Facebook Sentiment Analysis of MIT
Robert Slattery
November 28, 2016
MIT Background Information
MIT Facebook Page Total Likes:904,977
Mission: The mission of MIT is to advance knowledge and educate students in science, technology, and other areas of scholarship that will best serve the nation and the world in the 21st century.
Story: MIT is independent, coeducational, and privately endowed. Its five schools and one college encompass numerous academic departments, divisions and degree-granting programs, as well as interdisciplinary centers, laboratories and programs whose work cuts across traditional departmental boundaries.
About: MIT is a world-class educational institution where teaching and research — with relevance to the practical world as a guiding principle — continue to be its primary purpose.
General Information: Degrees earned at MIT include bachelor’s, master’s and doctorates. In the 2009-2010 academic year, there were 4,232 undergraduate students and 6,152 graduate students
Awards: Faculty of MIT have won numerous awards, including the Nobel Prize, Medal of Science, John Bates Clark Medal, Fields Medal, PECASE awards, MacArthur ‘genius’ grants, and many more. Many faculty are members of the AAAS, IEEE, American Academy of Arts and Sciences, and other prestigious societies. Students from MIT are regularly named Fulbright Fellows, Marshall Scholars, Rhodes Scholars, Truman Scholars and Gates Scholars.
This is the code to download data from Facebook
library(Rfacebook)## Loading required package: httr## Loading required package: rjson## Loading required package: httpuv##
## Attaching package: 'Rfacebook'## The following object is masked from 'package:methods':
##
## getGrouplibrary(scales)
library(ggplot2)
library(plyr)
library(sentimentr)
library(reshape2)
library(syuzhet)##
## Attaching package: 'syuzhet'## The following object is masked from 'package:sentimentr':
##
## get_sentences## The following object is masked from 'package:scales':
##
## rescale#setwd("/Users/robertslattery")
#load("my_oath")
#mit <- getPage(page = "mitnews",token = my_oath, n=25, reactions = T)
#write.csv(mit,"/Users/robertslattery/Facebook/mit.csv")#Code to read a previously saved group of 25 Posts
mit <- read.csv(file.path("/Users/robertslattery/Facebook/mit.csv"), stringsAsFactors = FALSE) This code was executed to pull the posts, comments and likes from each of the 25 Facebook Posts.
It was modified appropriately for each post
#mit.1 <- getPost(post=mit[1,1],n=500, token=my_oath)
#mit.posts.1 <- as.data.frame(mit.1$post)
#mit.comments.1 <- as.data.frame(mit.1$comments)
#mit.likes.1 <- as.data.frame(mit.1$likes)
#mit.likes.1$id <- mit.posts.1$id
#mit.comments.1$id <- mit.posts.1$idThese next lines combined the data from each post into data frames of Posts, Comments and Likes and then Saves them
#Combine the Comments and Likes into large data frames and save them and also save the posts
#mit.all.comments <- rbind(mit.comments.1, mit.comments.2, mit.comments.3, mit.comments.4, mit.comments.5, #mit.comments.6, mit.comments.7, mit.comments.8, mit.comments.9, mit.comments.10, mit.comments.11, mit.comments.12,
#mit.comments.13, mit.comments.14, mit.comments.15, mit.comments.16, mit.comments.17, mit.comments.18,
#mit.comments.19, mit.comments.20, mit.comments.21, mit.comments.22, mit.comments.23, #mit.comments.24,mit.comments.25)
#save(mit.all.comments, file = "/Users/robertslattery/Facebook/comments.RData")
#mit.all.likes <- rbind(mit.likes.1, mit.likes.2, mit.likes.3, mit.likes.4, mit.likes.5, mit.likes.6,
# mit.likes.7, mit.likes.8, mit.likes.9, mit.likes.10, mit.likes.11, mit.likes.12,
# mit.likes.13, mit.likes.14, mit.likes.15, mit.likes.16, mit.likes.17, mit.likes.18,
# mit.likes.19, mit.likes.20, mit.likes.21, mit.likes.22, mit.likes.23, mit.likes.24,
# mit.likes.25)
#save(mit.all.likes, file = "/Users/robertslattery/Facebook/likes.RData")
#save(mit, file = "/Users/robertslattery/Facebook/posts.RData") Here I will load the actual data files and run the analysis in Markdown.
load("/Users/robertslattery/Facebook/Slatteryposts.RData")
load("/Users/robertslattery/Facebook/Slatterycomments.RData")
load("/Users/robertslattery/Facebook/Slatterylikes.RData")First I will clean the Comments of any strange symbols such as Emojis which can not be analyzed by the Syuzhet Package used here
mit.all.comments$message <- iconv(mit.all.comments$message, sub="", 'UTF-8', 'ASCII')Now to actually do the analysis. This calculates the sentiment and emotions of each comment and then aggregates all the sentiments of each post.
#Do the Sentiment Analysis
mit.all.sentiment <- as.data.frame(get_nrc_sentiment(mit.all.comments$message))
mit.all.sentiment$id <- mit.all.comments$id
mit.all.sentiment.total <- aggregate(mit.all.sentiment[,1:10], list(mit.all.sentiment$id), sum)Now here I map the Syuzhet emotions to equivalent or similar Facebook Reactions for comparison
#Map Syuzhet Sentiments to Facebook Reactions
mit.all.sentiment.total$Angry <- mit.all.sentiment.total$disgust + mit.all.sentiment.total$anger
mit.all.sentiment.total$Wow <- mit.all.sentiment.total$surprise + mit.all.sentiment.total$anticipation
mit.all.sentiment.total$Sad <- mit.all.sentiment.total$fear + mit.all.sentiment.total$sadness
mit.all.sentiment.total$Haha <- mit.all.sentiment.total$joy
mit.all.sentiment.total$Love <- mit.all.sentiment.total$trustHowever it is important to note that these mappings are not precise and may not be entirely valid for comparison purposes but unless there is a package that can directly map word sentiment to Facebook reactions this is likely as good as it gets.
Now I sum the positive and subtract the negative Facebook Reactions to get a general Sentiment Rating, Positive or negative
#Find a total Sentiment by summing the positive and negative reactions
mit.sent <- as.data.frame(mit$love_count + mit$haha_count + mit$wow_count - mit$sad_count - mit$angry)
mit.sent$id <- mit$id
mit.sent$Sentiment <- mit.sent$`mit$love_count + mit$haha_count + mit$wow_count - mit$sad_count - mit$angry`
syuz.sent <- as.data.frame(mit.all.sentiment.total$Love + mit.all.sentiment.total$Haha + mit.all.sentiment.total$Wow - mit.all.sentiment.total$Angry - mit.all.sentiment.total$Sad)
syuz.sent$id <- mit.all.sentiment.total$Group.1
syuz.sent$Sentiment <- syuz.sent$`mit.all.sentiment.total$Love + mit.all.sentiment.total$Haha + mit.all.sentiment.total$Wow - mit.all.sentiment.total$Angry - mit.all.sentiment.total$Sad`So now I calculate the Pearson Correlation and plot the Sentiments against eachother.
#Plotting the Sentiments and Calculating Pearson Correlation
syuz.sent <- syuz.sent[order(syuz.sent$id),]
mit.sent <- mit.sent[order(mit.sent$id),]
syuz.sent$`mit.all.sentiment.total$Love + mit.all.sentiment.total$Haha + mit.all.sentiment.total$Wow - mit.all.sentiment.total$Angry - mit.all.sentiment.total$Sad` <- NULL
mit.sent$`mit$love_count + mit$haha_count + mit$wow_count - mit$sad_count - mit$angry` <- NULL
cor(mit.sent[,2], syuz.sent[,2], method="pearson")## [1] 0.5111269plot(syuz.sent$Sentiment, mit.sent$Sentiment)
The results here show a Pearson Correlation of 0.511 which is a moderate Correlation and we can see from the plot that there is definitely correlation between the Comment Sentiments and the Facebook Reactions on any given post. However there are several outliers which are heavily skewing this correlation. Posts which are much more popular seem to be more polarized and there isn’t much agreement between the moods of the posters and those who just click a reaction and make no comments. There is probably some interesting Social Psychology at work here but that can’t be investigated with the available data.
Now plot the overall Sentiment and total Like Counts for each Post
ggplot(syuz.sent, aes(id, Sentiment)) + geom_point() 
ggplot(mit, aes(id, likes_count)) + geom_point()
From these two graphs we can see that there are a number of outlying posts which have much higher like counts but the analysis of the sentiment of all the posts shows that the sentiment is generally positive and with the range from 0 to 40. So the total popularity of the post doesn’t have much influence on the overall mood of the comments.
The most common Posters
#Most Frequent Posters
sort(table(mit.all.comments$from_name))##
## AA-Mohammed Mostafa Abbie Lee
## 1 1
## Abdul Hadi Abe Ledesma
## 1 1
## Abhijeet Kumar Ahmed Sayed
## 1 1
## Aimée Gillespie Ajay Sharma
## 1 1
## Akshay K Rahul Alex Rodríguez
## 1 1
## Alfredo de la Fuente Ali Mclay
## 1 1
## Allan John Sluis Amir Ebrahimnia
## 1 1
## Anjana Sharma Arlene Rivera
## 1 1
## Aroon Boonsiri Arshad Hussain
## 1 1
## Ashu Savani Asit Parikh
## 1 1
## Asmatullah Khan Assamin Nour
## 1 1
## Astrid Rodríguez Vélez Attique UR Rehman
## 1 1
## Aviral Jain Barbara Holliday
## 1 1
## Barbara Thomas Benoit Rheault
## 1 1
## Bernhard Kratzwald Blas Navarrete Araujo
## 1 1
## Book: A Patriot's A to Z of America Brad Philipbar
## 1 1
## Brenda Monson Brian Passineau
## 1 1
## Caridad Garcia Carol Lukacs Hallowell
## 1 1
## Charlie McDonald Charlie Parker
## 1 1
## Chejinis Rivera Christhian Lima
## 1 1
## Christopher Linn Cindy Zhang
## 1 1
## Clarise Snyder CR Sergh
## 1 1
## Daniel Byers Daniel Ovalle
## 1 1
## Daniel Peña Danish Yasin
## 1 1
## Darius Baltazar Daulat Elias
## 1 1
## David Cardill David Jones
## 1 1
## David Tassoni Davon Hunter
## 1 1
## Dean Johdy Debby Edgar
## 1 1
## Denise Segovia Derek Smart
## 1 1
## Diana Fernández Sánchez Diana Towle
## 1 1
## Diego Ríos Dior Dna
## 1 1
## Donald Earner Elder Escribano
## 1 1
## Elena Colombo Eleni Kanatsouli
## 1 1
## Eleonora Fusco Elisavet Mist
## 1 1
## Elizabeth Anne Brock Emmanuel López
## 1 1
## Estrella Espinoza Fernando Mujica
## 1 1
## Fernando Yordan Flavio Chuahy
## 1 1
## Floyd Waite Francisco Matias Schaer
## 1 1
## Francisco Proskauer Valerio Frank Barcus
## 1 1
## Frank Fanyi Zhang Frank H. Nein Jr.
## 1 1
## Fredy Saie Gabby Valdes
## 1 1
## Génesis García Santini Gerd Moe-Behrens
## 1 1
## Giorgia Rosa Fernanda Zunino Glyn Louk
## 1 1
## Grace C Young Grecia Pisté
## 1 1
## Gry Folge Gura Grillo
## 1 1
## Haemi Lee Hafizuddin Aero Astro
## 1 1
## Handy heater amazon Harsh Bhatnagar
## 1 1
## Henry Wulff Ignacio Zúñiga
## 1 1
## IRfan Ahmed Iris Schilke
## 1 1
## Irshad Shaikh Ismail Odetokun
## 1 1
## Jagmeet Kaur Jason L. Berlowitz
## 1 1
## Jason Lessard Jason Trichel
## 1 1
## Javir Twell A JB Franceschi
## 1 1
## Jeffrey Schantz Jefri Tamba
## 1 1
## Jen Neely Jeremy Jay Liao
## 1 1
## Jessica Noviello Jessie-Emma Smith
## 1 1
## Jesus González Pastor JJ Prieto Rivera
## 1 1
## João Lúcio Gomes Joe Campbell
## 1 1
## John Bond John J Chiki
## 1 1
## John Lee Jonathas Kerber
## 1 1
## Joni Predi Siregar Jordan Albertus
## 1 1
## Jordan Sean Hughes Jorge Andrés Medina
## 1 1
## Jorge Silén Jose Manuel Sanchez Soldevila
## 1 1
## Joseph George Jude Morgan
## 1 1
## Julio Daz Kabir Khan
## 1 1
## Kaiwal Panchal Karam Da
## 1 1
## Kaya Ni Kenny Man
## 1 1
## Khalid Mohamed Khanh Nguyen
## 1 1
## Laura Sadhana Fricke Leo Pose
## 1 1
## Liliana Ortiz Liu Tsz Fun
## 1 1
## Liu Yinsong Loïc Crobeddu
## 1 1
## Lon DeGraw Lori Rose
## 1 1
## Louise Martins Luck Benny Toussaint
## 1 1
## Lucy Molinari Ludmi Ratnayake
## 1 1
## Lynn Otto M M Shams India
## 1 1
## Maddie Garcia Maia Weinstock
## 1 1
## Maíra Yasmin Manoj Kumar
## 1 1
## Manolo Gonzalez Manuel Rivera
## 1 1
## María Carolina Poveda Amaya Marianela Méndez García
## 1 1
## Maribel Guzmán Marilyn Andino Santana
## 1 1
## MariYan VG Mark Rekuc
## 1 1
## Martha Chow Matteo Uccio Arlotti
## 1 1
## Matthew Bradford Matthew McInerney
## 1 1
## Mauricio Torres Mayank Kumar
## 1 1
## Mayank Vishwabandhu MD Shahidul Islam
## 1 1
## Md Shamsuzzoha Mehmed Truman Kukavica
## 1 1
## Mehran Ali Meral Ekincioglu
## 1 1
## Michael Feuerstein Michael Kapteyn
## 1 1
## Michael Williams Min Jae Kim
## 1 1
## Mitko Mitkov Mohammad Nassar
## 1 1
## Muhammad Umer Myo Hein Thet
## 1 1
## Nacquia Smith Johnson Nameera Azim
## 1 1
## Natalia Siena Brody Neftali Rosado
## 1 1
## Neicy Browneyez Newton Meter
## 1 1
## Newton Pinedo Quintana Nikitas Gagas
## 1 1
## Nishit Jain Nitin Goley
## 1 1
## Norma R. Yount Olumide Johnson
## 1 1
## Oruche Goodluck Ovais Crimson
## 1 1
## Pablo Aqueveque Patrick Dobson
## 1 1
## Patrick Neuls Patrick T DiCaprio
## 1 1
## Patsy Fordjour Pedd Raam
## 1 1
## Pedro LV Pha Thai
## 1 1
## Pranav Agrawal Pratham Bhargava
## 1 1
## Rafaela PS Ram Gopalan
## 1 1
## Rashonda Stubblefield Raúl Tello
## 1 1
## Rhiana Rivas Rishabh Awasthi
## 1 1
## Rita Silva Robert Jason
## 1 1
## Robert Olson Roberto Bellas
## 1 1
## Robin Rogerfeld Rodrigo Ruz
## 1 1
## Rohit Korde Ronaldo Talagtag
## 1 1
## Ruth Telschow Ryan Faber
## 1 1
## Sabrina Madera Sabyasachi Mukhopadhyay
## 1 1
## Sahabuddin Hossain Sambhuti Anand
## 1 1
## Sanjay Biswas Santos Alejandro Camarena
## 1 1
## Sayam Kumar Das Sebasthian Santiago
## 1 1
## Seif Alaa Shaba Shams
## 1 1
## Shannon Peng Sharegist
## 1 1
## Shawn Keally Shawn Mcdowell
## 1 1
## Shubham Bhushan Silvia Patricia Rivas Poma
## 1 1
## Simone Pifferi Sinushine Kathare
## 1 1
## Soumyadeep Roy Srimanta Roy
## 1 1
## Stael Naseri Stelios Angelidis
## 1 1
## Stephanie Li Stuart Boben
## 1 1
## Surender Gupta Suzanne Cable
## 1 1
## Sylvain Proux Tarridode Marmol
## 1 1
## Teresa Nguyen Terre Hernandez
## 1 1
## Thatiana Soares Thomas Percipient Olum
## 1 1
## Thuto Douglas Maebe Tobias Mathew
## 1 1
## Tom Humphrey Toño Lozano
## 1 1
## Turkey Visit Guide Uwe Bürgin
## 1 1
## Vachagan Gevorgyan Vaggelis Goumas
## 1 1
## Vanessa Persaud Vatsal Dhaka
## 1 1
## Venkatesh Les Paul Vineet K Kashyap
## 1 1
## Vinita Seth Rampuria Vivek Chaurasia
## 1 1
## Wayne A. Seltzer William Allen
## 1 1
## William Casey Wells Yacoub Jomaa
## 1 1
## Yardley Yanira Rosado Yiovani Burbano
## 1 1
## Zaza Asatiani Zeeshan Haider
## 1 1
## Zokou Pasteur Amedée Zadi Ztevan Whyte
## 1 1
## Фёдор Иванов अदिती भारद्वाज़
## 1 1
## Aaron Puah Abdellah EL Goutbi
## 2 2
## Adroit Toriq Nur Fajar Andrea Messidoro
## 2 2
## Dan Bierwirth Dibakar Kachari DK
## 2 2
## Indi Go Jim O'Donnell
## 2 2
## Johnathan Roberts Johnny Cox
## 2 2
## Juan de Souza Julian R. Rodriguez-Bird
## 2 2
## Marie D'Ambrosio King Mark E. Wilcox
## 2 2
## Milind Ayush Tiwari Paul L. Rand
## 2 2
## Paula Zelaya Pinkee Devi
## 2 2
## Prabhakar Marshall Ramandeep Gill
## 2 2
## Randy Ramtahell Range Gowda A
## 2 2
## Rodrigue Mentz Samia Siddique
## 2 2
## Sara Gomez Arancibia Sunday Wingling
## 2 2
## Yong Li Dich Keon Jae Jeong
## 2 3
## Adri Ferent Jack B Srimof
## 4 4
## Marcy Fabian Neha Arora
## 4 4
## Sidharth Sidharth 신진실
## 4 8
## Peter Ossimini Sonu Kumar Yadav
## 12 12sort(table(mit.all.comments$from_id))##
## 1005335636262343 1009874822450335 10100470135102967 10101040731697138
## 1 1 1 1
## 10101040731702128 10101040751966518 10153920495281962 10153924618766073
## 1 1 1 1
## 10153972552556625 10154016028215978 10154028489056932 10154087582706762
## 1 1 1 1
## 10154095175207467 10154108564817023 10154128959878946 10154135631937217
## 1 1 1 1
## 10154174347342712 10154177951356089 10154184173862921 10154235332296848
## 1 1 1 1
## 10154249120209895 10154285792925186 10154352660049024 10154357882264864
## 1 1 1 1
## 10154366290215910 10154369117049335 10154465173424279 10154478355140129
## 1 1 1 1
## 10154535943500630 10154552991545664 10154647055451031 10154654096581702
## 1 1 1 1
## 10154657829563116 10154676953218632 10154680029834477 10154698834368186
## 1 1 1 1
## 10154717017934785 10154725675648794 10154734798788996 10154743180124910
## 1 1 1 1
## 10154773203642840 10154780155704549 10154833251740739 10155568217109899
## 1 1 1 1
## 10157803671890182 10157810110185284 10157849230800707 10158110089910508
## 1 1 1 1
## 1019127974879461 10202486727804772 10202547958535891 10202706117171727
## 1 1 1 1
## 10202767526865126 10202801393473675 10205793771402874 10205823810193545
## 1 1 1 1
## 10205901679979520 10205925141646607 10206460212898877 10207148719945830
## 1 1 1 1
## 10207396502691075 10207398575914725 10207692240290546 10207708890583532
## 1 1 1 1
## 10207936801474099 10207977884310089 10208088101748768 10208143049153809
## 1 1 1 1
## 10208212916666878 10208610294958695 10209034593413549 10209095669563622
## 1 1 1 1
## 10209194503918466 10209212714728011 10209319896447247 10209362566834216
## 1 1 1 1
## 10209575457997283 10209600598286273 10209628906093976 10209755735103040
## 1 1 1 1
## 10209756403738812 10209766556857161 10209806231425556 10209865900958274
## 1 1 1 1
## 10209952957153288 10209961155019928 10209986427311009 10210139840886608
## 1 1 1 1
## 10210587123304670 10210706902564706 10210889094608865 10210889502216475
## 1 1 1 1
## 10210955445026723 10210961922515245 10211018734365617 10211132843061308
## 1 1 1 1
## 10211155179695599 10211163819393660 10211177961025402 10211232093624059
## 1 1 1 1
## 10211245279638169 10211417900102597 10211444572540158 10211503556569274
## 1 1 1 1
## 10211535189921644 10211550084254620 10211572102803244 10211607550530033
## 1 1 1 1
## 10211670157895448 10211818286358289 10211878119168151 1024865174306185
## 1 1 1 1
## 1025669610876529 1035145383277815 1060442287400470 1108917305891940
## 1 1 1 1
## 1114473781993678 1118796894905982 1124642284317804 1130549213726430
## 1 1 1 1
## 1136102666485032 1143570552378031 1150665301635406 1150682355000631
## 1 1 1 1
## 1151958251548320 1154433601271556 1155783354535558 1156578711128153
## 1 1 1 1
## 1162600440491860 1163317450421501 1164052486976613 1168898813146966
## 1 1 1 1
## 1174637825957036 1178418048914674 1181855568573762 1185495348208530
## 1 1 1 1
## 1190747487659719 1196261330463257 1198527176850806 1217483488317538
## 1 1 1 1
## 1218961521530917 1219428791471589 1220560331315836 1231883463521717
## 1 1 1 1
## 1234318079940519 1236705879708761 1239902309399804 1243602995686521
## 1 1 1 1
## 1244457378926261 1248996735123721 1251109678282161 1253236658030029
## 1 1 1 1
## 126361417847573 1266144466791996 126719187811491 1274175072640050
## 1 1 1 1
## 1275863075768265 1280045872036484 1281916821861222 1282756665131568
## 1 1 1 1
## 1287336384650831 1290446041007298 1297567140293528 1297917143573765
## 1 1 1 1
## 1301491443215681 1306026776083072 1309154922469200 1324145387598604
## 1 1 1 1
## 1324273620940497 1337509462933947 1340103302707875 1343373759027714
## 1 1 1 1
## 1346286885416253 1350542344979044 1357529657598626 1358228737523453
## 1 1 1 1
## 1370359009671419 137440866738408 1378411782204133 1382005635144353
## 1 1 1 1
## 1386533524713732 1398358740182083 1429569080405510 1447237315300114
## 1 1 1 1
## 1461993480495507 1470383576322559 1473416886069232 1517667934916951
## 1 1 1 1
## 1521505754532483 1530368370517757 1535951083088225 1553798424634013
## 1 1 1 1
## 1605163079793039 1609196712716171 161045964368636 1610722615898006
## 1 1 1 1
## 1634303626869992 1677776155866869 1681013502228063 1691569051155672
## 1 1 1 1
## 1717387481911389 1734129003580037 1758192827775334 177352519395363
## 1 1 1 1
## 1785591898360625 1790714971153381 1807858449435168 1808749909407841
## 1 1 1 1
## 1810271735917136 1813033518936811 1818736085073658 1830959240483290
## 1 1 1 1
## 1832290746990049 1833397443540442 1846217645613609 1851687081784709
## 1 1 1 1
## 1855908234696422 1862320664054546 1884412548447458 2030259217200483
## 1 1 1 1
## 206766696446238 2090187334339684 209450126167467 2164289330462233
## 1 1 1 1
## 2165521507007330 217323085373489 2178245632400478 219071688536235
## 1 1 1 1
## 2201224720102709 224212801333562 224460234652481 237468896673428
## 1 1 1 1
## 245648415473995 282347362166683 294785087582982 332757750089121
## 1 1 1 1
## 333689340338316 335364996832472 338614076508056 339451113089341
## 1 1 1 1
## 346764375693054 348871608814156 349741735390716 351580048526201
## 1 1 1 1
## 351810308509309 359344801083361 366224583726502 370400063302536
## 1 1 1 1
## 378192882523040 391850634480331 548015678741067 572450926298749
## 1 1 1 1
## 577174752468079 582727348588581 591182031065200 593907067459797
## 1 1 1 1
## 601606573374036 629533903891376 651714261701330 658798090961810
## 1 1 1 1
## 666494306854036 676108895904333 693098840846841 695784570587451
## 1 1 1 1
## 706118832887728 706621429488093 714961098534565 720102724820184
## 1 1 1 1
## 723804454440159 732890474486 756934541121644 758771454262392
## 1 1 1 1
## 791845187622831 885020874932245 900976603365930 918059224990847
## 1 1 1 1
## 924784767656323 933903250076352 941005456044081 965782590232882
## 1 1 1 1
## 10153775283196706 10154039988416186 10156235380903644 10207873511579965
## 2 2 2 2
## 10210238917146463 10211047144672162 10211079684578726 10211302633506687
## 2 2 2 2
## 10211628985986662 1083220691790674 1085548221564825 1147986465308880
## 2 2 2 2
## 118762135276589 1243601109019989 1292616897476929 1335096529834596
## 2 2 2 2
## 1344500692249445 1359565654056418 1492236807466949 1625686177727586
## 2 2 2 2
## 1665419620416826 1796832877232506 1811554795759290 312626092464170
## 2 2 2 2
## 340732296293093 538872049644425 544533385737231 1263013277103876
## 2 2 2 3
## 10208462824674279 137193320096344 1603945726576809 1763572040570768
## 4 4 4 4
## 199379500518737 114175139070126 1717784775205289 953351858103164
## 4 8 12 12names(which(table(mit.all.comments$from_name) == max(table(mit.all.comments$from_name))))## [1] "Peter Ossimini" "Sonu Kumar Yadav"max(table(mit.all.comments$from_name))## [1] 12This code shows that the Most prolific posters are Sonu Kumar Yadav and Peter Ossimini with 12 comments each.
This next block of code combines the comments of the top 10 commenters into a large Data Frame
common.posters <- as.data.frame(mit.all.comments[mit.all.comments == "114175139070126" | mit.all.comments == "953351858103164" |
mit.all.comments == "1603945726576809" | mit.all.comments == "1717784775205289" |
mit.all.comments == "137193320096344" | mit.all.comments == "10208462824674279" |
mit.all.comments == "1263013277103876" | mit.all.comments == "544533385737231" |
mit.all.comments == "1763572040570768" | mit.all.comments == "199379500518737",])Printing 30 of the comments yields some insight.
print(common.posters[1:30,2:3])## from_name
## 5 Sidharth Sidharth
## 8 Sidharth Sidharth
## 16 Sidharth Sidharth
## 19 Sidharth Sidharth
## 29 Sonu Kumar Yadav
## 45 Sonu Kumar Yadav
## 48 Peter Ossimini
## 68 Sonu Kumar Yadav
## 79 Peter Ossimini
## 84 Sonu Kumar Yadav
## 103 Marcy Fabian
## 146 Jack B Srimof
## 147 Sonu Kumar Yadav
## 148 Peter Ossimini
## 149 Peter Ossimini
## 155 신진실
## 156 신진실
## 160 Peter Ossimini
## 166 Sonu Kumar Yadav
## 167 Keon Jae Jeong
## 170 Peter Ossimini
## 171 Neha Arora
## 172 Neha Arora
## 186 Marcy Fabian
## 191 Sonu Kumar Yadav
## 192 Keon Jae Jeong
## 194 신진실
## 195 Jack B Srimof
## 218 Peter Ossimini
## 221 Sonu Kumar Yadav
## message
## 5 My first interview with Keith about Nobel prizes\nhttps://m.facebook.com/story.php?story_fbid=10154154560767309&id=651752308
## 8 My first interview with Keith about Nobel prizes\nhttps://m.facebook.com/story.php?story_fbid=10154154560767309&id=651752308
## 16 My first interview with Keith about Nobel prizes\nhttps://m.facebook.com/story.php?story_fbid=10154154560767309&id=651752308
## 19 My first interview with Keith about Nobel prizes\nhttps://m.facebook.com/story.php?story_fbid=10154154560767309&id=651752308
## 29 That's great
## 45 Nice
## 48 How many christian organizations exist in the middle east in order to you allow this? https://www.facebook.com/mitmsa/?fref=ts
## 68 Very nice
## 79 How many christian organizations exist in the middle east in order to you allow this? https://www.facebook.com/mitmsa/?fref=ts
## 84 Amazing picture of moon a great day of world history..
## 103 Beautiful!!
## 146 above , a joke .
## 147 Nice
## 148 How many christian organizations exist in the middle east in order to you allow this? https://www.facebook.com/mitmsa/?fref=ts
## 149 How many christian organizations exist in the middle east in order to you allow this? https://www.facebook.com/mitmsa/?fref=ts
## 155
## 156 ?
## 160 How many christian organizations exist in the middle east in order to you allow this? https://www.facebook.com/mitmsa/?fref=ts
## 166
## 167 thank you!
## 170 How many christian organizations exist in the middle east in order to you allow this? https://www.facebook.com/mitmsa/?fref=ts
## 171 http://kalmanserve.com/lan1.aspx?r=regular\thttp://kalmanserve.com/images/set1/125.gif
## 172 http://kalmanserve.com/lan1.aspx?r=regular\thttp://kalmanserve.com/images/set1/125.gif
## 186 Congratulations!!
## 191 Lovely
## 192 thank you!
## 194 ?
## 195 Puerto Rico is not part of US .
## 218 How many christian organizations exist in the middle east in order to you allow this? https://www.facebook.com/mitmsa/?fref=ts
## 221 LovelyIt appears that many of the comments of the top posters are spam type messages, advertising a website or containing political/religous or other irrelvant rantings.
Let’s get the Sentiment and number of posts for each of the top commenters
#Sentiment of common posters
common.posters.sentiment <- as.data.frame(get_nrc_sentiment(common.posters$message))
common.posters.sentiment$from_name <- common.posters$from_name
common.posters.total <- aggregate(common.posters.sentiment[,1:10], list(common.posters.sentiment$from_name), sum)
common.posters.total2 <- colSums(common.posters.sentiment[,1:10])
num.comments <- as.data.frame(sort(table(common.posters$from_name)))Plotting the top commenter sentiment and number of posts
#Plot habits and sentiment of top 10 posters
ggplot(num.comments, aes(Var1, Freq)) + geom_point() + labs(x="Name of Poster", y="Frequency of Posts") + ylim(0, 15)
barplot(common.posters.total2)
So we can see that the top commenters post much more than the other commenters and that generally the sentiment of these comments are quite positive.
Brand Recommendations
MIT has one of the most influential brands in the World and its scope goes well beyond that of a normal University. The results of this analysis shows that it enjoys relatively stable popularity as shown by the plots of Likes and occasionally has “Viral” posts which are many times more popular than it’s usual posts. The analysis also shows that there are occasional posts which have very negative reactions as shown by the Syuzhet sentiment and Facebook Reactions. It is likely that these posts are cover topics that are highly politicized and evoke strong reactions from individuals on Facebook. Investigations of the top commenters show that there is a problem with spam posts. Both the top posters post the same comment many times and seem to be advertising something or contain Political ramblings. Though in general the other top commenters are strongly positive in their comments. MIT as a brand is simply impeccable and I have no recommendations in that regard. However its Facebook page could be more carefully curated to ensure that its atmosphere is one of Scholarship and highlights its Research. They should be careful to avoid posting topics that are too political if they wish to avoid backlash, though many topics such as Climate Change which are highly politicized are important areas of research at a school like MIT, and investigate the nature of the posts which are much more popular than others in order to get a sense of what are the most popular topics. They should also curate the comment sections since there generally appears to be a problem with people posting irrelevant spam in many different posts. This degrades from the quality of the discussions in the comments and should be monitored. MIT’s name is as strong as it has ever been and aside from tweaking their Social Media presences there is little that can be done to improve its standing.