entities Example: Ben lives at 5 Geroge St, Sydney. 15 / 61
- 18. Entity and Relationship Extraction Named Entity Recognition (NER): identify named entities in text into pre-defined categories, such as person names, organizations, locations, date and time, etc. Relationship Extraction: identify associations among entities Example: Ben lives at 5 Geroge St, Sydney. 15 / 61 Ben 5 Geroge St, Sydney
- 19. Contents Text Mining Concept Tasks Twitter Data Analysis with R Twitter Extracting Tweets Text Cleaning Frequent Words and Word Cloud Word Associations Clustering Topic Modelling Sentiment Analysis Follower Analysis Retweeting Analysis R Packages Wrap Up Further Readings and Online Resources 16 / 61
- 20. Twitter An online social networking service that enables users to send and read short 280-character (used to be 140 before November 2017) messages called “tweets” (Wikipedia) Over 300 million monthly active users (as of 2018) Creating over 500 million tweets per day 17 / 61
- 21. RDataMining Twitter Account 18 / 61
- 22. Process† 1. Extract tweets and followers from the Twitter website with R and the twitteR package 2. With the tm package, clean text by removing punctuations, numbers, hyperlinks and stop words, followed by stemming and stem completion 3. Build a term-document matrix 4. Cluster Tweets with text clustering 5. Analyse topics with the topicmodels package 6. Analyse sentiment with the sentiment140 package 7. Analyse following/followed and retweeting relationships with the igraph package † More details in paper titled Analysing Twitter Data with Text Mining and Social Network Analysis [Zhao, 2013]. 19 / 61
- 23. Retrieve Tweets ## Option 1: retrieve tweets from Twitter library(twitteR) library(ROAuth) ## Twitter authentication setup_twitter_oauth(consumer_key, consumer_secret, access_token, access_ ## 3200 is the maximum to retrieve tweets <- "RDataMining" %>% userTimeline(n = 3200) See details of Twitter Authentication with OAuth in Section 3 of http://geoffjentry.hexdump.org/twitteR.pdf. ## Option 2: download @RDataMining tweets from RDataMining.com library(twitteR) url <- "http://www.rdatamining.com/data/RDataMining-Tweets-20160212.rds" download.file(url, destfile = "./data/RDataMining-Tweets-20160212.rds") ## load tweets into R tweets <- readRDS("./data/RDataMining-Tweets-20160212.rds") 20 / 61
- 24. (n.tweet <- tweets %>% length()) ## [1] 448 # convert tweets to a data frame tweets.df <- tweets %>% twListToDF() # tweet #1 tweets.df[1, c("id", "created", "screenName", "replyToSN", "favoriteCount", "retweetCount", "longitude", "latitude", "text")] ## id created screenName replyToSN ## 1 697031245503418368 2016-02-09 12:16:13 RDataMining <NA> ## favoriteCount retweetCount longitude latitude ## 1 13 14 NA NA ## ... ## 1 A Twitter dataset for text mining: @RDataMining Tweets ex... # print tweet #1 and make text fit for slide width tweets.df$text[1] %>% strwrap(60) %>% writeLines() ## A Twitter dataset for text mining: @RDataMining Tweets ## extracted on 3 February 2016. Download it at ## https://t.co/lQp94IvfPf 21 / 61
- 25. Text Cleaning Functions Convert to lower case: tolower Remove punctuation: removePunctuation Remove numbers: removeNumbers Remove URLs Remove stop words (like ’a’, ’the’, ’in’): removeWords, stopwords Remove extra white space: stripWhitespace ## text cleaning library(tm) # function for removing URLs, i.e., # "http" followed by any non-space letters removeURL <- function(x) gsub("http[^[:space:]]*", "", x) # function for removing anything other than English letters or space removeNumPunct <- function(x) gsub("[^[:alpha:][:space:]]*", "", x) # customize stop words myStopwords <- c(setdiff(stopwords('english'), c("r", "big")), "use", "see", "used", "via", "amp") See details of regular expressions by running ?regex in R console. 22 / 61
- 26. Text Cleaning # build a corpus and specify the source to be character vectors corpus.raw <- tweets.df$text %>% VectorSource() %>% Corpus() # text cleaning corpus.cleaned <- corpus.raw %>% # convert to lower case tm_map(content_transformer(tolower)) %>% # remove URLs tm_map(content_transformer(removeURL)) %>% # remove numbers and punctuations tm_map(content_transformer(removeNumPunct)) %>% # remove stopwords tm_map(removeWords, myStopwords) %>% # remove extra whitespace tm_map(stripWhitespace) 23 / 61
- 27. Stemming and Stem Completion ‡ ## stem words corpus.stemmed <- corpus.cleaned %>% tm_map(stemDocument) ## stem completion stemCompletion2 <- function(x, dictionary) { x <- unlist(strsplit(as.character(x), " ")) x <- x[x != ""] x <- stemCompletion(x, dictionary=dictionary) x <- paste(x, sep="", collapse=" ") stripWhitespace(x) } corpus.completed <- corpus.stemmed %>% lapply(stemCompletion2, dictionary=corpus.cleaned) %>% VectorSource() %>% Corpus() ‡ http://stackoverflow.com/questions/25206049/stemcompletion-is-not-working 24 / 61
- 28. Before/After Text Cleaning and Stemming ## compare text before/after cleaning # original text corpus.raw[[1]]$content %>% strwrap(60) %>% writeLines() ## A Twitter dataset for text mining: @RDataMining Tweets ## extracted on 3 February 2016. Download it at ## https://t.co/lQp94IvfPf # after basic cleaning corpus.cleaned[[1]]$content %>% strwrap(60) %>% writeLines() ## twitter dataset text mining rdatamining tweets extracted ## february download # stemmed text corpus.stemmed[[1]]$content %>% strwrap(60) %>% writeLines() ## twitter dataset text mine rdatamin tweet extract februari ## download # after stem completion corpus.completed[[1]]$content %>% strwrap(60) %>% writeLines() ## twitter dataset text miner rdatamining tweet extract ## download 25 / 61
- 29. Issues in Stem Completion: “Miner” vs “Mining” # count word frequence wordFreq <- function(corpus, word) { results <- lapply(corpus, function(x) grep(as.character(x), pattern=paste0("<",word)) ) sum(unlist(results)) } n.miner <- corpus.cleaned %>% wordFreq("miner") n.mining <- corpus.cleaned %>% wordFreq("mining") cat(n.miner, n.mining) ## 9 104 # replace old word with new word replaceWord <- function(corpus, oldword, newword) { tm_map(corpus, content_transformer(gsub), pattern=oldword, replacement=newword) } corpus.completed <- corpus.completed %>% replaceWord("miner", "mining") %>% replaceWord("universidad", "university") %>% replaceWord("scienc", "science") 26 / 61
- 30. Build Term Document Matrix ## Build Term Document Matrix tdm <- corpus.completed %>% TermDocumentMatrix(control = list(wordLengths = c(1, Inf))) %>% print ## <<TermDocumentMatrix (terms: 1073, documents: 448)>> ## Non-/sparse entries: 3594/477110 ## Sparsity : 99% ## Maximal term length: 23 ## Weighting : term frequency (tf) idx <- which(dimnames(tdm)$Terms %in% c("r", "data", "mining")) tdm[idx, 21:30] %>% as.matrix() ## Docs ## Terms 21 22 23 24 25 26 27 28 29 30 ## mining 0 0 0 0 1 0 0 0 0 1 ## data 0 1 0 0 1 0 0 0 0 1 ## r 1 1 1 1 0 1 0 1 1 1 27 / 61
- 31. Top Frequent Terms # inspect frequent words freq.terms <- tdm %>% findFreqTerms(lowfreq = 20) %>% print ## [1] "mining" "rdatamining" "text" "analytics" ## [5] "australia" "data" "canberra" "group" ## [9] "university" "science" "slide" "tutorial" ## [13] "big" "learn" "package" "r" ## [17] "network" "course" "introduction" "talk" ## [21] "analysing" "research" "position" "example" term.freq <- tdm %>% as.matrix() %>% rowSums() term.freq <- term.freq %>% subset(term.freq >= 20) df <- data.frame(term = names(term.freq), freq = term.freq) 28 / 61
- 32. ## plot frequent words library(ggplot2) ggplot(df, aes(x=term, y=freq)) + geom_bar(stat="identity") + xlab("Terms") + ylab("Count") + coord_flip() + theme(axis.text=element_text(size=7)) analysing analytics australia big canberra course data example group introduction learn mining network package position r rdatamining research science slide talk text tutorial university 0 50 100 150 200 Count Terms 29 / 61
- 33. Wordcloud ## word cloud m <- tdm %>% as.matrix # calculate the frequency of words and sort it by frequency word.freq <- m %>% rowSums() %>% sort(decreasing = T) # colors library(RColorBrewer) pal <- brewer.pal(9, "BuGn")[-(1:4)] # plot word cloud library(wordcloud) wordcloud(words = names(word.freq), freq = word.freq, min.freq = 3, random.order = F, colors = pal) 30 / 61
- 34. data r mining slide big analysing package research analytics example position university canberra network australia group tutorial rdatamining course talk introduction text science learn scientist useful social series free computational application online code ausdm statistical book program modeling available present submission conference pdf time start cluster lecture workshop user august mapreduce language videosoftware hadoop join twitter seminar machine kdnuggets poll associate dataset melbourne visualisations open postdoctoral vacancies thanks sydney will top th process get cfp april parallel classification job due postdoc knowledge graph document linkedin predicting rstudio may analyst us provided extract new pm event spark informal can easier iapa system senior database web feb large technological stanford now detailed ieee rule detection outlier thursday follow close page find business chapter create engine guidance intern keynote call forecasting google week recent card referencefunction map technique add apache tricks excel kdd access china list acm coursera handling webinar nice today deadline fellow file lab experience sigkdd version distributed area francisco san titled step give ranked sentiment tool build topic industrial notes graphical center download dr little tuesday sept go vs june public extended retrieval search singapore member share high jan natural seattle skills case studies nov visit developed regression risk quick advanced nd spatial interacting crancloud canada published california short fast performance wwwrdataminingcom tweet healthcare support decision facebook make sunday friday algorithm aug sas competition various datacamp australasian official task summit format load please run updated v survey neoj tree march world prof fit together massive sna amazon iselect looking australian oct participation america dynamic mexico source credit management paper forest random improve check link pls simple edited mode result media project contain track initial mid dmapps youtube state southern plot snowfall website answers comment 31 / 61
- 35. Associations # which words are associated with 'r'? tdm %>% findAssocs("r", 0.2) ## $r ## code example series user markdown ## 0.27 0.21 0.21 0.20 0.20 # which words are associated with 'data'? tdm %>% findAssocs("data", 0.2) ## $data ## mining big analytics science poll ## 0.48 0.44 0.31 0.29 0.24 32 / 61
- 36. Network of Terms ## network of terms library(graph) library(Rgraphviz) plot(tdm, term = freq.terms, corThreshold = 0.1, weighting = T) mining rdataminingtextanalytics australia data canberra group university science slide tutorial big learn package r network course introduction talk analysing research position example 33 / 61
- 37. Hierarchical Clustering of Terms ## clustering of terms remove sparse terms m2 <- tdm %>% removeSparseTerms(sparse = 0.95) %>% as.matrix() # calculate distance matrix dist.matrix <- m2 %>% scale() %>% dist() # hierarchical clustering fit <- dist.matrix %>% hclust(method = "ward") 34 / 61
- 38. plot(fit) fit %>% rect.hclust(k = 6) # cut tree into 6 clusters groups <- fit %>% cutree(k = 6) slide big australia analytics canberra research university position package example analysing tutorial network r mining data 10305070 Cluster Dendrogram hclust (*, "ward.D") . Height 35 / 61
- 39. ## k-means clustering of documents m3 <- m2 %>% t() # transpose the matrix to cluster documents (tweets) set.seed(122) # set a fixed random seed to make the result reproducible k <- 6 # number of clusters kmeansResult <- kmeans(m3, k) round(kmeansResult$centers, digits = 3) # cluster centers ## mining analytics australia data canberra university slide ## 1 0.435 0.000 0.000 0.217 0.000 0.000 0.087 ## 2 1.128 0.154 0.000 1.333 0.026 0.051 0.179 ## 3 0.055 0.018 0.009 0.164 0.027 0.009 0.227 ## 4 0.083 0.014 0.056 0.000 0.035 0.097 0.090 ## 5 0.412 0.206 0.098 1.196 0.137 0.039 0.078 ## 6 0.167 0.133 0.133 0.567 0.033 0.233 0.000 ## tutorial big package r network analysing research ## 1 0.043 0.000 0.043 1.130 0.087 0.174 0.000 ## 2 0.026 0.077 0.282 1.103 0.000 0.051 0.000 ## 3 0.064 0.018 0.109 1.127 0.045 0.109 0.000 ## 4 0.056 0.007 0.090 0.000 0.090 0.111 0.000 ## 5 0.059 0.333 0.010 0.020 0.020 0.059 0.020 ## 6 0.000 0.167 0.033 0.000 0.067 0.100 1.233 ## position example ## 1 0.000 1.043 ## 2 0.000 0.026 ## 3 0.000 0.000 36 / 61
- 40. for (i in 1:k) { cat(paste("cluster ", i, ": ", sep = "")) s <- sort(kmeansResult$centers[i, ], decreasing = T) cat(names(s)[1:5], "n") # print the tweets of every cluster # print(tweets[which(kmeansResult£cluster==i)]) } ## cluster 1: r example mining data analysing ## cluster 2: data mining r package slide ## cluster 3: r slide data package analysing ## cluster 4: analysing university slide package network ## cluster 5: data mining big analytics canberra ## cluster 6: research data position university mining 37 / 61
- 41. Topic Modelling dtm <- tdm %>% as.DocumentTermMatrix() library(topicmodels) lda <- LDA(dtm, k = 8) # find 8 topics term <- terms(lda, 7) # first 7 terms of every topic term <- apply(term, MARGIN = 2, paste, collapse = ", ") %>% print ## ... ## "data, big, mining, r, research, group,... ## ... ## "analysing, network, r, canberra, data, social,... ## ... ## "r, talk, slide, series, learn, rdatamini... ## ... ## "r, data, course, introduction, free, online... ## ... ## "data, mining, r, application, book, dataset, a... ## ... ## "r, package, example, useful, program, sli... ## ... ## "data, university, analytics, mining, position, research, s... ## ... ## "australia, data, ausdm, submission, workshop, mining... 38 / 61
- 42. Topic Modelling rdm.topics <- topics(lda) # 1st topic identified for every document (twe rdm.topics <- data.frame(date=as.IDate(tweets.df$created), topic=rdm.topics) ggplot(rdm.topics, aes(date, fill = term[topic])) + geom_density(position = "stack") 0.000 0.002 0.004 0.006 2012 2013 2014 2015 2016 date density term[topic] analysing, network, r, canberra, data, social, present australia, data, ausdm, submission, workshop, mining, august data, big, mining, r, research, group, science data, mining, r, application, book, dataset, associate data, university, analytics, mining, position, research, scientist r, data, course, introduction, free, online, mining r, package, example, useful, program, slide, code r, talk, slide, series, learn, rdatamining, time Another way to plot steam graph: http://menugget.blogspot.com.au/2013/12/data-mountains-and-streams-stacked-area.html 39 / 61
- 43. Sentiment Analysis ## sentiment analysis install package sentiment140 require(devtools) install_github("sentiment140", "okugami79") # sentiment analysis library(sentiment) sentiments <- sentiment(tweets.df$text) table(sentiments$polarity) # sentiment plot sentiments$score <- 0 sentiments$score[sentiments$polarity == "positive"] <- 1 sentiments$score[sentiments$polarity == "negative"] <- -1 sentiments$date <- as.IDate(tweets.df$created) result <- aggregate(score ~ date, data = sentiments, sum) 40 / 61
- 44. Retrieve User Info and Followers ## follower analysis user <- getUser("RDataMining") user$toDataFrame() friends <- user$getFriends() # who this user follows followers <- user$getFollowers() # this user's followers followers2 <- followers[[1]]$getFollowers() # a follower's followers ## [,1] ... ## description "R and Data Mining. Group on LinkedIn: ht... ## statusesCount "583" ... ## followersCount "2376" ... ## favoritesCount "6" ... ## friendsCount "72" ... ## url "http://t.co/LwL50uRmPd" ... ## name "Yanchang Zhao" ... ## created "2011-04-04 09:15:43" ... ## protected "FALSE" ... ## verified "FALSE" ... ## screenName "RDataMining" ... ## location "Australia" ... ## lang "en" ... 41 / 61
- 45. Follower Map§ @RDataMining Followers (#: 2376) qqq q q qq q q qqq q qqqq qqqqq q q q q q qq q q qq qqq qq q qqq q qq qq qq qqq qqq qqqqq q qq q q q q q qq q qq qq q q q q qq qq q q qq qqq qq qq q qq q qqq q qq q qq q q q q q q q qqqqqqq qqq q qq q qqqqqq qq q q q q q q q qq q q q q q q q q qq q qq q q q q q q q qq qq q qq q q qq q q qq qq qq q q q q qq q qqqqq qq q qq q q q q qq qq q q qq q q q q q q q qq qq q q q q q q q qq q qqqqq q q q q q qqqqqqqqq qqq q q qqqqq q qqq q qqq q qq qqqq qqqqq qq q qqq qq q q qqq q q q q q q qqq qqq q qq q q qqq qqq qq qqqqqqqq q q qq q qq q q qq q q qqq q qq q q q q q qqq q q q q qqq q qq q q q q q qqqq qq q q qq qq q qq q q q qq qq q q q qq q q q q qq q q qqq qq qqq q q q qq qq qqq q qqq q qq qq q q q qqqqqqqqqq q qq q q qqqqqqqqqqq q q qq qqqq q q q q q qqq qq q q qqqq qqqq q q qq q qq q q q q q qqq qqq q q q q q q q qq q q q q qq q q qq q qq q qqq qq qqq q qqqq qqqqq q q q q q q q q q qqqq q qqq q qqq qqqq q qqqqq q q qq q q qqqqqqqq q qqqqq q qqq qq q q qq q qqqqq q q qq q q qq q q qqq q qqqq q q qq qqqqq qq q q q qq q qqq qq qqq qqqq q q qq qq q qqq q q qq qqq q q q q q q qqqqq q q q q q q qq q q qqq qq qq qqqq q q qq q qq qqqqq qqq q q qqqqqq qqq q qqqq qqq qqqqqq q qqq qq qqqq q qqq q qqqqq q q q qqqq q q q q q q qqqq q q q q qqq qqqq q qq q qq q q qqqq q q qq q q qqqqqqqq q q qqqq q q qqq qqqqqqqqqq q q q qqqqqqqq qqqqqqqqqqqqqqq q q q qqq q q qq q q q q q qqqqqq qq qqq qq q q q q q q q q q qqq q qqq qqqqq qq qq q qq qqq q q q qq q qqqq qq qqqq q q qqq qq q qq q qqq qq qq qq q qq qqq § Based on Jeff Leek’s twitterMap function at http://biostat.jhsph.edu/~jleek/code/twitterMap.R 42 / 61
- 46. Active Influential Followers q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q 5 10 20 50 100 0.20.51.02.05.010.020.0 #followers / #friends #Tweetsperday #AI PR Girl Marcel Molina Rahul Kapil David Smith Zac S. Christopher D. Long Data Science London Ryan Rosario Roby Learn R Statistics Blog Robert Penner Prof. Diego Kuonen DataCamp Derecho Internet Murari Bhartia Sharon Machlis Rob J Hyndman StatsBlogs Mitch Sanders Michal Illich ................................. RDataMining pavel jašek biao Daniel D. Gutierrez Data Mining M Kautzar Ichramsyah Yichuan Wang Prithwis Mukerjee Antonio Piccolboni Duccio Schiavon LearnDataAnalysis RDataMining 43 / 61
- 47. Top Retweeted Tweets ## retweet analysis ## select top retweeted tweets table(tweets.df$retweetCount) selected <- which(tweets.df$retweetCount >= 9) ## plot them dates <- strptime(tweets.df$created, format="%Y-%m-%d") plot(x=dates, y=tweets.df$retweetCount, type="l", col="grey", xlab="Date", ylab="Times retweeted") colors <- rainbow(10)[1:length(selected)] points(dates[selected], tweets.df$retweetCount[selected], pch=19, col=colors) text(dates[selected], tweets.df$retweetCount[selected], tweets.df$text[selected], col=colors, cex=.9) 44 / 61
- 48. Top Retweeted Tweets 2012 2013 2014 2015 2016 051015 Date Timesretweeted q qq q q q Free online course on Computing for Data Analysis (with R), to start on 24 Sept 2012 https://t.co/Y617n30y Lecture videos of natural language processing course at Stanford University: 18 videos, with each of over 1 hr length http://t.co/VKKdA9Tykm The R Reference Card for Data Mining now provides links to packages on CRAN. Packages for MapReduce and Hadoop added. http://t.co/RrFypol8kw Slides in 8 PDF files on Getting Data from the Web with R http://t.co/epT4Jv07WD Handling and Processing Strings in R −− an ebook in PDF format, 105 pages. http://t.co/UXnetU7k87 A Twitter dataset for text mining: @RDataMining Tweets extracted on 3 February 2016. Download it at https://t.co/lQp94IvfPf 45 / 61
- 49. Tracking Message Propagation tweets[[1]] retweeters(tweets[[1]]$id) retweets(tweets[[1]]$id) ## [1] "RDataMining: A Twitter dataset for text mining: @RData... ## [1] "197489286" "316875164" "229796464" "3316009302" ## [5] "244077734" "16900353" "2404767650" "222061895" ## [9] "11686382" "190569306" "49413866" "187048879" ## [13] "6146692" "2591996912" ## [[1]] ## [1] "bobaiKato: RT @RDataMining: A Twitter dataset for text... ## ## [[2]] ## [1] "VipulMathur: RT @RDataMining: A Twitter dataset for te... ## ## [[3]] ## [1] "tau_phoenix: RT @RDataMining: A Twitter dataset for te... The tweet potentially reached around 120,000 users. 46 / 61
- 50. 47 / 61
- 51. Contents Text Mining Concept Tasks Twitter Data Analysis with R Twitter Extracting Tweets Text Cleaning Frequent Words and Word Cloud Word Associations Clustering Topic Modelling Sentiment Analysis Follower Analysis Retweeting Analysis R Packages Wrap Up Further Readings and Online Resources 48 / 61
- 52. R Packages Twitter data extraction: twitteR Text cleaning and mining: tm Word cloud: wordcloud Topic modelling: topicmodels, lda Sentiment analysis: sentiment140 Social network analysis: igraph, sna Visualisation: wordcloud, Rgraphviz, ggplot2 49 / 61
- 53. Twitter Data Extraction – Package twitteR ¶ userTimeline, homeTimeline, mentions, retweetsOfMe: retrive various timelines getUser, lookupUsers: get information of Twitter user(s) getFollowers, getFollowerIDs: retrieve followers (or their IDs) getFriends, getFriendIDs: return a list of Twitter users (or user IDs) that a user follows retweets, retweeters: return retweets or users who retweeted a tweet searchTwitter: issue a search of Twitter getCurRateLimitInfo: retrieve current rate limit information twListToDF: convert into data.frame ¶ https://cran.r-project.org/package=twitteR 50 / 61
- 54. Text Mining – Package tm removeNumbers, removePunctuation, removeWords, removeSparseTerms, stripWhitespace: remove numbers, punctuations, words or extra whitespaces removeSparseTerms: remove sparse terms from a term-document matrix stopwords: various kinds of stopwords stemDocument, stemCompletion: stem words and complete stems TermDocumentMatrix, DocumentTermMatrix: build a term-document matrix or a document-term matrix termFreq: generate a term frequency vector findFreqTerms, findAssocs: find frequent terms or associations of terms weightBin, weightTf, weightTfIdf, weightSMART, WeightFunction: various ways to weight a term-document matrix https://cran.r-project.org/package=tm 51 / 61
- 55. Topic Modelling and Sentiment Analysis – Packages topicmodels & sentiment140 Package topicmodels ∗∗ LDA: build a Latent Dirichlet Allocation (LDA) model CTM: build a Correlated Topic Model (CTM) model terms: extract the most likely terms for each topic topics: extract the most likely topics for each document Package sentiment140 †† sentiment: sentiment analysis with the sentiment140 API, tune to Twitter text analysis ∗∗ https://cran.r-project.org/package=topicmodels †† https://github.com/okugami79/sentiment140 52 / 61
- 56. Social Network Analysis and Visualization – Package igraph ‡‡ degree, betweenness, closeness, transitivity: various centrality scores neighborhood: neighborhood of graph vertices cliques, largest.cliques, maximal.cliques, clique.number: find cliques, ie. complete subgraphs clusters, no.clusters: maximal connected components of a graph and the number of them fastgreedy.community, spinglass.community: community detection cohesive.blocks: calculate cohesive blocks induced.subgraph: create a subgraph of a graph (igraph) read.graph, write.graph: read and writ graphs from and to files of various formats ‡‡ https://cran.r-project.org/package=igraph 53 / 61
- 57. Contents Text Mining Concept Tasks Twitter Data Analysis with R Twitter Extracting Tweets Text Cleaning Frequent Words and Word Cloud Word Associations Clustering Topic Modelling Sentiment Analysis Follower Analysis Retweeting Analysis R Packages Wrap Up Further Readings and Online Resources 54 / 61
- 58. Wrap Up Transform unstructured data into structured data (i.e., term-document matrix), and then apply traditional data mining algorithms like clustering and classification Feature extraction: term frequency, TF-IDF and many others Text cleaning: lower case, removing numbers, puntuations and URLs, stop words, stemming and stem completion Stem completion may not always work as expected. Documents in languages other than English 55 / 61
- 59. Contents Text Mining Concept Tasks Twitter Data Analysis with R Twitter Extracting Tweets Text Cleaning Frequent Words and Word Cloud Word Associations Clustering Topic Modelling Sentiment Analysis Follower Analysis Retweeting Analysis R Packages Wrap Up Further Readings and Online Resources 56 / 61
- 60. Further Readings Text Mining https://en.wikipedia.org/wiki/Text_mining TF-IDF https://en.wikipedia.org/wiki/TfOT1textendashidf Topic Modelling https://en.wikipedia.org/wiki/Topic_model Sentiment Analysis https://en.wikipedia.org/wiki/Sentiment_analysis Document Summarization https://en.wikipedia.org/wiki/Automatic_summarization Natural Language Processing https://en.wikipedia.org/wiki/Natural_language_processing An introduction to text mining by Ian Witten http://www.cs.waikato.ac.nz/%7Eihw/papers/04-IHW-Textmining.pdf 57 / 61
- 61. Online Resources Book titled R and Data Mining: Examples and Case Studies [Zhao, 2012] http://www.rdatamining.com/docs/RDataMining-book.pdf R Reference Card for Data Mining http://www.rdatamining.com/docs/RDataMining-reference-card.pdf Free online courses and documents http://www.rdatamining.com/resources/ RDataMining Group on LinkedIn (27,000+ members) http://group.rdatamining.com Twitter (3,300+ followers) @RDataMining 58 / 61
- 62. The End Thanks! Email: yanchang(at)RDataMining.com Twitter: @RDataMining 59 / 61
- 63. How to Cite This Work Citation Yanchang Zhao. R and Data Mining: Examples and Case Studies. ISBN 978-0-12-396963-7, December 2012. Academic Press, Elsevier. 256 pages. URL: http://www.rdatamining.com/docs/RDataMining-book.pdf. BibTex @BOOK{Zhao2012R, title = {R and Data Mining: Examples and Case Studies}, publisher = {Academic Press, Elsevier}, year = {2012}, author = {Yanchang Zhao}, pages = {256}, month = {December}, isbn = {978-0-123-96963-7}, keywords = {R, data mining}, url = {http://www.rdatamining.com/docs/RDataMining-book.pdf} } 60 / 61
- 64. References I Zhao, Y. (2012). R and Data Mining: Examples and Case Studies, ISBN 978-0-12-396963-7. Academic Press, Elsevier. Zhao, Y. (2013). Analysing twitter data with text mining and social network analysis. In Proc. of the 11th Australasian Data Mining Conference (AusDM 2013), Canberra, Australia. 61 / 61