SlideShare a Scribd company logo
1 of 48
Download to read offline
ABUSIVE LANGUAGE
DETECTION—
COMPARATIVE STUDY AND
ITS APPLICATIONS
YOUNGHUN LEE
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
BEFORE WE BEGIN..
▸ B.S. in Electrical and Computer Engineering
at Seoul National University
▸ M.S. in Information at the University of
Michigan
▸ (Current) Research Intern at Machine
Intelligence Lab @ Seoul National University
▸ (Current) Ph.D. applicant in Machine
Learning and Natural Language Processing
2
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
WHAT THIS PRESENTATION IS COVERING
I. Introduction to Abusive Language Detection
▸ What is it? Why is it important to study?
▸ Recent studies
II. About the paper, “Comparative Studies of Detecting Abusive Language on Twitter”
▸ Research idea and its novelty
▸ Empirical results and findings
III. Future relevance with Naver
▸ Research possibilities and their challenges
3
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
ACKNOWLEDGEMENTS
▸ Zeerak Waseem
▸ Ph.D. Candidate at the University of Sheffield
▸ Organizer of the Abusive Language Online Workshop
4
I. INTRODUCTION TO
ABUSIVE LANGUAGE DETECTION
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
▸ Varying definition and terminology
6
EARLY DEFINITION
Messages that most users consider to
be annoying or upsetting1
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
▸ Varying definition and terminology
7
RECENT
Messages that under-rate a person or a
group on the basis of characteristics
(race, ethnicity, gender, religion, etc.)2
EARLY DEFINITION
Messages that most users consider to
be annoying or upsetting1
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
▸ Varying definition and terminology
8
EXAMPLE
1) Go fucking kill yourself useless
scumbag
2) Hell yeah! Go bitches!
3) Jews are lower class pigs
EARLY DEFINITION
Messages that most users consider to
be annoying or upsetting1
RECENT
Messages that under-rate a person or a
group on the basis of characteristics
(race, ethnicity, gender, religion, etc.)2
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
▸ Varying definition and terminology
9
EXAMPLE
1) Go fucking kill yourself useless
scumbag
2) Hell yeah! Go bitches!
3) Jews are lower class pigs
EARLY DEFINITION
Messages that most users consider to
be annoying or upsetting1
RECENT
Messages that under-rate a person or a
group on the basis of characteristics
(race, ethnicity, gender, religion, etc.)2
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
▸ Varying definition and terminology
10
EXAMPLE
1) Go fucking kill yourself useless
scumbag
2) Hell yeah! Go bitches!
3) Jews are lower class pigs
EARLY DEFINITION
Messages that most users consider to
be annoying or upsetting1
RECENT
Messages that under-rate a person or a
group on the basis of characteristics
(race, ethnicity, gender, religion, etc.)2
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
▸ Varying definition and terminology
11
TERMINOLOGY
abusive language
hate speech
offensive language
cyberbullying
hostile flames
vulgar language
insults, profanity
…
EARLY DEFINITION
Messages that most users consider to
be annoying or upsetting1
RECENT
Messages that under-rate a person or a
group on the basis of characteristics
(race, ethnicity, gender, religion, etc.)2
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
▸ Varying definition and terminology
12
TERMINOLOGY
abusive language
hate speech
offensive language
cyberbullying
hostile flames
vulgar language
insults, profanity
…
EARLY DEFINITION
Messages that most users consider to
be annoying or upsetting1
RECENT
Messages that under-rate a person or a
group on the basis of characteristics
(race, ethnicity, gender, religion, etc.)2
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
PRO)
Enables considering
diverse situations
CON)
Confusing definition
-> hard to annotate
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
▸ Why is it important these days?
13
1) Increasing usage of social media3
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
▸ Why is it important these days?
14
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
1) Increasing usage of social media
2) Social media significantly affects
current day society
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
▸ Why is it important these days?
15
4
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
▸ Why is it important these days?
16
5
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
1) Increasing usage of social media
2) Social media significantly affects
current day society
3) More and more users have been
experiencing online harassment
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
▸ Why is it important these days?
17
1) Increasing usage of social media
2) Social media significantly affects
current day society
3) More and more users have been
experiencing online harassment
4) Yet, major social media companies
fail to successfully resolve the issueArtist stencils hate speech tweets
outside Twitter HQ to highlight failure
to deal with offensive messages
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
▸ Characteristics of Abusive Language Online
▸ Abusive messages might be influenced by…6
18
- its discourse context
- its co-occurring media (images, videos)
- world events
- identity of the author and target
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
▸ Characteristics of Abusive Language Online
▸ Abusive messages might be influenced by…6
▸ What is it different from other advanced text classification tasks such as
sarcasm detection and fake news detection?
19
- its discourse context
- its co-occurring media (images, videos)
- world events
- identity of the author and target
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
- CONTEXT is its core (e.g. ‘nigger’, ‘bitch’)
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
▸ Recent studies
▸ Data crawling and annotation
▸ Feature-engineering models for accurate classification
▸ Using meta-information such as age, gender, location as additional features
▸ Applying deep models
▸ Concerning fairness issues
▸ Resolving unintentional bias issue (Woman, Jew being abusive?)
20
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
▸ Future focus
▸ Creating a gold-standard test set
▸ More reliable annotations
▸ Getting more context information
▸ Multi-modal abusive detection
21
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
▸ Future focus
▸ Creating a gold-standard test set
▸ More reliable annotations
▸ Getting more context information
▸ Multi-modal abusive detection
22
II. ABOUT THE PAPER,
“COMPARATIVE STUDIES OF DETECTING
ABUSIVE LANGUAGE ON TWITTER”
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
▸ Brief overview of the paper
Comparing the accuracy (F1 scores) of different machine learning models and
different features in detecting abusive language on a recently released Twitter
dataset
24
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
▸ About the dataset, “Hate and Abusive Speech on Twitter” 7
▸ Approximately 100K labeled tweets (other datasets: 10K to 35K)
▸ Reduced overlapping labels by calculating correlation coefficients
▸ E.g. ‘abusive’, ‘offensive’, ‘aggressive’, ‘cyberbullying’ ‘abusive’
▸ Multi-label dataset (4 labels: abusive, hateful, spam, none)
25
Label distribution of crawled tweets
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
▸ Implemented models and features
26
• Naïve Bayes

• Logistic Regression

• Support Vector Machine

• Random Forests

• Gradient Boosted Trees

• Word-level / Char-level
FEATURE-ENGINEERING ML NEURAL NETWORKS
• Conv Neural Network

• Recurrent Neural Network

• Word-level / Char-level

• Hybrid CNN

• Self-matching attention RNN

• Latent Topic Clustering

• Context Tweets
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
▸ Implemented models and features
27
NEURAL NETWORKS
• Conv Neural Network

• Recurrent Neural Network

• Word-level / Char-level

• Hybrid CNN
• Self-matching attention RNN

• Latent Topic Clustering

• Context Tweets
Architecture of Hybrid CNN8
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
▸ Implemented models and features
28
NEURAL NETWORKS
• Conv Neural Network

• Recurrent Neural Network

• Word-level / Char-level

• Hybrid CNN

• Self-matching attention RNN
• Latent Topic Clustering

• Context Tweets
Gated self-matching attention networks (Wang et al.,)9
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
▸ Implemented models and features
29
NEURAL NETWORKS
• Conv Neural Network

• Recurrent Neural Network

• Word-level / Char-level

• Hybrid CNN

• Self-matching attention RNN

• Latent Topic Clustering
• Context Tweets
Hierarchical Recurrent Dual Encoder with
Latent Topic Clustering module10
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
▸ Implemented models and features
30
NEURAL NETWORKS
• Conv Neural Network

• Recurrent Neural Network

• Word-level / Char-level

• Hybrid CNN

• Self-matching attention RNN

• Latent Topic Clustering

• Context Tweets
WHAT ARE CONTEXT TWEETS?
‣ Looking at the tweet one has
replied to or has quoted provides
significant contextual information
‣ Assumption: ML models benefit
from taking context tweets into
account in detecting abusive
language.
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
▸ Implemented models and features
31
NEURAL NETWORKS
• Conv Neural Network

• Recurrent Neural Network

• Word-level / Char-level

• Hybrid CNN

• Self-matching attention RNN

• Latent Topic Clustering

• Context Tweets
SAMPLE CONTEXT TWEET
Who the HELL is “LIKE” ING this post?
Sick people….
@user_B LABELED TWEET
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
▸ Implemented models and features
32
NEURAL NETWORKS
• Conv Neural Network

• Recurrent Neural Network

• Word-level / Char-level

• Hybrid CNN

• Self-matching attention RNN

• Latent Topic Clustering

• Context Tweets
SAMPLE CONTEXT TWEET
Survivors of #Syria Gas Attack Recount
‘a Cruel Scene’.
@user_A
Who the HELL is “LIKE” ING this post?
Sick people….
@user_B
CONTEXT TWEET
LABELED TWEET
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
▸ Implemented models and features
33
NEURAL NETWORKS
• Conv Neural Network

• Recurrent Neural Network

• Word-level / Char-level

• Hybrid CNN

• Self-matching attention RNN

• Latent Topic Clustering

• Context Tweets
HOW TO INTEGRATE CONTEXT TWEETS
LABELED
TWEET
CONTEXT
TWEET
RNN
CNN
max_pooled
max_pooled
last_hidden
last_hidden
σ
σ
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
34
Experimental results of learning models and their variants, followed by the context tweet models.
The top 2 scores are marked as bold for each metric.
‣ Neural network
models with word-
level features are
accurate in general
compared to feature-
engineering models
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
35
Experimental results of learning models and their variants, followed by the context tweet models.
The top 2 scores are marked as bold for each metric.
‣ Neural network
models with word-
level features are
accurate in general
compared to feature-
engineering models
‣ Char-level features
benefit feature-
engineering ML
models while cripple
neural models
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
36
Experimental results of learning models and their variants, followed by the context tweet models.
The top 2 scores are marked as bold for each metric.
‣ Neural network
models with word-
level features are
accurate in general
compared to feature-
engineering models
‣ Char-level features
benefit feature-
engineering ML
models while cripple
neural models
‣ Context data doesn’t
improve the overall
accuracy, however, it
is more effective in
classifying lower-
distributed labels
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
37
▸ Discussion
BASELINE CONTEXT
p_normal

.1614

p_abusive
.5958
p_normal

.3709

p_abusive

.2582
LABEL: abusive
8 dudes added me in a group chat
telling me I should kill myself,
lmfaooo I think I reached a whole
new level of twitter.
@user_A
They hatin cause you a damn
queen who doesn’t answer to
anybody. Let em be mad. Keep
rolling mamas.
@user_B
‣ Only baseline model
was correct, but is
‘abusive’ label for
this tweet accurate?
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
38
▸ Discussion
BASELINE CONTEXT
LABEL: normal
Oklahoma home invasion shooting:
No charges against man who killed
3 intruders.
@user_C
You have every right to protect your
family and home from worthless
thieves.
@user_D
‣ Context tweets gave
more information to
accurately classify
the tweet
p_normal

.3287

p_abusive

.3718
p_normal
.8104
p_abusive

.0962
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER
39
▸ Discussion
▸ How to efficiently train the classifier for imbalanced datasets?
▸ How can we better incorporate context tweets into the original data? How to
handle ‘no_context’?
▸ Can we annotate the dataset regarding context data? Specifically, can we
make annotators consider the content of its context tweet when deciding
abusiveness of the tweet?
▸ What are other applications that could benefit from using context data?
III. FUTURE RELEVANCE
WITH NAVER
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
III. FUTURE RELEVANCE WITH NAVER
▸ Hate Speech in Korea
▸ 남성혐오 / 여성혐오 (워마드, 일베저장소)
▸ 극단적 진보 / 보수 정치 성향
▸ 사회적 문제로 확장
41
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
III. FUTURE RELEVANCE WITH NAVER
▸ Possible Contribution
▸ DATASET!!!
42
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
III. FUTURE RELEVANCE WITH NAVER
43
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
III. FUTURE RELEVANCE WITH NAVER
44
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
III. FUTURE RELEVANCE WITH NAVER
45
RICH TEXTUAL & CONTEXTUAL
INFORMATION FROM ONE NEWS ARTICLE
1) Article, title, and its topic
2) Comments and comments of comments
2) Emotions on the article
3) Number of comments, demographic
distribution of commenters
4) Upvotes and downvotes of comments
5) User meta-information
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
III. FUTURE RELEVANCE WITH NAVER
▸ Possible applications
▸ Detecting abusive language in news article comments
▸ Enabling personalized settings for not encountering toxic messages
▸ CLOVA: read comments of news articles that are not abusive
▸ Challenges
▸ Korean text: not extensively studied compared to English
▸ Lack of user meta-information
46
/ 48
ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS
REFERENCE
1. Ellen Spertus. 1997. Smokey: Automatic recognition of hostile messages. In AAAI/IAAI, pages 1058– 1065.
2. John T. Nockleby. 2000. Hate Speech. In Leonard W. Levy, Kenneth L. Karst, and Dennis J. Mahoney, editors, Encyclopedia of
the American Constitution, pages 1277–1279. Macmillan, 2nd edition.
3. Aaron Smith and Monica Anderson. 2018. Social Media Use in 2018. Pew Research Center; accessed 8-December-2018.
4. Monica Anderson and Skye Toor. How social media users have discussed sexual harassment since #MeToo went viral. Pew
Research Center; accessed 8-December-2018.
5. Maeve Duggan. 2017. Online harassment 2017. Pew Research Center; accessed 8-December-2018.
6. Anna Schmidt and Michael Wiegand. 2017. A survey on hate speech detection using natural language pro- cessing. In
Proceedings of the Fifth International Workshop on Natural Language Processing for So- cial Media, pages 1–10.
7. Antigoni Founta, Constantinos Djouvas, Despoina Chatzakou, Ilias Leontiadis, Jeremy Blackburn, Gianluca Stringhini, Athena
Vakali, Michael Sirivianos, and Nicolas Kourtellis. 2018. Large scale crowdsourcing and characterization of twitter abusive
behavior. In Proceedings of the International AAAI Conference on Web and Social Media.
8. Ji Ho Park and Pascale Fung. 2017. One-step and two- step classification for abusive language detection on twitter. In
Proceedings of the First Workshop on Abusive Language Online, pages 41–45.
9. Wenhui Wang, Nan Yang, Furu Wei, Baobao Chang, and Ming Zhou. 2017. Gated self-matching networks for reading
comprehension and question answering. In Proceedings of the 55th Annual Meeting of the Association for Computational
Linguistics, volume 1, pages 189–198.
10.Seunghyun Yoon, Joongbo Shin, and Kyomin Jung. 2018. Learning to rank question-answer pairs using hierarchical recurrent
encoder with latent topic clustering. In Proceedings of the 2018 Conference of the North American Chapter of the Association
for Computational Linguistics: Human Language Technologies, volume 1, pages 1575–1584.
47
THANK YOU!!

More Related Content

What's hot

Substitution cipher and Its Cryptanalysis
Substitution cipher and Its CryptanalysisSubstitution cipher and Its Cryptanalysis
Substitution cipher and Its CryptanalysisSunil Meena
 
Asymptotic Notation and Complexity
Asymptotic Notation and ComplexityAsymptotic Notation and Complexity
Asymptotic Notation and ComplexityRajandeep Gill
 
Recursion with Python [Rev]
Recursion with Python [Rev]Recursion with Python [Rev]
Recursion with Python [Rev]Dennis Walangadi
 
Toxic Comment Classification
Toxic Comment ClassificationToxic Comment Classification
Toxic Comment Classificationijtsrd
 
Solving recurrences
Solving recurrencesSolving recurrences
Solving recurrencesMegha V
 
Cryptography and network security
Cryptography and network securityCryptography and network security
Cryptography and network securitypatisa
 
Divide and Conquer - Part 1
Divide and Conquer - Part 1Divide and Conquer - Part 1
Divide and Conquer - Part 1Amrinder Arora
 
Introduction to NumPy (PyData SV 2013)
Introduction to NumPy (PyData SV 2013)Introduction to NumPy (PyData SV 2013)
Introduction to NumPy (PyData SV 2013)PyData
 
OpenGL Mini Projects With Source Code [ Computer Graphics ]
OpenGL Mini Projects With Source Code [ Computer Graphics ]OpenGL Mini Projects With Source Code [ Computer Graphics ]
OpenGL Mini Projects With Source Code [ Computer Graphics ]Daffodil International University
 
Stressen's matrix multiplication
Stressen's matrix multiplicationStressen's matrix multiplication
Stressen's matrix multiplicationKumar
 
Rabin karp string matching algorithm
Rabin karp string matching algorithmRabin karp string matching algorithm
Rabin karp string matching algorithmGajanand Sharma
 
Telephone directory in c
Telephone directory in cTelephone directory in c
Telephone directory in cUpendra Sengar
 

What's hot (20)

Asymptotic notation
Asymptotic notationAsymptotic notation
Asymptotic notation
 
Substitution cipher and Its Cryptanalysis
Substitution cipher and Its CryptanalysisSubstitution cipher and Its Cryptanalysis
Substitution cipher and Its Cryptanalysis
 
Asymptotic Notation and Complexity
Asymptotic Notation and ComplexityAsymptotic Notation and Complexity
Asymptotic Notation and Complexity
 
Preprocessor
PreprocessorPreprocessor
Preprocessor
 
Deep Learning for Computer Vision: Data Augmentation (UPC 2016)
Deep Learning for Computer Vision: Data Augmentation (UPC 2016)Deep Learning for Computer Vision: Data Augmentation (UPC 2016)
Deep Learning for Computer Vision: Data Augmentation (UPC 2016)
 
Recursion with Python [Rev]
Recursion with Python [Rev]Recursion with Python [Rev]
Recursion with Python [Rev]
 
Toxic Comment Classification
Toxic Comment ClassificationToxic Comment Classification
Toxic Comment Classification
 
Solving recurrences
Solving recurrencesSolving recurrences
Solving recurrences
 
Public key cryptography and RSA
Public key cryptography and RSAPublic key cryptography and RSA
Public key cryptography and RSA
 
Divide and Conquer
Divide and ConquerDivide and Conquer
Divide and Conquer
 
Generative adversarial text to image synthesis
Generative adversarial text to image synthesisGenerative adversarial text to image synthesis
Generative adversarial text to image synthesis
 
Cryptography and network security
Cryptography and network securityCryptography and network security
Cryptography and network security
 
Diffiehellman
DiffiehellmanDiffiehellman
Diffiehellman
 
Divide and Conquer - Part 1
Divide and Conquer - Part 1Divide and Conquer - Part 1
Divide and Conquer - Part 1
 
Hash function
Hash function Hash function
Hash function
 
Introduction to NumPy (PyData SV 2013)
Introduction to NumPy (PyData SV 2013)Introduction to NumPy (PyData SV 2013)
Introduction to NumPy (PyData SV 2013)
 
OpenGL Mini Projects With Source Code [ Computer Graphics ]
OpenGL Mini Projects With Source Code [ Computer Graphics ]OpenGL Mini Projects With Source Code [ Computer Graphics ]
OpenGL Mini Projects With Source Code [ Computer Graphics ]
 
Stressen's matrix multiplication
Stressen's matrix multiplicationStressen's matrix multiplication
Stressen's matrix multiplication
 
Rabin karp string matching algorithm
Rabin karp string matching algorithmRabin karp string matching algorithm
Rabin karp string matching algorithm
 
Telephone directory in c
Telephone directory in cTelephone directory in c
Telephone directory in c
 

Similar to Comparative studies on detecting abusive language on twitter

2206 FAccT_inperson
2206 FAccT_inperson2206 FAccT_inperson
2206 FAccT_inpersonWarNik Chow
 
Computational Social Science: Programming Skills for Social Scientists
Computational Social Science: Programming Skills for Social ScientistsComputational Social Science: Programming Skills for Social Scientists
Computational Social Science: Programming Skills for Social ScientistsJames Allen-Robertson
 
AI in between online and offline discourse - and what has ChatGPT to do with ...
AI in between online and offline discourse - and what has ChatGPT to do with ...AI in between online and offline discourse - and what has ChatGPT to do with ...
AI in between online and offline discourse - and what has ChatGPT to do with ...Stefan Dietze
 
An Analytical Survey on Hate Speech Recognition through NLP and Deep Learning
An Analytical Survey on Hate Speech Recognition through NLP and Deep LearningAn Analytical Survey on Hate Speech Recognition through NLP and Deep Learning
An Analytical Survey on Hate Speech Recognition through NLP and Deep LearningIRJET Journal
 
Studying Cybercrime: Raising Awareness of Objectivity & Bias
Studying Cybercrime: Raising Awareness of Objectivity & BiasStudying Cybercrime: Raising Awareness of Objectivity & Bias
Studying Cybercrime: Raising Awareness of Objectivity & Biasgloriakt
 
Qualitative Methods in International Relations - Chapters 5, 8, 10
Qualitative Methods in International Relations - Chapters 5, 8, 10Qualitative Methods in International Relations - Chapters 5, 8, 10
Qualitative Methods in International Relations - Chapters 5, 8, 10Bahria University, Islamabad
 
Netnography and Research Ethics: From ACR 2015 Doctoral Symposium
Netnography and Research Ethics: From ACR 2015 Doctoral SymposiumNetnography and Research Ethics: From ACR 2015 Doctoral Symposium
Netnography and Research Ethics: From ACR 2015 Doctoral SymposiumUniversity of Southern California
 
Aspects of Impoliteness during 2007 and 2013 Presidential Campaigns in Kenya
Aspects of Impoliteness during 2007 and 2013 Presidential Campaigns in KenyaAspects of Impoliteness during 2007 and 2013 Presidential Campaigns in Kenya
Aspects of Impoliteness during 2007 and 2013 Presidential Campaigns in KenyaAJSSMTJournal
 
Octopus and Midget in the Israeli-Palestinian Peace Process: Who Determines W...
Octopus and Midget in the Israeli-Palestinian Peace Process: Who Determines W...Octopus and Midget in the Israeli-Palestinian Peace Process: Who Determines W...
Octopus and Midget in the Israeli-Palestinian Peace Process: Who Determines W...AJSSMTJournal
 
Ibm cog institutetalk_diab
Ibm cog institutetalk_diabIbm cog institutetalk_diab
Ibm cog institutetalk_diabdiannepatricia
 
The Networked Creativity in the Censored Web 2.0
The Networked Creativity in the Censored Web 2.0The Networked Creativity in the Censored Web 2.0
The Networked Creativity in the Censored Web 2.0Weiai Wayne Xu
 
QE. Strength of Ties under conditions of anonymity
QE. Strength of Ties under conditions of anonymityQE. Strength of Ties under conditions of anonymity
QE. Strength of Ties under conditions of anonymityHerbert Eng
 
UCWI-Safety & Security Final Presentation-5
UCWI-Safety & Security Final Presentation-5UCWI-Safety & Security Final Presentation-5
UCWI-Safety & Security Final Presentation-5晰 王
 
Objectification Is A Word That Has Many Negative Connotations
Objectification Is A Word That Has Many Negative ConnotationsObjectification Is A Word That Has Many Negative Connotations
Objectification Is A Word That Has Many Negative ConnotationsBeth Johnson
 
Gatekeeping framing theory summary
Gatekeeping framing theory summaryGatekeeping framing theory summary
Gatekeeping framing theory summaryMinCheol Shin
 

Similar to Comparative studies on detecting abusive language on twitter (20)

A systematic literature review of academic cyberbullying 2021
A systematic literature review of academic cyberbullying 2021A systematic literature review of academic cyberbullying 2021
A systematic literature review of academic cyberbullying 2021
 
2206 FAccT_inperson
2206 FAccT_inperson2206 FAccT_inperson
2206 FAccT_inperson
 
Computational Social Science: Programming Skills for Social Scientists
Computational Social Science: Programming Skills for Social ScientistsComputational Social Science: Programming Skills for Social Scientists
Computational Social Science: Programming Skills for Social Scientists
 
New literacies Definition and Components
New literacies Definition and ComponentsNew literacies Definition and Components
New literacies Definition and Components
 
cuhk-fb-mi-talk.pdf
cuhk-fb-mi-talk.pdfcuhk-fb-mi-talk.pdf
cuhk-fb-mi-talk.pdf
 
AI in between online and offline discourse - and what has ChatGPT to do with ...
AI in between online and offline discourse - and what has ChatGPT to do with ...AI in between online and offline discourse - and what has ChatGPT to do with ...
AI in between online and offline discourse - and what has ChatGPT to do with ...
 
An Analytical Survey on Hate Speech Recognition through NLP and Deep Learning
An Analytical Survey on Hate Speech Recognition through NLP and Deep LearningAn Analytical Survey on Hate Speech Recognition through NLP and Deep Learning
An Analytical Survey on Hate Speech Recognition through NLP and Deep Learning
 
Metlit bru
Metlit bruMetlit bru
Metlit bru
 
Studying Cybercrime: Raising Awareness of Objectivity & Bias
Studying Cybercrime: Raising Awareness of Objectivity & BiasStudying Cybercrime: Raising Awareness of Objectivity & Bias
Studying Cybercrime: Raising Awareness of Objectivity & Bias
 
Qualitative Methods in International Relations - Chapters 5, 8, 10
Qualitative Methods in International Relations - Chapters 5, 8, 10Qualitative Methods in International Relations - Chapters 5, 8, 10
Qualitative Methods in International Relations - Chapters 5, 8, 10
 
Netnography and Research Ethics: From ACR 2015 Doctoral Symposium
Netnography and Research Ethics: From ACR 2015 Doctoral SymposiumNetnography and Research Ethics: From ACR 2015 Doctoral Symposium
Netnography and Research Ethics: From ACR 2015 Doctoral Symposium
 
Small Ideas for ICRC
Small Ideas for ICRCSmall Ideas for ICRC
Small Ideas for ICRC
 
Aspects of Impoliteness during 2007 and 2013 Presidential Campaigns in Kenya
Aspects of Impoliteness during 2007 and 2013 Presidential Campaigns in KenyaAspects of Impoliteness during 2007 and 2013 Presidential Campaigns in Kenya
Aspects of Impoliteness during 2007 and 2013 Presidential Campaigns in Kenya
 
Octopus and Midget in the Israeli-Palestinian Peace Process: Who Determines W...
Octopus and Midget in the Israeli-Palestinian Peace Process: Who Determines W...Octopus and Midget in the Israeli-Palestinian Peace Process: Who Determines W...
Octopus and Midget in the Israeli-Palestinian Peace Process: Who Determines W...
 
Ibm cog institutetalk_diab
Ibm cog institutetalk_diabIbm cog institutetalk_diab
Ibm cog institutetalk_diab
 
The Networked Creativity in the Censored Web 2.0
The Networked Creativity in the Censored Web 2.0The Networked Creativity in the Censored Web 2.0
The Networked Creativity in the Censored Web 2.0
 
QE. Strength of Ties under conditions of anonymity
QE. Strength of Ties under conditions of anonymityQE. Strength of Ties under conditions of anonymity
QE. Strength of Ties under conditions of anonymity
 
UCWI-Safety & Security Final Presentation-5
UCWI-Safety & Security Final Presentation-5UCWI-Safety & Security Final Presentation-5
UCWI-Safety & Security Final Presentation-5
 
Objectification Is A Word That Has Many Negative Connotations
Objectification Is A Word That Has Many Negative ConnotationsObjectification Is A Word That Has Many Negative Connotations
Objectification Is A Word That Has Many Negative Connotations
 
Gatekeeping framing theory summary
Gatekeeping framing theory summaryGatekeeping framing theory summary
Gatekeeping framing theory summary
 

More from NAVER Engineering

디자인 시스템에 직방 ZUIX
디자인 시스템에 직방 ZUIX디자인 시스템에 직방 ZUIX
디자인 시스템에 직방 ZUIXNAVER Engineering
 
진화하는 디자인 시스템(걸음마 편)
진화하는 디자인 시스템(걸음마 편)진화하는 디자인 시스템(걸음마 편)
진화하는 디자인 시스템(걸음마 편)NAVER Engineering
 
서비스 운영을 위한 디자인시스템 프로젝트
서비스 운영을 위한 디자인시스템 프로젝트서비스 운영을 위한 디자인시스템 프로젝트
서비스 운영을 위한 디자인시스템 프로젝트NAVER Engineering
 
BPL(Banksalad Product Language) 무야호
BPL(Banksalad Product Language) 무야호BPL(Banksalad Product Language) 무야호
BPL(Banksalad Product Language) 무야호NAVER Engineering
 
이번 생에 디자인 시스템은 처음이라
이번 생에 디자인 시스템은 처음이라이번 생에 디자인 시스템은 처음이라
이번 생에 디자인 시스템은 처음이라NAVER Engineering
 
날고 있는 여러 비행기 넘나 들며 정비하기
날고 있는 여러 비행기 넘나 들며 정비하기날고 있는 여러 비행기 넘나 들며 정비하기
날고 있는 여러 비행기 넘나 들며 정비하기NAVER Engineering
 
쏘카프레임 구축 배경과 과정
 쏘카프레임 구축 배경과 과정 쏘카프레임 구축 배경과 과정
쏘카프레임 구축 배경과 과정NAVER Engineering
 
플랫폼 디자이너 없이 디자인 시스템을 구축하는 프로덕트 디자이너의 우당탕탕 고통 연대기
플랫폼 디자이너 없이 디자인 시스템을 구축하는 프로덕트 디자이너의 우당탕탕 고통 연대기플랫폼 디자이너 없이 디자인 시스템을 구축하는 프로덕트 디자이너의 우당탕탕 고통 연대기
플랫폼 디자이너 없이 디자인 시스템을 구축하는 프로덕트 디자이너의 우당탕탕 고통 연대기NAVER Engineering
 
200820 NAVER TECH CONCERT 15_Code Review is Horse(코드리뷰는 말이야)(feat.Latte)
200820 NAVER TECH CONCERT 15_Code Review is Horse(코드리뷰는 말이야)(feat.Latte)200820 NAVER TECH CONCERT 15_Code Review is Horse(코드리뷰는 말이야)(feat.Latte)
200820 NAVER TECH CONCERT 15_Code Review is Horse(코드리뷰는 말이야)(feat.Latte)NAVER Engineering
 
200819 NAVER TECH CONCERT 03_화려한 코루틴이 내 앱을 감싸네! 코루틴으로 작성해보는 깔끔한 비동기 코드
200819 NAVER TECH CONCERT 03_화려한 코루틴이 내 앱을 감싸네! 코루틴으로 작성해보는 깔끔한 비동기 코드200819 NAVER TECH CONCERT 03_화려한 코루틴이 내 앱을 감싸네! 코루틴으로 작성해보는 깔끔한 비동기 코드
200819 NAVER TECH CONCERT 03_화려한 코루틴이 내 앱을 감싸네! 코루틴으로 작성해보는 깔끔한 비동기 코드NAVER Engineering
 
200819 NAVER TECH CONCERT 10_맥북에서도 아이맥프로에서 빌드하는 것처럼 빌드 속도 빠르게 하기
200819 NAVER TECH CONCERT 10_맥북에서도 아이맥프로에서 빌드하는 것처럼 빌드 속도 빠르게 하기200819 NAVER TECH CONCERT 10_맥북에서도 아이맥프로에서 빌드하는 것처럼 빌드 속도 빠르게 하기
200819 NAVER TECH CONCERT 10_맥북에서도 아이맥프로에서 빌드하는 것처럼 빌드 속도 빠르게 하기NAVER Engineering
 
200819 NAVER TECH CONCERT 08_성능을 고민하는 슬기로운 개발자 생활
200819 NAVER TECH CONCERT 08_성능을 고민하는 슬기로운 개발자 생활200819 NAVER TECH CONCERT 08_성능을 고민하는 슬기로운 개발자 생활
200819 NAVER TECH CONCERT 08_성능을 고민하는 슬기로운 개발자 생활NAVER Engineering
 
200819 NAVER TECH CONCERT 05_모르면 손해보는 Android 디버깅/분석 꿀팁 대방출
200819 NAVER TECH CONCERT 05_모르면 손해보는 Android 디버깅/분석 꿀팁 대방출200819 NAVER TECH CONCERT 05_모르면 손해보는 Android 디버깅/분석 꿀팁 대방출
200819 NAVER TECH CONCERT 05_모르면 손해보는 Android 디버깅/분석 꿀팁 대방출NAVER Engineering
 
200819 NAVER TECH CONCERT 09_Case.xcodeproj - 좋은 동료로 거듭나기 위한 노하우
200819 NAVER TECH CONCERT 09_Case.xcodeproj - 좋은 동료로 거듭나기 위한 노하우200819 NAVER TECH CONCERT 09_Case.xcodeproj - 좋은 동료로 거듭나기 위한 노하우
200819 NAVER TECH CONCERT 09_Case.xcodeproj - 좋은 동료로 거듭나기 위한 노하우NAVER Engineering
 
200820 NAVER TECH CONCERT 14_야 너두 할 수 있어. 비전공자, COBOL 개발자를 거쳐 네이버에서 FE 개발하게 된...
200820 NAVER TECH CONCERT 14_야 너두 할 수 있어. 비전공자, COBOL 개발자를 거쳐 네이버에서 FE 개발하게 된...200820 NAVER TECH CONCERT 14_야 너두 할 수 있어. 비전공자, COBOL 개발자를 거쳐 네이버에서 FE 개발하게 된...
200820 NAVER TECH CONCERT 14_야 너두 할 수 있어. 비전공자, COBOL 개발자를 거쳐 네이버에서 FE 개발하게 된...NAVER Engineering
 
200820 NAVER TECH CONCERT 13_네이버에서 오픈 소스 개발을 통해 성장하는 방법
200820 NAVER TECH CONCERT 13_네이버에서 오픈 소스 개발을 통해 성장하는 방법200820 NAVER TECH CONCERT 13_네이버에서 오픈 소스 개발을 통해 성장하는 방법
200820 NAVER TECH CONCERT 13_네이버에서 오픈 소스 개발을 통해 성장하는 방법NAVER Engineering
 
200820 NAVER TECH CONCERT 12_상반기 네이버 인턴을 돌아보며
200820 NAVER TECH CONCERT 12_상반기 네이버 인턴을 돌아보며200820 NAVER TECH CONCERT 12_상반기 네이버 인턴을 돌아보며
200820 NAVER TECH CONCERT 12_상반기 네이버 인턴을 돌아보며NAVER Engineering
 
200820 NAVER TECH CONCERT 11_빠르게 성장하는 슈퍼루키로 거듭나기
200820 NAVER TECH CONCERT 11_빠르게 성장하는 슈퍼루키로 거듭나기200820 NAVER TECH CONCERT 11_빠르게 성장하는 슈퍼루키로 거듭나기
200820 NAVER TECH CONCERT 11_빠르게 성장하는 슈퍼루키로 거듭나기NAVER Engineering
 
200819 NAVER TECH CONCERT 07_신입 iOS 개발자 개발업무 적응기
200819 NAVER TECH CONCERT 07_신입 iOS 개발자 개발업무 적응기200819 NAVER TECH CONCERT 07_신입 iOS 개발자 개발업무 적응기
200819 NAVER TECH CONCERT 07_신입 iOS 개발자 개발업무 적응기NAVER Engineering
 

More from NAVER Engineering (20)

React vac pattern
React vac patternReact vac pattern
React vac pattern
 
디자인 시스템에 직방 ZUIX
디자인 시스템에 직방 ZUIX디자인 시스템에 직방 ZUIX
디자인 시스템에 직방 ZUIX
 
진화하는 디자인 시스템(걸음마 편)
진화하는 디자인 시스템(걸음마 편)진화하는 디자인 시스템(걸음마 편)
진화하는 디자인 시스템(걸음마 편)
 
서비스 운영을 위한 디자인시스템 프로젝트
서비스 운영을 위한 디자인시스템 프로젝트서비스 운영을 위한 디자인시스템 프로젝트
서비스 운영을 위한 디자인시스템 프로젝트
 
BPL(Banksalad Product Language) 무야호
BPL(Banksalad Product Language) 무야호BPL(Banksalad Product Language) 무야호
BPL(Banksalad Product Language) 무야호
 
이번 생에 디자인 시스템은 처음이라
이번 생에 디자인 시스템은 처음이라이번 생에 디자인 시스템은 처음이라
이번 생에 디자인 시스템은 처음이라
 
날고 있는 여러 비행기 넘나 들며 정비하기
날고 있는 여러 비행기 넘나 들며 정비하기날고 있는 여러 비행기 넘나 들며 정비하기
날고 있는 여러 비행기 넘나 들며 정비하기
 
쏘카프레임 구축 배경과 과정
 쏘카프레임 구축 배경과 과정 쏘카프레임 구축 배경과 과정
쏘카프레임 구축 배경과 과정
 
플랫폼 디자이너 없이 디자인 시스템을 구축하는 프로덕트 디자이너의 우당탕탕 고통 연대기
플랫폼 디자이너 없이 디자인 시스템을 구축하는 프로덕트 디자이너의 우당탕탕 고통 연대기플랫폼 디자이너 없이 디자인 시스템을 구축하는 프로덕트 디자이너의 우당탕탕 고통 연대기
플랫폼 디자이너 없이 디자인 시스템을 구축하는 프로덕트 디자이너의 우당탕탕 고통 연대기
 
200820 NAVER TECH CONCERT 15_Code Review is Horse(코드리뷰는 말이야)(feat.Latte)
200820 NAVER TECH CONCERT 15_Code Review is Horse(코드리뷰는 말이야)(feat.Latte)200820 NAVER TECH CONCERT 15_Code Review is Horse(코드리뷰는 말이야)(feat.Latte)
200820 NAVER TECH CONCERT 15_Code Review is Horse(코드리뷰는 말이야)(feat.Latte)
 
200819 NAVER TECH CONCERT 03_화려한 코루틴이 내 앱을 감싸네! 코루틴으로 작성해보는 깔끔한 비동기 코드
200819 NAVER TECH CONCERT 03_화려한 코루틴이 내 앱을 감싸네! 코루틴으로 작성해보는 깔끔한 비동기 코드200819 NAVER TECH CONCERT 03_화려한 코루틴이 내 앱을 감싸네! 코루틴으로 작성해보는 깔끔한 비동기 코드
200819 NAVER TECH CONCERT 03_화려한 코루틴이 내 앱을 감싸네! 코루틴으로 작성해보는 깔끔한 비동기 코드
 
200819 NAVER TECH CONCERT 10_맥북에서도 아이맥프로에서 빌드하는 것처럼 빌드 속도 빠르게 하기
200819 NAVER TECH CONCERT 10_맥북에서도 아이맥프로에서 빌드하는 것처럼 빌드 속도 빠르게 하기200819 NAVER TECH CONCERT 10_맥북에서도 아이맥프로에서 빌드하는 것처럼 빌드 속도 빠르게 하기
200819 NAVER TECH CONCERT 10_맥북에서도 아이맥프로에서 빌드하는 것처럼 빌드 속도 빠르게 하기
 
200819 NAVER TECH CONCERT 08_성능을 고민하는 슬기로운 개발자 생활
200819 NAVER TECH CONCERT 08_성능을 고민하는 슬기로운 개발자 생활200819 NAVER TECH CONCERT 08_성능을 고민하는 슬기로운 개발자 생활
200819 NAVER TECH CONCERT 08_성능을 고민하는 슬기로운 개발자 생활
 
200819 NAVER TECH CONCERT 05_모르면 손해보는 Android 디버깅/분석 꿀팁 대방출
200819 NAVER TECH CONCERT 05_모르면 손해보는 Android 디버깅/분석 꿀팁 대방출200819 NAVER TECH CONCERT 05_모르면 손해보는 Android 디버깅/분석 꿀팁 대방출
200819 NAVER TECH CONCERT 05_모르면 손해보는 Android 디버깅/분석 꿀팁 대방출
 
200819 NAVER TECH CONCERT 09_Case.xcodeproj - 좋은 동료로 거듭나기 위한 노하우
200819 NAVER TECH CONCERT 09_Case.xcodeproj - 좋은 동료로 거듭나기 위한 노하우200819 NAVER TECH CONCERT 09_Case.xcodeproj - 좋은 동료로 거듭나기 위한 노하우
200819 NAVER TECH CONCERT 09_Case.xcodeproj - 좋은 동료로 거듭나기 위한 노하우
 
200820 NAVER TECH CONCERT 14_야 너두 할 수 있어. 비전공자, COBOL 개발자를 거쳐 네이버에서 FE 개발하게 된...
200820 NAVER TECH CONCERT 14_야 너두 할 수 있어. 비전공자, COBOL 개발자를 거쳐 네이버에서 FE 개발하게 된...200820 NAVER TECH CONCERT 14_야 너두 할 수 있어. 비전공자, COBOL 개발자를 거쳐 네이버에서 FE 개발하게 된...
200820 NAVER TECH CONCERT 14_야 너두 할 수 있어. 비전공자, COBOL 개발자를 거쳐 네이버에서 FE 개발하게 된...
 
200820 NAVER TECH CONCERT 13_네이버에서 오픈 소스 개발을 통해 성장하는 방법
200820 NAVER TECH CONCERT 13_네이버에서 오픈 소스 개발을 통해 성장하는 방법200820 NAVER TECH CONCERT 13_네이버에서 오픈 소스 개발을 통해 성장하는 방법
200820 NAVER TECH CONCERT 13_네이버에서 오픈 소스 개발을 통해 성장하는 방법
 
200820 NAVER TECH CONCERT 12_상반기 네이버 인턴을 돌아보며
200820 NAVER TECH CONCERT 12_상반기 네이버 인턴을 돌아보며200820 NAVER TECH CONCERT 12_상반기 네이버 인턴을 돌아보며
200820 NAVER TECH CONCERT 12_상반기 네이버 인턴을 돌아보며
 
200820 NAVER TECH CONCERT 11_빠르게 성장하는 슈퍼루키로 거듭나기
200820 NAVER TECH CONCERT 11_빠르게 성장하는 슈퍼루키로 거듭나기200820 NAVER TECH CONCERT 11_빠르게 성장하는 슈퍼루키로 거듭나기
200820 NAVER TECH CONCERT 11_빠르게 성장하는 슈퍼루키로 거듭나기
 
200819 NAVER TECH CONCERT 07_신입 iOS 개발자 개발업무 적응기
200819 NAVER TECH CONCERT 07_신입 iOS 개발자 개발업무 적응기200819 NAVER TECH CONCERT 07_신입 iOS 개발자 개발업무 적응기
200819 NAVER TECH CONCERT 07_신입 iOS 개발자 개발업무 적응기
 

Recently uploaded

Streamlining Python Development: A Guide to a Modern Project Setup
Streamlining Python Development: A Guide to a Modern Project SetupStreamlining Python Development: A Guide to a Modern Project Setup
Streamlining Python Development: A Guide to a Modern Project SetupFlorian Wilhelm
 
Scanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL CertsScanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL CertsRizwan Syed
 
Bluetooth Controlled Car with Arduino.pdf
Bluetooth Controlled Car with Arduino.pdfBluetooth Controlled Car with Arduino.pdf
Bluetooth Controlled Car with Arduino.pdfngoud9212
 
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks..."LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...Fwdays
 
Build your next Gen AI Breakthrough - April 2024
Build your next Gen AI Breakthrough - April 2024Build your next Gen AI Breakthrough - April 2024
Build your next Gen AI Breakthrough - April 2024Neo4j
 
Connect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationConnect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationSlibray Presentation
 
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticsKotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticsAndrey Dotsenko
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersThousandEyes
 
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024BookNet Canada
 
CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):comworks
 
Science&tech:THE INFORMATION AGE STS.pdf
Science&tech:THE INFORMATION AGE STS.pdfScience&tech:THE INFORMATION AGE STS.pdf
Science&tech:THE INFORMATION AGE STS.pdfjimielynbastida
 
Unblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesUnblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesSinan KOZAK
 
My Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationMy Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationRidwan Fadjar
 
Artificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning eraArtificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning eraDeakin University
 
Advanced Test Driven-Development @ php[tek] 2024
Advanced Test Driven-Development @ php[tek] 2024Advanced Test Driven-Development @ php[tek] 2024
Advanced Test Driven-Development @ php[tek] 2024Scott Keck-Warren
 
Understanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitectureUnderstanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitecturePixlogix Infotech
 
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptxMaking_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptxnull - The Open Security Community
 

Recently uploaded (20)

Streamlining Python Development: A Guide to a Modern Project Setup
Streamlining Python Development: A Guide to a Modern Project SetupStreamlining Python Development: A Guide to a Modern Project Setup
Streamlining Python Development: A Guide to a Modern Project Setup
 
Scanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL CertsScanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL Certs
 
Bluetooth Controlled Car with Arduino.pdf
Bluetooth Controlled Car with Arduino.pdfBluetooth Controlled Car with Arduino.pdf
Bluetooth Controlled Car with Arduino.pdf
 
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks..."LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
 
Build your next Gen AI Breakthrough - April 2024
Build your next Gen AI Breakthrough - April 2024Build your next Gen AI Breakthrough - April 2024
Build your next Gen AI Breakthrough - April 2024
 
Connect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationConnect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck Presentation
 
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticsKotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
 
DMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special EditionDMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special Edition
 
Hot Sexy call girls in Panjabi Bagh 🔝 9953056974 🔝 Delhi escort Service
Hot Sexy call girls in Panjabi Bagh 🔝 9953056974 🔝 Delhi escort ServiceHot Sexy call girls in Panjabi Bagh 🔝 9953056974 🔝 Delhi escort Service
Hot Sexy call girls in Panjabi Bagh 🔝 9953056974 🔝 Delhi escort Service
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
 
E-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptx
E-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptxE-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptx
E-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptx
 
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
 
CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):
 
Science&tech:THE INFORMATION AGE STS.pdf
Science&tech:THE INFORMATION AGE STS.pdfScience&tech:THE INFORMATION AGE STS.pdf
Science&tech:THE INFORMATION AGE STS.pdf
 
Unblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesUnblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen Frames
 
My Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationMy Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 Presentation
 
Artificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning eraArtificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning era
 
Advanced Test Driven-Development @ php[tek] 2024
Advanced Test Driven-Development @ php[tek] 2024Advanced Test Driven-Development @ php[tek] 2024
Advanced Test Driven-Development @ php[tek] 2024
 
Understanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitectureUnderstanding the Laravel MVC Architecture
Understanding the Laravel MVC Architecture
 
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptxMaking_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
 

Comparative studies on detecting abusive language on twitter

  • 1. ABUSIVE LANGUAGE DETECTION— COMPARATIVE STUDY AND ITS APPLICATIONS YOUNGHUN LEE
  • 2. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS BEFORE WE BEGIN.. ▸ B.S. in Electrical and Computer Engineering at Seoul National University ▸ M.S. in Information at the University of Michigan ▸ (Current) Research Intern at Machine Intelligence Lab @ Seoul National University ▸ (Current) Ph.D. applicant in Machine Learning and Natural Language Processing 2
  • 3. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS WHAT THIS PRESENTATION IS COVERING I. Introduction to Abusive Language Detection ▸ What is it? Why is it important to study? ▸ Recent studies II. About the paper, “Comparative Studies of Detecting Abusive Language on Twitter” ▸ Research idea and its novelty ▸ Empirical results and findings III. Future relevance with Naver ▸ Research possibilities and their challenges 3
  • 4. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ACKNOWLEDGEMENTS ▸ Zeerak Waseem ▸ Ph.D. Candidate at the University of Sheffield ▸ Organizer of the Abusive Language Online Workshop 4
  • 5. I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
  • 6. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION ▸ Varying definition and terminology 6 EARLY DEFINITION Messages that most users consider to be annoying or upsetting1
  • 7. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ▸ Varying definition and terminology 7 RECENT Messages that under-rate a person or a group on the basis of characteristics (race, ethnicity, gender, religion, etc.)2 EARLY DEFINITION Messages that most users consider to be annoying or upsetting1 I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
  • 8. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ▸ Varying definition and terminology 8 EXAMPLE 1) Go fucking kill yourself useless scumbag 2) Hell yeah! Go bitches! 3) Jews are lower class pigs EARLY DEFINITION Messages that most users consider to be annoying or upsetting1 RECENT Messages that under-rate a person or a group on the basis of characteristics (race, ethnicity, gender, religion, etc.)2 I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
  • 9. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ▸ Varying definition and terminology 9 EXAMPLE 1) Go fucking kill yourself useless scumbag 2) Hell yeah! Go bitches! 3) Jews are lower class pigs EARLY DEFINITION Messages that most users consider to be annoying or upsetting1 RECENT Messages that under-rate a person or a group on the basis of characteristics (race, ethnicity, gender, religion, etc.)2 I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
  • 10. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ▸ Varying definition and terminology 10 EXAMPLE 1) Go fucking kill yourself useless scumbag 2) Hell yeah! Go bitches! 3) Jews are lower class pigs EARLY DEFINITION Messages that most users consider to be annoying or upsetting1 RECENT Messages that under-rate a person or a group on the basis of characteristics (race, ethnicity, gender, religion, etc.)2 I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
  • 11. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ▸ Varying definition and terminology 11 TERMINOLOGY abusive language hate speech offensive language cyberbullying hostile flames vulgar language insults, profanity … EARLY DEFINITION Messages that most users consider to be annoying or upsetting1 RECENT Messages that under-rate a person or a group on the basis of characteristics (race, ethnicity, gender, religion, etc.)2 I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
  • 12. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ▸ Varying definition and terminology 12 TERMINOLOGY abusive language hate speech offensive language cyberbullying hostile flames vulgar language insults, profanity … EARLY DEFINITION Messages that most users consider to be annoying or upsetting1 RECENT Messages that under-rate a person or a group on the basis of characteristics (race, ethnicity, gender, religion, etc.)2 I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION PRO) Enables considering diverse situations CON) Confusing definition -> hard to annotate
  • 13. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ▸ Why is it important these days? 13 1) Increasing usage of social media3 I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
  • 14. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ▸ Why is it important these days? 14 I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION 1) Increasing usage of social media 2) Social media significantly affects current day society
  • 15. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ▸ Why is it important these days? 15 4 I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
  • 16. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ▸ Why is it important these days? 16 5 I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION 1) Increasing usage of social media 2) Social media significantly affects current day society 3) More and more users have been experiencing online harassment
  • 17. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ▸ Why is it important these days? 17 1) Increasing usage of social media 2) Social media significantly affects current day society 3) More and more users have been experiencing online harassment 4) Yet, major social media companies fail to successfully resolve the issueArtist stencils hate speech tweets outside Twitter HQ to highlight failure to deal with offensive messages I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
  • 18. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ▸ Characteristics of Abusive Language Online ▸ Abusive messages might be influenced by…6 18 - its discourse context - its co-occurring media (images, videos) - world events - identity of the author and target I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION
  • 19. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS ▸ Characteristics of Abusive Language Online ▸ Abusive messages might be influenced by…6 ▸ What is it different from other advanced text classification tasks such as sarcasm detection and fake news detection? 19 - its discourse context - its co-occurring media (images, videos) - world events - identity of the author and target I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION - CONTEXT is its core (e.g. ‘nigger’, ‘bitch’)
  • 20. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION ▸ Recent studies ▸ Data crawling and annotation ▸ Feature-engineering models for accurate classification ▸ Using meta-information such as age, gender, location as additional features ▸ Applying deep models ▸ Concerning fairness issues ▸ Resolving unintentional bias issue (Woman, Jew being abusive?) 20
  • 21. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION ▸ Future focus ▸ Creating a gold-standard test set ▸ More reliable annotations ▸ Getting more context information ▸ Multi-modal abusive detection 21
  • 22. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS I. INTRODUCTION TO ABUSIVE LANGUAGE DETECTION ▸ Future focus ▸ Creating a gold-standard test set ▸ More reliable annotations ▸ Getting more context information ▸ Multi-modal abusive detection 22
  • 23. II. ABOUT THE PAPER, “COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER”
  • 24. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER ▸ Brief overview of the paper Comparing the accuracy (F1 scores) of different machine learning models and different features in detecting abusive language on a recently released Twitter dataset 24
  • 25. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER ▸ About the dataset, “Hate and Abusive Speech on Twitter” 7 ▸ Approximately 100K labeled tweets (other datasets: 10K to 35K) ▸ Reduced overlapping labels by calculating correlation coefficients ▸ E.g. ‘abusive’, ‘offensive’, ‘aggressive’, ‘cyberbullying’ ‘abusive’ ▸ Multi-label dataset (4 labels: abusive, hateful, spam, none) 25 Label distribution of crawled tweets
  • 26. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER ▸ Implemented models and features 26 • Naïve Bayes • Logistic Regression • Support Vector Machine • Random Forests • Gradient Boosted Trees • Word-level / Char-level FEATURE-ENGINEERING ML NEURAL NETWORKS • Conv Neural Network • Recurrent Neural Network • Word-level / Char-level • Hybrid CNN • Self-matching attention RNN • Latent Topic Clustering • Context Tweets
  • 27. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER ▸ Implemented models and features 27 NEURAL NETWORKS • Conv Neural Network • Recurrent Neural Network • Word-level / Char-level • Hybrid CNN • Self-matching attention RNN • Latent Topic Clustering • Context Tweets Architecture of Hybrid CNN8
  • 28. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER ▸ Implemented models and features 28 NEURAL NETWORKS • Conv Neural Network • Recurrent Neural Network • Word-level / Char-level • Hybrid CNN • Self-matching attention RNN • Latent Topic Clustering • Context Tweets Gated self-matching attention networks (Wang et al.,)9
  • 29. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER ▸ Implemented models and features 29 NEURAL NETWORKS • Conv Neural Network • Recurrent Neural Network • Word-level / Char-level • Hybrid CNN • Self-matching attention RNN • Latent Topic Clustering • Context Tweets Hierarchical Recurrent Dual Encoder with Latent Topic Clustering module10
  • 30. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER ▸ Implemented models and features 30 NEURAL NETWORKS • Conv Neural Network • Recurrent Neural Network • Word-level / Char-level • Hybrid CNN • Self-matching attention RNN • Latent Topic Clustering • Context Tweets WHAT ARE CONTEXT TWEETS? ‣ Looking at the tweet one has replied to or has quoted provides significant contextual information ‣ Assumption: ML models benefit from taking context tweets into account in detecting abusive language.
  • 31. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER ▸ Implemented models and features 31 NEURAL NETWORKS • Conv Neural Network • Recurrent Neural Network • Word-level / Char-level • Hybrid CNN • Self-matching attention RNN • Latent Topic Clustering • Context Tweets SAMPLE CONTEXT TWEET Who the HELL is “LIKE” ING this post? Sick people…. @user_B LABELED TWEET
  • 32. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER ▸ Implemented models and features 32 NEURAL NETWORKS • Conv Neural Network • Recurrent Neural Network • Word-level / Char-level • Hybrid CNN • Self-matching attention RNN • Latent Topic Clustering • Context Tweets SAMPLE CONTEXT TWEET Survivors of #Syria Gas Attack Recount ‘a Cruel Scene’. @user_A Who the HELL is “LIKE” ING this post? Sick people…. @user_B CONTEXT TWEET LABELED TWEET
  • 33. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER ▸ Implemented models and features 33 NEURAL NETWORKS • Conv Neural Network • Recurrent Neural Network • Word-level / Char-level • Hybrid CNN • Self-matching attention RNN • Latent Topic Clustering • Context Tweets HOW TO INTEGRATE CONTEXT TWEETS LABELED TWEET CONTEXT TWEET RNN CNN max_pooled max_pooled last_hidden last_hidden σ σ
  • 34. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER 34 Experimental results of learning models and their variants, followed by the context tweet models. The top 2 scores are marked as bold for each metric. ‣ Neural network models with word- level features are accurate in general compared to feature- engineering models
  • 35. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER 35 Experimental results of learning models and their variants, followed by the context tweet models. The top 2 scores are marked as bold for each metric. ‣ Neural network models with word- level features are accurate in general compared to feature- engineering models ‣ Char-level features benefit feature- engineering ML models while cripple neural models
  • 36. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER 36 Experimental results of learning models and their variants, followed by the context tweet models. The top 2 scores are marked as bold for each metric. ‣ Neural network models with word- level features are accurate in general compared to feature- engineering models ‣ Char-level features benefit feature- engineering ML models while cripple neural models ‣ Context data doesn’t improve the overall accuracy, however, it is more effective in classifying lower- distributed labels
  • 37. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER 37 ▸ Discussion BASELINE CONTEXT p_normal .1614 p_abusive .5958 p_normal .3709 p_abusive .2582 LABEL: abusive 8 dudes added me in a group chat telling me I should kill myself, lmfaooo I think I reached a whole new level of twitter. @user_A They hatin cause you a damn queen who doesn’t answer to anybody. Let em be mad. Keep rolling mamas. @user_B ‣ Only baseline model was correct, but is ‘abusive’ label for this tweet accurate?
  • 38. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER 38 ▸ Discussion BASELINE CONTEXT LABEL: normal Oklahoma home invasion shooting: No charges against man who killed 3 intruders. @user_C You have every right to protect your family and home from worthless thieves. @user_D ‣ Context tweets gave more information to accurately classify the tweet p_normal .3287 p_abusive .3718 p_normal .8104 p_abusive .0962
  • 39. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS II. COMPARATIVE STUDIES OF DETECTING ABUSIVE LANGUAGE ON TWITTER 39 ▸ Discussion ▸ How to efficiently train the classifier for imbalanced datasets? ▸ How can we better incorporate context tweets into the original data? How to handle ‘no_context’? ▸ Can we annotate the dataset regarding context data? Specifically, can we make annotators consider the content of its context tweet when deciding abusiveness of the tweet? ▸ What are other applications that could benefit from using context data?
  • 41. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS III. FUTURE RELEVANCE WITH NAVER ▸ Hate Speech in Korea ▸ 남성혐오 / 여성혐오 (워마드, 일베저장소) ▸ 극단적 진보 / 보수 정치 성향 ▸ 사회적 문제로 확장 41
  • 42. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS III. FUTURE RELEVANCE WITH NAVER ▸ Possible Contribution ▸ DATASET!!! 42
  • 43. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS III. FUTURE RELEVANCE WITH NAVER 43
  • 44. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS III. FUTURE RELEVANCE WITH NAVER 44
  • 45. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS III. FUTURE RELEVANCE WITH NAVER 45 RICH TEXTUAL & CONTEXTUAL INFORMATION FROM ONE NEWS ARTICLE 1) Article, title, and its topic 2) Comments and comments of comments 2) Emotions on the article 3) Number of comments, demographic distribution of commenters 4) Upvotes and downvotes of comments 5) User meta-information
  • 46. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS III. FUTURE RELEVANCE WITH NAVER ▸ Possible applications ▸ Detecting abusive language in news article comments ▸ Enabling personalized settings for not encountering toxic messages ▸ CLOVA: read comments of news articles that are not abusive ▸ Challenges ▸ Korean text: not extensively studied compared to English ▸ Lack of user meta-information 46
  • 47. / 48 ABUSIVE LANGUAGE DETECTION—COMPARATIVE STUDY AND ITS APPLICATIONS REFERENCE 1. Ellen Spertus. 1997. Smokey: Automatic recognition of hostile messages. In AAAI/IAAI, pages 1058– 1065. 2. John T. Nockleby. 2000. Hate Speech. In Leonard W. Levy, Kenneth L. Karst, and Dennis J. Mahoney, editors, Encyclopedia of the American Constitution, pages 1277–1279. Macmillan, 2nd edition. 3. Aaron Smith and Monica Anderson. 2018. Social Media Use in 2018. Pew Research Center; accessed 8-December-2018. 4. Monica Anderson and Skye Toor. How social media users have discussed sexual harassment since #MeToo went viral. Pew Research Center; accessed 8-December-2018. 5. Maeve Duggan. 2017. Online harassment 2017. Pew Research Center; accessed 8-December-2018. 6. Anna Schmidt and Michael Wiegand. 2017. A survey on hate speech detection using natural language pro- cessing. In Proceedings of the Fifth International Workshop on Natural Language Processing for So- cial Media, pages 1–10. 7. Antigoni Founta, Constantinos Djouvas, Despoina Chatzakou, Ilias Leontiadis, Jeremy Blackburn, Gianluca Stringhini, Athena Vakali, Michael Sirivianos, and Nicolas Kourtellis. 2018. Large scale crowdsourcing and characterization of twitter abusive behavior. In Proceedings of the International AAAI Conference on Web and Social Media. 8. Ji Ho Park and Pascale Fung. 2017. One-step and two- step classification for abusive language detection on twitter. In Proceedings of the First Workshop on Abusive Language Online, pages 41–45. 9. Wenhui Wang, Nan Yang, Furu Wei, Baobao Chang, and Ming Zhou. 2017. Gated self-matching networks for reading comprehension and question answering. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, volume 1, pages 189–198. 10.Seunghyun Yoon, Joongbo Shin, and Kyomin Jung. 2018. Learning to rank question-answer pairs using hierarchical recurrent encoder with latent topic clustering. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, volume 1, pages 1575–1584. 47