Text Analytics & Semantic Network
Rhowena Vespa
2022-10-27
library(readr)## Warning: package 'readr' was built under R version 4.2.1library(dplyr)## Warning: package 'dplyr' was built under R version 4.2.1##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(quanteda)## Warning: package 'quanteda' was built under R version 4.2.1## Warning in .recacheSubclasses(def@className, def, env): undefined subclass
## "unpackedMatrix" of class "mMatrix"; definition not updated## Warning in .recacheSubclasses(def@className, def, env): undefined subclass
## "unpackedMatrix" of class "replValueSp"; definition not updated## Package version: 3.2.3
## Unicode version: 13.0
## ICU version: 69.1## Parallel computing: 4 of 4 threads used.## See https://quanteda.io for tutorials and examples.library(quanteda.textstats)## Warning: package 'quanteda.textstats' was built under R version 4.2.1## Warning in .recacheSubclasses(def@className, def, env): undefined subclass
## "unpackedMatrix" of class "mMatrix"; definition not updated## Warning in .recacheSubclasses(def@className, def, env): undefined subclass
## "unpackedMatrix" of class "replValueSp"; definition not updatedlibrary(quanteda.textplots)## Warning: package 'quanteda.textplots' was built under R version 4.2.1library(seededlda)## Warning: package 'seededlda' was built under R version 4.2.1## Loading required package: proxyC## Warning: package 'proxyC' was built under R version 4.2.1##
## Attaching package: 'proxyC'## The following object is masked from 'package:stats':
##
## dist##
## Attaching package: 'seededlda'## The following object is masked from 'package:stats':
##
## termslibrary(ggplot2)## Warning: package 'ggplot2' was built under R version 4.2.1library(DT)## Warning: package 'DT' was built under R version 4.2.1#KWIC (keywords-in-context)
Umass <- read_csv("https://curiositybits.cc/files/umass-instagram.csv")## Rows: 1421 Columns: 21
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (10): Account, User Name, Followers at Posting, Post Created, Type, URL...
## dbl (4): Likes, Comments, Views, Overperforming Score (weighted — Likes ...
## lgl (4): Like and View Counts Disabled, Title, Sponsor Id, Sponsor Name
## date (1): Post Created Date
## time (1): Post Created Time
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.corpus_Umass <- corpus(Umass,text_field = "Description") ## Warning: NA is replaced by empty stringdata_tokens <- tokens(corpus_Umass)kw <- kwic(data_tokens, pattern = "student*")
head(kw, 10) # SHOW 10 RECORDS OF DOCUMENTS THAT CONTAIN WORDS BEGINNING WITH 'student'## Keyword-in-context with 10 matches.
## [text3, 94] giving, financial resources and | student |
## [text4, 21] , September 13 in the | Student |
## [text8, 19] of care to UMass Amherst | students |
## [text9, 5] As part of New | Student |
## [text9, 10] Student Orientation and Transitions, | students |
## [text10, 23] Thank you to all the | students |
## [text10, 36] and to the amazing New | Student |
## [text11, 91] : 00 p.m. - | Students |
## [text12, 4] The UMass Amherst | Student |
## [text12, 16] for the freshmen Senator and | Student |
##
## selectivity. Read more with
## Union Ballroom. UMass Amherst
## to help cope with stress
## Orientation and Transitions, students
## were invited to Athletics 101
## who joined us on Sunday
## Orientation and Transitions team for
## - 4: 15 to
## Government Association is now accepting
## Trustee elections until Tuesday,DFM based on the UMass Instagram account’s posts.
insta_dfm <- tokens(corpus_Umass, remove_punct = TRUE, remove_numbers = TRUE, remove_symbols = TRUE, remove_url=TRUE) %>%
tokens_remove(stopwords("en")) %>%
dfm()
topfeatures(insta_dfm) ## this shows top words by basic word count## campus umass students amherst link
## 474 462 234 210 169
## day bio #sceneatumass today student
## 169 163 157 154 140Top words ranked by TF-IDF
insta_dfm <- dfm_tfidf(insta_dfm) #create a new DFM by tf-idf scores
topfeatures(insta_dfm) ## this shows top words by tf-idf## campus umass students amherst day health link bio
## 274.7885 271.0768 212.8361 183.0277 167.5172 158.4141 157.1493 153.7219
## student today
## 152.8655 151.2792Semantic network- DFM that contains hashtags and top 30 hashtags in the DFM
tag_dfm <- dfm_select(insta_dfm, pattern = "#*")
toptag <- names(topfeatures(tag_dfm, 30)) # get top 30 results
head(toptag)## [1] "#sceneatumass" "#umassamherst" "#flashbackfriday" "#college"
## [5] "#university" "#umassgives"#Construct feature-occurrence matrix of hashtags
tag_fcm <- fcm(tag_dfm)
head(tag_fcm)## Feature co-occurrence matrix of: 6 by 333 features.
## features
## features #swamyselfie #throwback #welcometoumass #flashbackfriday
## #swamyselfie 3.393128 0 0 0
## #throwback 0 3.393128 0 0
## #welcometoumass 0 0 3.393128 0
## #flashbackfriday 0 0 0 12.63902
## #tbt 0 0 0 0
## #throwbackthursday 0 0 0 0
## features
## features #tbt #throwbackthursday #umasshistory
## #swamyselfie 0 0 0
## #throwback 0 0 0
## #welcometoumass 0 0 0
## #flashbackfriday 0 0 5.517717
## #tbt 11.24306 4.945201 4.945201
## #throwbackthursday 0 11.854791 33.829247
## features
## features #umasssummerphoto22 #indigenousday #indigenouspeoplesday
## #swamyselfie 0 0 0
## #throwback 0 0 0
## #welcometoumass 0 0 0
## #flashbackfriday 0 0 0
## #tbt 0 0 0
## #throwbackthursday 0 0 0
## [ reached max_nfeat ... 323 more features ]#Visualize the semantic network based on hashtag co-occurrence
topgat_fcm <- fcm_select(tag_fcm, pattern = toptag)
textplot_network(topgat_fcm, min_freq = 0.1, edge_alpha = 0.5, edge_size = 1, edge_color = "orange")
LDA model based on the Instagram posts. Set the number of topics to 25. Show top 10 terms from each topic
library(seededlda)
tmod_lda <- textmodel_lda(insta_dfm, k = 25) # Set k=10 and look for 25 topics in the dfm.
terms(tmod_lda, 10) #Extract the 10 most important terms for each topic from the topic model ## topic1 topic2 topic3 topic4 topic5
## [1,] "make" "orchard" "black" "farm" "counseling"
## [2,] "#umassgives" "sunsets" "presence" "just" "health"
## [3,] "gift" "fall" "check" "gear" "center"
## [4,] "favorite" "warm" "look" "aj" "community"
## [5,] "giving" "huge" "research" "hadley" "help"
## [6,] "together" "hill" "website" "cute" "spring"
## [7,] "changed" "spring" "excited" "seen" "semester"
## [8,] "give" "colors" "wanted" "days" "support"
## [9,] "best" "engagement" "radical" "gone" "learn"
## [10,] "thank" "met" "videos" "last" "psychological"
## topic6 topic7 topic8 topic9 topic10
## [1,] "chancellor" "a.m" "nursing" "sexual" "really"
## [2,] "inclusion" "graduating" "@umass_scua" "current" "people"
## [3,] "emails" "senior" "#flashbackfriday" "assault" "major"
## [4,] "racism" "new" "college" "known" "class"
## [5,] "racist" "united" "marieb" "challenge" "like"
## [6,] "black" "ice" "student" "part" "get"
## [7,] "community" "washington" "elaine" "initiative" "able"
## [8,] "subbaswamy" "attacks" "innovation" "review" "thought"
## [9,] "investigation" "food" "#throwbackthursday" "last" "think"
## [10,] "equity" "sean" "#umasshistory" "arts" "got"
## topic11 topic12 topic13 topic14 topic15
## [1,] "summer" "rising" "energy" "best" "happy"
## [2,] "us" "researcher" "worked" "universities" "closed"
## [3,] "photos" "lab" "assistant" "magazine" "offices"
## [4,] "photo" "gateway" "project" "know" "holiday"
## [5,] "win" "rising-researchers" "black" "student" "monday"
## [6,] "tag" "helping" "program" "u.s" "observance"
## [7,] "contest" "read" "art" "cap" "classes"
## [8,] "card" "chemistry" "chemical" "hear" "garden"
## [9,] "now" "umass.edu" "research" "year" "enjoy"
## [10,] "chance" "use" "served" "graduation" "dining"
## topic16 topic17 topic18 topic19 topic20
## [1,] "juneteenth" "vibes" "contact" "view" "dogs"
## [2,] "justice" "finals" "time" "pond" "student"
## [3,] "june" "hot" "can" "image" "makes"
## [4,] "society" "sports" "tracing" "library" "like"
## [5,] "sunwheel" "way" "moodle" "description" "thanks"
## [6,] "racism" "pride" "household" "@douhan_photography" "check"
## [7,] "americans" "making" "virtually" "@shtewpac" "one"
## [8,] "rain" "happy" "us" "du" "last"
## [9,] "police" "southwest" "campus" "old" "union"
## [10,] "say" "family" "outside" "bois" "ballroom"
## topic21 topic22 topic23 topic24
## [1,] "testing" "game" "commencement" "band"
## [2,] "center" "#flagship" "back" "minuteman"
## [3,] "covid-19" "#flashbackfriday" "welcome" "international"
## [4,] "health" "set" "class" "parade"
## [5,] "public" "hockey" "soon" "marching"
## [6,] "mask" "@umassband" "college" "s"
## [7,] "students" "@umasshockey" "weekend" "rose"
## [8,] "remain" "music" "honors" "@sam_minuteman"
## [9,] "in-person" "love" "graduates" "sam"
## [10,] "open" "home" "commonwealth" "#roseparade"
## topic25
## [1,] "#umassamherst"
## [2,] "#college"
## [3,] "#university"
## [4,] "#massachusetts"
## [5,] "#campus"
## [6,] "contest"
## [7,] "rules"
## [8,] "#snow"
## [9,] "photo"
## [10,] "#umass"