SlideShare a Scribd company logo
1 of 16
Download to read offline
@qutdmrc
Detecting Twitter Bots That Share SoundCloud Tracks
Axel Bruns, Brenda Moon, Felix Victor Münch, Patrik Wikström
Digital Media Research Centre, Queensland University of Technology
Stefan Stieglitz, Florian Brachten, Björn Ross
Research Group Professional Communication in Electronic Media / Social Media, University of Duisburg-Essen
SM&S 2018, Copenhagen
@snurb_dot_info
@qutdmrc
Bots
● Bot types:
● Benign and legitimate
● Nefarious and subversive
● Political, commercial, …
● Exploiting affordances of social media platforms
● Main focus on Facebook and Twitter
● Bots as an industry:
● Fake followers
● Like, comment, share, retweet campaigns
● Trending topic generation
● URL promotion campaigns
@qutdmrc
Bots on / for SoundCloud
@qutdmrc
Bots on / for SoundCloud
● Native SoundCloud bots
● Plays, likes, comments, followers, etc.
● Addressed in Ross et al. (2018)
● SoundCloud-promoting Twitter bots:
● Sharing SoundCloud URLs, boosting SoundCloud metrics
● Can we detect such Twitter bots by their URL sharing patterns?
@qutdmrc
Datasets
● Data:
● Tweets that share soundcloud.com URLs (after unshortening)
● Selection of tweets posted during March / April 2017
● 11.5m tweets posted by 2m Twitter accounts, sharing 2.1m tracks
● URLs standardised to soundcloud.com/[user name]/[track name]
● Top Tracks dataset:
● Tracks shared ≥ 1,000 times: 233 tracks, 914,131 tweets
● Top Accounts dataset:
● Accounts sharing tracks in ≥ 500 tweets: 649 accounts, 1m tweets
Preliminary Metrics
@qutdmrc
Metrics 1: Account Diversity
● Account Diversity:
𝐴𝑐𝑐𝑜𝑢𝑛𝑡 𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦 =
# 𝑈𝑛𝑖𝑞𝑢𝑒 𝑇𝑤𝑖𝑡𝑡𝑒𝑟 𝐴𝑐𝑐𝑜𝑢𝑛𝑡𝑠
# 𝑇𝑤𝑒𝑒𝑡𝑠
● Calculated per track:
● ≈0: persistent promotion of track by small number of accounts
● ≈1: many accounts sharing the same track
@qutdmrc
Metrics 2: Tweet Originality
● Tweet Originality:
𝑇𝑤𝑒𝑒𝑡 𝑂𝑟𝑖𝑔𝑖𝑛𝑎𝑙𝑖𝑡𝑦 =
# 𝑁𝑜𝑛−𝑅𝑒𝑡𝑤𝑒𝑒𝑡𝑠 − # 𝑅𝑒𝑡𝑤𝑒𝑒𝑡𝑠
# 𝑇𝑤𝑒𝑒𝑡𝑠
● Calculated per track:
● –1: all tweets sharing the track are retweets
● +1: all tweets sharing the track are original tweets / @mentions
● Calculated per account:
● –1: all SoundCloud URL tweets by account are retweets
● +1: all SoundCloud URL tweets are original tweets / @mentions
@qutdmrc
Metrics 3: Tweet Text Diversity
● Tweet Text Diversity:
𝑇𝑤𝑒𝑒𝑡 𝑇𝑒𝑥𝑡 𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦 =
# 𝑈𝑛𝑖𝑞𝑢𝑒 𝑇𝑤𝑒𝑒𝑡 𝑇𝑒𝑥𝑡𝑠
# 𝑇𝑤𝑒𝑒𝑡𝑠
● Calculated per track:
● ≈0: all tweets linking to SoundCloud track use identical text
● ≈1: all tweets linking to SoundCloud track use different texts
● Calculated per account:
● ≈0: all SoundCloud tweets posted by account use identical text
● ≈1: all SoundCloud tweets posted by account use different texts
@qutdmrc
Metrics 4: SoundCloud URL Diversity
● SoundCloud URL Diversity:
𝑆𝑜𝑢𝑛𝑑𝐶𝑙𝑜𝑢𝑑 𝑈𝑅𝐿 𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦 =
# 𝑇𝑟𝑎𝑐𝑘𝑠 𝑆ℎ𝑎𝑟𝑒𝑑
# 𝑇𝑤𝑒𝑒𝑡𝑠
● Calculated per account:
● ≈0: all SoundCloud tweets by account share the same URL
● ≈1: all SoundCloud tweets by account share a different URL
Exploratory Analysis
@qutdmrc
Analysis
● Preliminary analysis:
● Three per-track metrics:
● Account Diversity
● Tweet Originality
● Tweet Text Diversity
● Three per-account metrics:
● Tweet Originality
● Tweet Text Diversity
● SoundCloud URL Diversity
● Pairwise exploration of correlation patterns amongst metrics
● k-means clustering of tracks / accounts across each three metrics
● Abductive interpretation of patterns towards hypothesis formation
@qutdmrc
Metrics per Track
retweeting
many
retweets by
few
accounts
many original
tweets by
few accounts
many original
tweets by few
accounts, using
many different
tweet texts
@qutdmrc
Metrics per Accountretweets
originaltweets
many original
tracks, many
original texts
Few original
tracks, many
original texts
repeated sharing
but few retweets
@qutdmrc
Observations
● Proposed metrics:
● Promising distinctions between tracks and accounts
● Need to further validate patterns through qualitative assessment
● Need to correlate between per-track and per-account metrics
● Need to explore patterns around SoundCloud users, not just tracks
● Further application:
● Test metrics on larger datasets
● Develop and test bot detection approaches on SoundCloud itself,
and compare findings with present study (Ross et al. 2018)
@qutdmrc
SM&S 2018, Copenhagen
@snurb_dot_info
Acknowledgments
This research is supported by the UA-DAAD project Emergent
Music Engagement Practices via SoundCloud and the ARC LIEF
project TrISMA: Tracking Infrastructure for Social Media Analysis.
Reference
B. Ross, F. Brachten, S. Stieglitz, P. Wikström, B. Moon, F.
Münch, and A. Bruns. 2018. Social Bots in a Commercial
Context – A Case Study on SoundCloud. Paper presented at
European Conference on Information Systems (ECIS),
Portsmouth, 23-28 July 2018.

More Related Content

Similar to Detecting Twitter Bots That Promote SoundCloud Tracks

Social media analytics as a service: tools from GATE
Social media analytics as a service: tools from GATESocial media analytics as a service: tools from GATE
Social media analytics as a service: tools from GATEDiana Maynard
 
Twitter Sentiment Prediction.pptx
Twitter Sentiment Prediction.pptxTwitter Sentiment Prediction.pptx
Twitter Sentiment Prediction.pptxKrishnesh Pujari
 
What do you do with 280 million tweets from the 2016 U.S. election?
What do you do with 280 million tweets from the 2016 U.S. election?What do you do with 280 million tweets from the 2016 U.S. election?
What do you do with 280 million tweets from the 2016 U.S. election?Justin Littman
 
How to grow your business using Social Media
How to grow your business using Social Media How to grow your business using Social Media
How to grow your business using Social Media Tomo360, LLC
 
Social Media - MA Journalism - Week Four
Social Media - MA Journalism - Week FourSocial Media - MA Journalism - Week Four
Social Media - MA Journalism - Week Fourpaulwould
 
Social Media Strategy: Kristen Botica's Personal Brand
Social Media Strategy: Kristen Botica's Personal BrandSocial Media Strategy: Kristen Botica's Personal Brand
Social Media Strategy: Kristen Botica's Personal BrandKristen Botica
 
New Methodologies for Capturing and Working with Publicly Available Twitter Data
New Methodologies for Capturing and Working with Publicly Available Twitter DataNew Methodologies for Capturing and Working with Publicly Available Twitter Data
New Methodologies for Capturing and Working with Publicly Available Twitter DataAxel Bruns
 
Social Media & Metrics (Digital Marketing Today)
Social Media & Metrics (Digital Marketing Today)Social Media & Metrics (Digital Marketing Today)
Social Media & Metrics (Digital Marketing Today)Julian Gamboa
 
Predict Interestingness of An Article Using Twitter
Predict Interestingness of An Article Using TwitterPredict Interestingness of An Article Using Twitter
Predict Interestingness of An Article Using TwitterYash Girdhar
 
Final Year PPT on Twitter App
Final Year PPT on Twitter AppFinal Year PPT on Twitter App
Final Year PPT on Twitter Appscorpionking257
 
Twitter: A Hands-On Learning Session for Researcher
Twitter: A Hands-On Learning Session for ResearcherTwitter: A Hands-On Learning Session for Researcher
Twitter: A Hands-On Learning Session for ResearcherKMb Unit, York University
 
CTM Social Media Strategy
CTM Social Media StrategyCTM Social Media Strategy
CTM Social Media StrategyElise Schimke
 
Honest marketing
Honest marketingHonest marketing
Honest marketingbobsumnerjr
 
Twitter Sentiment Analysis.pdf
Twitter Sentiment Analysis.pdfTwitter Sentiment Analysis.pdf
Twitter Sentiment Analysis.pdfRachanasamal3
 
The Social Media Cheat Sheet - The Daily Social Media Workouts v3
The Social Media Cheat Sheet - The Daily Social Media Workouts v3The Social Media Cheat Sheet - The Daily Social Media Workouts v3
The Social Media Cheat Sheet - The Daily Social Media Workouts v3Lightspan Digital
 
How to: Advanced Social Media Techniques
How to: Advanced Social Media TechniquesHow to: Advanced Social Media Techniques
How to: Advanced Social Media TechniquesMandy Jenkins
 
A Multi-Institutional Approach to ‘Big Social Data’: The TrISMA Project
A Multi-Institutional Approach to ‘Big Social Data’: The TrISMA ProjectA Multi-Institutional Approach to ‘Big Social Data’: The TrISMA Project
A Multi-Institutional Approach to ‘Big Social Data’: The TrISMA ProjectAxel Bruns
 
Using Twitter for Business Engagement
Using Twitter for Business EngagementUsing Twitter for Business Engagement
Using Twitter for Business EngagementSue Beckingham
 

Similar to Detecting Twitter Bots That Promote SoundCloud Tracks (20)

Social media analytics as a service: tools from GATE
Social media analytics as a service: tools from GATESocial media analytics as a service: tools from GATE
Social media analytics as a service: tools from GATE
 
Twitter Sentiment Prediction.pptx
Twitter Sentiment Prediction.pptxTwitter Sentiment Prediction.pptx
Twitter Sentiment Prediction.pptx
 
What do you do with 280 million tweets from the 2016 U.S. election?
What do you do with 280 million tweets from the 2016 U.S. election?What do you do with 280 million tweets from the 2016 U.S. election?
What do you do with 280 million tweets from the 2016 U.S. election?
 
How to grow your business using Social Media
How to grow your business using Social Media How to grow your business using Social Media
How to grow your business using Social Media
 
Uaa social media strategy
Uaa social media strategyUaa social media strategy
Uaa social media strategy
 
Social Media - MA Journalism - Week Four
Social Media - MA Journalism - Week FourSocial Media - MA Journalism - Week Four
Social Media - MA Journalism - Week Four
 
Social Media Strategy: Kristen Botica's Personal Brand
Social Media Strategy: Kristen Botica's Personal BrandSocial Media Strategy: Kristen Botica's Personal Brand
Social Media Strategy: Kristen Botica's Personal Brand
 
New Methodologies for Capturing and Working with Publicly Available Twitter Data
New Methodologies for Capturing and Working with Publicly Available Twitter DataNew Methodologies for Capturing and Working with Publicly Available Twitter Data
New Methodologies for Capturing and Working with Publicly Available Twitter Data
 
Social Media & Metrics (Digital Marketing Today)
Social Media & Metrics (Digital Marketing Today)Social Media & Metrics (Digital Marketing Today)
Social Media & Metrics (Digital Marketing Today)
 
Predict Interestingness of An Article Using Twitter
Predict Interestingness of An Article Using TwitterPredict Interestingness of An Article Using Twitter
Predict Interestingness of An Article Using Twitter
 
Final Year PPT on Twitter App
Final Year PPT on Twitter AppFinal Year PPT on Twitter App
Final Year PPT on Twitter App
 
Twitter: A Hands-On Learning Session for Researcher
Twitter: A Hands-On Learning Session for ResearcherTwitter: A Hands-On Learning Session for Researcher
Twitter: A Hands-On Learning Session for Researcher
 
CTM Social Media Strategy
CTM Social Media StrategyCTM Social Media Strategy
CTM Social Media Strategy
 
Honest marketing
Honest marketingHonest marketing
Honest marketing
 
Twitter Sentiment Analysis.pdf
Twitter Sentiment Analysis.pdfTwitter Sentiment Analysis.pdf
Twitter Sentiment Analysis.pdf
 
The Social Media Cheat Sheet - The Daily Social Media Workouts v3
The Social Media Cheat Sheet - The Daily Social Media Workouts v3The Social Media Cheat Sheet - The Daily Social Media Workouts v3
The Social Media Cheat Sheet - The Daily Social Media Workouts v3
 
How to: Advanced Social Media Techniques
How to: Advanced Social Media TechniquesHow to: Advanced Social Media Techniques
How to: Advanced Social Media Techniques
 
Success on Twitter
Success on TwitterSuccess on Twitter
Success on Twitter
 
A Multi-Institutional Approach to ‘Big Social Data’: The TrISMA Project
A Multi-Institutional Approach to ‘Big Social Data’: The TrISMA ProjectA Multi-Institutional Approach to ‘Big Social Data’: The TrISMA Project
A Multi-Institutional Approach to ‘Big Social Data’: The TrISMA Project
 
Using Twitter for Business Engagement
Using Twitter for Business EngagementUsing Twitter for Business Engagement
Using Twitter for Business Engagement
 

More from Axel Bruns

Identifying the Symptoms of Destructive Polarisation
Identifying the Symptoms of Destructive PolarisationIdentifying the Symptoms of Destructive Polarisation
Identifying the Symptoms of Destructive PolarisationAxel Bruns
 
Voices on the Voice Referendum: A Computational Analysis of News and Audience...
Voices on the Voice Referendum: A Computational Analysis of News and Audience...Voices on the Voice Referendum: A Computational Analysis of News and Audience...
Voices on the Voice Referendum: A Computational Analysis of News and Audience...Axel Bruns
 
What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...
What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...
What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...Axel Bruns
 
What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...
What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...
What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...Axel Bruns
 
Types of Polarisation and Their Operationalisation in Digital and Social Medi...
Types of Polarisation and Their Operationalisation in Digital and Social Medi...Types of Polarisation and Their Operationalisation in Digital and Social Medi...
Types of Polarisation and Their Operationalisation in Digital and Social Medi...Axel Bruns
 
News Sharing and Partisanship: Tracking News Outlet Repertoires on Twitter ov...
News Sharing and Partisanship: Tracking News Outlet Repertoires on Twitter ov...News Sharing and Partisanship: Tracking News Outlet Repertoires on Twitter ov...
News Sharing and Partisanship: Tracking News Outlet Repertoires on Twitter ov...Axel Bruns
 
Determining the Drivers and Dynamics of Partisanship and Polarisation in Onli...
Determining the Drivers and Dynamics of Partisanship and Polarisation in Onli...Determining the Drivers and Dynamics of Partisanship and Polarisation in Onli...
Determining the Drivers and Dynamics of Partisanship and Polarisation in Onli...Axel Bruns
 
Towards a New Empiricism: Polarisation across Four Dimensions
Towards a New Empiricism: Polarisation across Four DimensionsTowards a New Empiricism: Polarisation across Four Dimensions
Towards a New Empiricism: Polarisation across Four DimensionsAxel Bruns
 
The Anatomy of Virality: How COVID-19 Conspiracy Theories Spread across Socia...
The Anatomy of Virality: How COVID-19 Conspiracy Theories Spread across Socia...The Anatomy of Virality: How COVID-19 Conspiracy Theories Spread across Socia...
The Anatomy of Virality: How COVID-19 Conspiracy Theories Spread across Socia...Axel Bruns
 
A Platform Policy Implementation Audit of Actions against Russia’s State-Cont...
A Platform Policy Implementation Audit of Actions against Russia’s State-Cont...A Platform Policy Implementation Audit of Actions against Russia’s State-Cont...
A Platform Policy Implementation Audit of Actions against Russia’s State-Cont...Axel Bruns
 
Networks of Agonism and Antagonism: Polarised Discourses about COP26 (and COP...
Networks of Agonism and Antagonism: Polarised Discourses about COP26 (and COP...Networks of Agonism and Antagonism: Polarised Discourses about COP26 (and COP...
Networks of Agonism and Antagonism: Polarised Discourses about COP26 (and COP...Axel Bruns
 
The Filter in Our (?) Heads: Digital Media and Polarisation
The Filter in Our (?) Heads: Digital Media and PolarisationThe Filter in Our (?) Heads: Digital Media and Polarisation
The Filter in Our (?) Heads: Digital Media and PolarisationAxel Bruns
 
Gatewatching 5: Weaponising Newssharing: ‘Fake News’ and Other Malinformation
Gatewatching 5: Weaponising Newssharing: ‘Fake News’ and Other MalinformationGatewatching 5: Weaponising Newssharing: ‘Fake News’ and Other Malinformation
Gatewatching 5: Weaponising Newssharing: ‘Fake News’ and Other MalinformationAxel Bruns
 
Gatewatching 10: New(s) Publics in the Public Sphere
Gatewatching 10: New(s) Publics in the Public SphereGatewatching 10: New(s) Publics in the Public Sphere
Gatewatching 10: New(s) Publics in the Public SphereAxel Bruns
 
Gatewatching 4: Random Acts of Gatewatching: Everyday Newssharing Practices
Gatewatching 4: Random Acts of Gatewatching: Everyday Newssharing PracticesGatewatching 4: Random Acts of Gatewatching: Everyday Newssharing Practices
Gatewatching 4: Random Acts of Gatewatching: Everyday Newssharing PracticesAxel Bruns
 
Gatewatching 11: Echo Chambers? Filter Bubbles? Reviewing the Evidence
Gatewatching 11: Echo Chambers? Filter Bubbles? Reviewing the EvidenceGatewatching 11: Echo Chambers? Filter Bubbles? Reviewing the Evidence
Gatewatching 11: Echo Chambers? Filter Bubbles? Reviewing the EvidenceAxel Bruns
 
Gatewatching 1: Introduction: What’s So Different about Journalism Today?
Gatewatching 1: Introduction: What’s So Different about Journalism Today?Gatewatching 1: Introduction: What’s So Different about Journalism Today?
Gatewatching 1: Introduction: What’s So Different about Journalism Today?Axel Bruns
 
Gatewatching 8: Hybrid News Coverage: Liveblogs
Gatewatching 8: Hybrid News Coverage: LiveblogsGatewatching 8: Hybrid News Coverage: Liveblogs
Gatewatching 8: Hybrid News Coverage: LiveblogsAxel Bruns
 
Gatewatching 2: From Gatekeeping to Gatewatching: The First Wave of Citizen M...
Gatewatching 2: From Gatekeeping to Gatewatching: The First Wave of Citizen M...Gatewatching 2: From Gatekeeping to Gatewatching: The First Wave of Citizen M...
Gatewatching 2: From Gatekeeping to Gatewatching: The First Wave of Citizen M...Axel Bruns
 
Gatewatching 9: ‘Real’ News and ‘Fake’ News: Fact-Checking and Media Literacy
Gatewatching 9: ‘Real’ News and ‘Fake’ News: Fact-Checking and Media LiteracyGatewatching 9: ‘Real’ News and ‘Fake’ News: Fact-Checking and Media Literacy
Gatewatching 9: ‘Real’ News and ‘Fake’ News: Fact-Checking and Media LiteracyAxel Bruns
 

More from Axel Bruns (20)

Identifying the Symptoms of Destructive Polarisation
Identifying the Symptoms of Destructive PolarisationIdentifying the Symptoms of Destructive Polarisation
Identifying the Symptoms of Destructive Polarisation
 
Voices on the Voice Referendum: A Computational Analysis of News and Audience...
Voices on the Voice Referendum: A Computational Analysis of News and Audience...Voices on the Voice Referendum: A Computational Analysis of News and Audience...
Voices on the Voice Referendum: A Computational Analysis of News and Audience...
 
What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...
What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...
What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...
 
What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...
What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...
What Is Lost When Twitter Is Lost? Reflections on the Impending Death of a Pl...
 
Types of Polarisation and Their Operationalisation in Digital and Social Medi...
Types of Polarisation and Their Operationalisation in Digital and Social Medi...Types of Polarisation and Their Operationalisation in Digital and Social Medi...
Types of Polarisation and Their Operationalisation in Digital and Social Medi...
 
News Sharing and Partisanship: Tracking News Outlet Repertoires on Twitter ov...
News Sharing and Partisanship: Tracking News Outlet Repertoires on Twitter ov...News Sharing and Partisanship: Tracking News Outlet Repertoires on Twitter ov...
News Sharing and Partisanship: Tracking News Outlet Repertoires on Twitter ov...
 
Determining the Drivers and Dynamics of Partisanship and Polarisation in Onli...
Determining the Drivers and Dynamics of Partisanship and Polarisation in Onli...Determining the Drivers and Dynamics of Partisanship and Polarisation in Onli...
Determining the Drivers and Dynamics of Partisanship and Polarisation in Onli...
 
Towards a New Empiricism: Polarisation across Four Dimensions
Towards a New Empiricism: Polarisation across Four DimensionsTowards a New Empiricism: Polarisation across Four Dimensions
Towards a New Empiricism: Polarisation across Four Dimensions
 
The Anatomy of Virality: How COVID-19 Conspiracy Theories Spread across Socia...
The Anatomy of Virality: How COVID-19 Conspiracy Theories Spread across Socia...The Anatomy of Virality: How COVID-19 Conspiracy Theories Spread across Socia...
The Anatomy of Virality: How COVID-19 Conspiracy Theories Spread across Socia...
 
A Platform Policy Implementation Audit of Actions against Russia’s State-Cont...
A Platform Policy Implementation Audit of Actions against Russia’s State-Cont...A Platform Policy Implementation Audit of Actions against Russia’s State-Cont...
A Platform Policy Implementation Audit of Actions against Russia’s State-Cont...
 
Networks of Agonism and Antagonism: Polarised Discourses about COP26 (and COP...
Networks of Agonism and Antagonism: Polarised Discourses about COP26 (and COP...Networks of Agonism and Antagonism: Polarised Discourses about COP26 (and COP...
Networks of Agonism and Antagonism: Polarised Discourses about COP26 (and COP...
 
The Filter in Our (?) Heads: Digital Media and Polarisation
The Filter in Our (?) Heads: Digital Media and PolarisationThe Filter in Our (?) Heads: Digital Media and Polarisation
The Filter in Our (?) Heads: Digital Media and Polarisation
 
Gatewatching 5: Weaponising Newssharing: ‘Fake News’ and Other Malinformation
Gatewatching 5: Weaponising Newssharing: ‘Fake News’ and Other MalinformationGatewatching 5: Weaponising Newssharing: ‘Fake News’ and Other Malinformation
Gatewatching 5: Weaponising Newssharing: ‘Fake News’ and Other Malinformation
 
Gatewatching 10: New(s) Publics in the Public Sphere
Gatewatching 10: New(s) Publics in the Public SphereGatewatching 10: New(s) Publics in the Public Sphere
Gatewatching 10: New(s) Publics in the Public Sphere
 
Gatewatching 4: Random Acts of Gatewatching: Everyday Newssharing Practices
Gatewatching 4: Random Acts of Gatewatching: Everyday Newssharing PracticesGatewatching 4: Random Acts of Gatewatching: Everyday Newssharing Practices
Gatewatching 4: Random Acts of Gatewatching: Everyday Newssharing Practices
 
Gatewatching 11: Echo Chambers? Filter Bubbles? Reviewing the Evidence
Gatewatching 11: Echo Chambers? Filter Bubbles? Reviewing the EvidenceGatewatching 11: Echo Chambers? Filter Bubbles? Reviewing the Evidence
Gatewatching 11: Echo Chambers? Filter Bubbles? Reviewing the Evidence
 
Gatewatching 1: Introduction: What’s So Different about Journalism Today?
Gatewatching 1: Introduction: What’s So Different about Journalism Today?Gatewatching 1: Introduction: What’s So Different about Journalism Today?
Gatewatching 1: Introduction: What’s So Different about Journalism Today?
 
Gatewatching 8: Hybrid News Coverage: Liveblogs
Gatewatching 8: Hybrid News Coverage: LiveblogsGatewatching 8: Hybrid News Coverage: Liveblogs
Gatewatching 8: Hybrid News Coverage: Liveblogs
 
Gatewatching 2: From Gatekeeping to Gatewatching: The First Wave of Citizen M...
Gatewatching 2: From Gatekeeping to Gatewatching: The First Wave of Citizen M...Gatewatching 2: From Gatekeeping to Gatewatching: The First Wave of Citizen M...
Gatewatching 2: From Gatekeeping to Gatewatching: The First Wave of Citizen M...
 
Gatewatching 9: ‘Real’ News and ‘Fake’ News: Fact-Checking and Media Literacy
Gatewatching 9: ‘Real’ News and ‘Fake’ News: Fact-Checking and Media LiteracyGatewatching 9: ‘Real’ News and ‘Fake’ News: Fact-Checking and Media Literacy
Gatewatching 9: ‘Real’ News and ‘Fake’ News: Fact-Checking and Media Literacy
 

Detecting Twitter Bots That Promote SoundCloud Tracks

  • 1. @qutdmrc Detecting Twitter Bots That Share SoundCloud Tracks Axel Bruns, Brenda Moon, Felix Victor Münch, Patrik Wikström Digital Media Research Centre, Queensland University of Technology Stefan Stieglitz, Florian Brachten, Björn Ross Research Group Professional Communication in Electronic Media / Social Media, University of Duisburg-Essen SM&S 2018, Copenhagen @snurb_dot_info
  • 2. @qutdmrc Bots ● Bot types: ● Benign and legitimate ● Nefarious and subversive ● Political, commercial, … ● Exploiting affordances of social media platforms ● Main focus on Facebook and Twitter ● Bots as an industry: ● Fake followers ● Like, comment, share, retweet campaigns ● Trending topic generation ● URL promotion campaigns
  • 3. @qutdmrc Bots on / for SoundCloud
  • 4. @qutdmrc Bots on / for SoundCloud ● Native SoundCloud bots ● Plays, likes, comments, followers, etc. ● Addressed in Ross et al. (2018) ● SoundCloud-promoting Twitter bots: ● Sharing SoundCloud URLs, boosting SoundCloud metrics ● Can we detect such Twitter bots by their URL sharing patterns?
  • 5. @qutdmrc Datasets ● Data: ● Tweets that share soundcloud.com URLs (after unshortening) ● Selection of tweets posted during March / April 2017 ● 11.5m tweets posted by 2m Twitter accounts, sharing 2.1m tracks ● URLs standardised to soundcloud.com/[user name]/[track name] ● Top Tracks dataset: ● Tracks shared ≥ 1,000 times: 233 tracks, 914,131 tweets ● Top Accounts dataset: ● Accounts sharing tracks in ≥ 500 tweets: 649 accounts, 1m tweets
  • 7. @qutdmrc Metrics 1: Account Diversity ● Account Diversity: 𝐴𝑐𝑐𝑜𝑢𝑛𝑡 𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦 = # 𝑈𝑛𝑖𝑞𝑢𝑒 𝑇𝑤𝑖𝑡𝑡𝑒𝑟 𝐴𝑐𝑐𝑜𝑢𝑛𝑡𝑠 # 𝑇𝑤𝑒𝑒𝑡𝑠 ● Calculated per track: ● ≈0: persistent promotion of track by small number of accounts ● ≈1: many accounts sharing the same track
  • 8. @qutdmrc Metrics 2: Tweet Originality ● Tweet Originality: 𝑇𝑤𝑒𝑒𝑡 𝑂𝑟𝑖𝑔𝑖𝑛𝑎𝑙𝑖𝑡𝑦 = # 𝑁𝑜𝑛−𝑅𝑒𝑡𝑤𝑒𝑒𝑡𝑠 − # 𝑅𝑒𝑡𝑤𝑒𝑒𝑡𝑠 # 𝑇𝑤𝑒𝑒𝑡𝑠 ● Calculated per track: ● –1: all tweets sharing the track are retweets ● +1: all tweets sharing the track are original tweets / @mentions ● Calculated per account: ● –1: all SoundCloud URL tweets by account are retweets ● +1: all SoundCloud URL tweets are original tweets / @mentions
  • 9. @qutdmrc Metrics 3: Tweet Text Diversity ● Tweet Text Diversity: 𝑇𝑤𝑒𝑒𝑡 𝑇𝑒𝑥𝑡 𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦 = # 𝑈𝑛𝑖𝑞𝑢𝑒 𝑇𝑤𝑒𝑒𝑡 𝑇𝑒𝑥𝑡𝑠 # 𝑇𝑤𝑒𝑒𝑡𝑠 ● Calculated per track: ● ≈0: all tweets linking to SoundCloud track use identical text ● ≈1: all tweets linking to SoundCloud track use different texts ● Calculated per account: ● ≈0: all SoundCloud tweets posted by account use identical text ● ≈1: all SoundCloud tweets posted by account use different texts
  • 10. @qutdmrc Metrics 4: SoundCloud URL Diversity ● SoundCloud URL Diversity: 𝑆𝑜𝑢𝑛𝑑𝐶𝑙𝑜𝑢𝑑 𝑈𝑅𝐿 𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦 = # 𝑇𝑟𝑎𝑐𝑘𝑠 𝑆ℎ𝑎𝑟𝑒𝑑 # 𝑇𝑤𝑒𝑒𝑡𝑠 ● Calculated per account: ● ≈0: all SoundCloud tweets by account share the same URL ● ≈1: all SoundCloud tweets by account share a different URL
  • 12. @qutdmrc Analysis ● Preliminary analysis: ● Three per-track metrics: ● Account Diversity ● Tweet Originality ● Tweet Text Diversity ● Three per-account metrics: ● Tweet Originality ● Tweet Text Diversity ● SoundCloud URL Diversity ● Pairwise exploration of correlation patterns amongst metrics ● k-means clustering of tracks / accounts across each three metrics ● Abductive interpretation of patterns towards hypothesis formation
  • 13. @qutdmrc Metrics per Track retweeting many retweets by few accounts many original tweets by few accounts many original tweets by few accounts, using many different tweet texts
  • 14. @qutdmrc Metrics per Accountretweets originaltweets many original tracks, many original texts Few original tracks, many original texts repeated sharing but few retweets
  • 15. @qutdmrc Observations ● Proposed metrics: ● Promising distinctions between tracks and accounts ● Need to further validate patterns through qualitative assessment ● Need to correlate between per-track and per-account metrics ● Need to explore patterns around SoundCloud users, not just tracks ● Further application: ● Test metrics on larger datasets ● Develop and test bot detection approaches on SoundCloud itself, and compare findings with present study (Ross et al. 2018)
  • 16. @qutdmrc SM&S 2018, Copenhagen @snurb_dot_info Acknowledgments This research is supported by the UA-DAAD project Emergent Music Engagement Practices via SoundCloud and the ARC LIEF project TrISMA: Tracking Infrastructure for Social Media Analysis. Reference B. Ross, F. Brachten, S. Stieglitz, P. Wikström, B. Moon, F. Münch, and A. Bruns. 2018. Social Bots in a Commercial Context – A Case Study on SoundCloud. Paper presented at European Conference on Information Systems (ECIS), Portsmouth, 23-28 July 2018.