73rd Annual Golden Globes Twitter Analysis
Boying Gong, Jianglong Huang, Peter Sujan, Shamindra Shrotriya, Tomofumi Ogawa
February 8, 2016 Data Sources and Limitations
I Metadata - Golden Globe Nominees
I 87 people nominees and 35 movie nominees
I Manually collected/ annotated list of all nominees
I Twitter Screen Names I Gender Flag I Film/ TV Show Flag I Age of Nominee/ Release Date
I Timelines
I Typically searched for top 3200 tweets from API
I Based on most recent tweets since Dec 10 2015
I NLP processing performed e.g. removing stopwords etc Quick summary of tweet data collected Key themes of our data exploration
I Twitter Influence and Temporal Patterns - Jianglong I Social Popularity of Winners and Nominees - Peter I Sentiment Analysis - Boying I Pre-Post-During Golden Globe Analysis - Tomo When Do They Tweet? Heatmap For Tweet Density All Tweets Heatmap 7Sun 6Sat count 5Fri 1000 4Thu 3Wed 500 2Tue
Day of the week Day 1Mon 0 5 10 15 20 Hour of the day Official Accounts Tweets Heatmap 7Sun 6Sat count 1000 5Fri 750 4Thu 500
3Wed 250 2Tue
Day of the week Day 1Mon 0 5 10 15 20 Hour of the day When Do Celebrities Tweet? By Gender Heatmap For Tweet Density Female Tweets Heatmap 7Sun 6Sat count
5Fri 300
4Thu 200
3Wed 100 2Tue
Day of the week Day 1Mon 0 5 10 15 20 Hour of the day Male Tweets Heatmap 7Sun 6Sat count
5Fri 150
4Thu 100 3Wed 50 2Tue
Day of the week Day 1Mon 0 5 10 15 20 Hour of the day Tweet “POWER” VS Twitter “TENURE” Scatter Plot for Tweet Power VS Tweet Tenure
10.0
7.5
factor(factor) Old and Powerful Old and Weak 5.0 Young and Powerful Young and Weak Tweet Power Tweet
2.5
0.0 1000 2000 3000 Tweet Tenure Profile of “Weak and Old” VS “Young and Powerful” Profile Plot
winner
as.factor(group) tv Old and Weak Young and Powerful Categories of Interest
female
0.0 0.2 0.4 0.6 0.8 Proportion in Each Group Follower/Following Behavior Exhibits Distinct Groups
Nominee Following vs. Follower Counts
4000 Wolf Hall
Rachel Bloom Not shown: 3000 Lady Gaga: 55029737 , 131197 The Fencer Game of Thrones: NA , NA
Result 2000 L Active Amy Schumer Power W Number followed
AnomalisaVeepMr. Robot Mark Ruffalo 1000
Taraji P. Henson
Queen Latifah Popular
Leonardo DiCaprio 0 Steve Carell Aziz Ansari
0.0e+00 5.0e+06 1.0e+07 1.5e+07 Number of followers Movie/TV Accounts Appear Most Active
Breakdown of User Type by Style of Twitter Use Active Other
20
10
0 WINNER L Popular Power count W
20
10
0 F FILM/SHOW M F FILM/SHOW M Type Following Similarity Distribution
Distribution of Following Similarity
Very dissimilar
2000 Frequency
1000
Similarity with self = 1
Similar pairs 0
0.00 0.25 0.50 0.75 1.00 Similarity Similarity Measures Match Real-life Connections
Following Similarity: Most Similar Pairs
Steve Carell, The Big Short
Queen Latifah, Viola Davis
Adam McKay, Amy Schumer
Rachel Bloom, Amy Schumer
Rachel Bloom, Adam McKay
Taraji P. Henson, Queen Latifah
Flesh and Bone, Sarah Hay
Felicity Huffman, American Crime Pair
Taraji P. Henson, Viola Davis
Inside Out, The Good Dinosaur
Spy, Spy
Fargo, American Horror Story: Hotel
Queen Latifah, Idris Elba
Taraji P. Henson, Idris Elba
Regina King, Viola Davis
0.0 0.1 0.2 0.3 0.4 Similarity Popularity Among All Users 6= Popularity Among Peers
4
log of mention count 2
0
6 8 10 12 14 16 log of retweet count Mention Counts Grouped by Celebrities
Caitriona Balfe Amy Schumer Jeffrey Tambor Melissa McCarthy Brie Larson Lady Gaga patrick wilson Kirsten Dunst Viola Davis ryuichi sakamoto Aziz Ansari Julia Louis−Dreyfus Jamie Lee Curtis Carter Burwell Uzo Aduba Regina King Mark Ruffalo Bryan Cranston Sarah Hay Liev Schreiber Leonardo DiCaprio Win Jane Seymour Fonda Idris Elba L
Name Tomlin and Wagner W Tobias Menzies Taraji P. Henson Steve Carell Rachel Bloom Queen Latifah Mr. Bob Odenkirk Gina Rodriguez Sylvester Stallone Robin Wright Rob Lowe Rami Malek Patrick Stewart JudithLight Joanne Froggatt Felicity Huffman Ennio Morricone Emma Donoghue Daniel Pemberton Christian Slater Alan Cumming Adam McKay 0 2 4 logMentionCount Top Domains of External Links Percentage of External Links Used by Gender
Male PercentageM Female PercentageF youtube 0.91 instagram 6.01 instagram 0.55 whattheflicka 1.56 whosay 0.40 youtube 0.39 apple 0.28 facebook 0.28 facebook 0.15 twimg 0.22 usanetwork 0.14 latina 0.14 ifc 0.14 ew 0.13 hollywoodreporter 0.12 variety 0.13 twimg 0.11 theguardian 0.13 ew 0.10 yahoo 0.10 Males tweeted more Post-Globes than Pre-Globes
Male Winners vs Male Losers (mean) lag vs retweet count lag vs tweet count 15
3000
WINNER_FLAG WINNER_FLAG 10 2000 black black blue blue L L tweet count tweet retweet count retweet 1000 W 5 W
0 −20 −10 0 10 20 −20 −10 0 10 20 lag (days) lag (days) lag vs favorite count lag vs tweet length 8000 140
6000 120 WINNER_FLAG WINNER_FLAG black black
4000 blue 100 blue L L tweet length tweet favorite count favorite W W 2000 80
0 60 −20 −10 0 10 20 −20 −10 0 10 20 lag (days) lag (days) Females winners were less favorited Post-Globes!
Female Winners vs Female Losers (mean) lag vs retweet count lag vs tweet count 4000 30
3000 WINNER_FLAG WINNER_FLAG black 20 black 2000 blue blue L L tweet count tweet
retweet count retweet W W 10 1000
0 0 −20 −10 0 10 20 −20 −10 0 10 20 lag (days) lag (days) lag vs favorite count lag vs tweet length 4000 120
3000 WINNER_FLAG 100 WINNER_FLAG black black 2000 blue blue L L 80 tweet length tweet favorite count favorite W W 1000
60 0 −20 −10 0 10 20 −20 −10 0 10 20 lag (days) lag (days) Actors are using Twitter for activism post-Globes! Conclusion and Next Steps
I Nominee Analysis shows distinct behaviour patterns when summarised by gender, age, temporal components I Next Steps:
I Do the analysis for Golden Globes 2015, 2014, 2013
I Look at nominee influence via external data e.g. box office
I Download large amount of historical follower analysis
I Analysis of twitter users the nominees follow