pith. sign in

arxiv: 1907.08873 · v1 · pith:D2I33HOPnew · submitted 2019-07-20 · 💻 cs.SI · cs.CY· cs.IR

Detecting Cyberbullying and Cyberaggression in Social Media

Despoina Chatzakou , Ilias Leontiadis , Jeremy Blackburn , Emiliano De Cristofaro , Gianluca Stringhini , Athena Vakali , Nicolas Kourtellis This is my paper

Pith reviewed 2026-05-24 18:24 UTC · model grok-4.3

classification 💻 cs.SI cs.CYcs.IR
keywords cyberbullyingcyberaggressionTwittermachine learning classificationsocial media abusenetwork analysisuser behavior
0
0 comments X

The pith

Text, user, and network attributes allow machine learning to separate bullies and aggressors from ordinary Twitter users with over 90 percent accuracy and AUC.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper sets out to build a detection system for cyberbullying and cyberaggression by studying large numbers of Twitter accounts. It contrasts users who join ordinary conversations, such as those about the NBA, with users active in controversial topics like Gamergate or BBC gender pay disputes, then narrows to one of those communities to examine specific abusive patterns. Features drawn from tweet text, account properties, and interaction networks feed into standard classifiers that label accounts as abusive or normal. If the approach holds, platforms gain a practical way to surface accounts that drive prolonged harassment affecting large numbers of users.

Core claim

The authors show that a methodology combining text-based, user-based, and network-based attributes, processed by several machine learning algorithms, can classify Twitter accounts as bullies, aggressors, or normal users at over 90 percent accuracy and AUC when the training data come from participants in hate-related discussions.

What carries the argument

The classification pipeline that fuses tweet text, account metadata, and social graph attributes to train supervised models distinguishing abusive from non-abusive accounts.

If this is right

  • Twitter could apply the same feature set to flag accounts for manual review before suspension.
  • The method separates cyberbullying from cyberaggression within the same community.
  • Performance of different suspension policies can be simulated on the labeled set.
  • Normal-topic users provide a baseline that highlights what changes when abuse appears.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same attribute combination might transfer to other platforms if their data allow extraction of comparable text, profile, and link features.
  • Future work could test whether adding victim-reported incidents improves label quality over topic-based proxies.
  • If network features prove dominant, early detection could occur before many abusive tweets are posted.

Load-bearing premise

Participation in discussions around hate-related topics serves as a reliable proxy for labeling users as bullies or aggressors without independent verification of their actual behavior.

What would settle it

Independent human raters label a held-out sample of the same accounts using only the visible tweets and then compare agreement with the model's output.

Figures

Figures reproduced from arXiv: 1907.08873 by Athena Vakali, Despoina Chatzakou, Emiliano De Cristofaro, Gianluca Stringhini, Ilias Leontiadis, Jeremy Blackburn, Nicolas Kourtellis.

Figure 1
Figure 1. Figure 1: Similarity distribution of duplicate posts across the datasets. duplications of posts are examined to detect the cutoff-limit above which a user will be characterized as spammer and con￾sequently will be removed from the datasets. Hashtags. Studying the hashtags distribution, we observe that users use on average 0 to 17 hashtags. Building on this, we examine various cutoffs to select a proper one above whi… view at source ↗
Figure 2
Figure 2. Figure 2: CCDF plots (log-log scale) of the Baseline, Gamergate, BBCpay, and NBA datasets. that models each document as a mixture of latent topics, where a topic is described by a distribution over words. The topic extraction was made based on the JSAT [77], i.e., a Java sta￾tistical analysis tool. The tool provides an implementation of LDA which is based on the Stochastic Variational Inference. To run the LDA model… view at source ↗
Figure 3
Figure 3. Figure 3: CDF distribution for various user profile features: (a) Account age, (b) Number of posts, (c) Hashtags, (d) Favorites, (e) Urls, and (f) Mentions. Metric Gamergate BBCpay NBA Baseline Account age (days) 982.94 / 788 / 772.49 1,043 / 865 / 913.04 996.58 / 780 / 837.07 834.39 / 522 / 652.42 Tweets 135,618 / 48,587 / 185,997 236,972 / 50,405 / 408,088 82,134 / 28,241 / 126,206 49,342 / 9,429 / 97,457 Hashtags… view at source ↗
Figure 4
Figure 4. Figure 4: CDF distribution of (a) Number of followers and (b) friends. a difference of 1-2 URLs, D = 0.27), while Gamergate users post more in an attempt to disseminate information about their “cause,” somewhat using Twitter like a news service. The use of urls on users posts shows the existence of a similar pattern with the number of used hashtags from the four different user categories with the users of the NBA co… view at source ↗
Figure 5
Figure 5. Figure 5: CDF distribution of (a) Sentiment, (b) Joy, and (c) Uppercases. are apparently less happy. The BBCpay dataset seems to con￾tain the less joyful users which can be justified by the fact that such a controversy has created a lot of frustration and disap￾pointment to the BBC female, and not only, community. The difference with the other three user categories is statistical sig￾nificant (D = 0.04 with baseline… view at source ↗
Figure 6
Figure 6. Figure 6: Overview of our sessionization process for constructing two consecutive sessions. In each session, the interarrival time between tweets does not exceed a predefined time threshold tl. sive, offensive, and sarcastic). Finally, the users of the NBA community seem to be very popular and with long activity on Twitter, something which is reasonable considering the popularity of the specific sport around the wor… view at source ↗
Figure 7
Figure 7. Figure 7: Example of the crowdsourcing user interface. tuition is that increasing the batch size provides more context to the workers to assess if a poster is acting in an aggressive or bullying behavior, however, too many tweets might confuse them. The best results with respect to labeling agreement – i.e., the number of workers that provide the same label for a batch – occur with 5-10 tweets per batch. Therefore, … view at source ↗
Figure 8
Figure 8. Figure 8: CDF of (a) Adjectives, (b) Adverbs, (c) Nouns, (d) Verbs, (e) Average words per sentece, (f) Average word length. from considering the followers and friends of each user in our dataset, we further extended the network by considering the contacts (followers and friends) of the followers/friends of the initial users. This way, we were able to expand the net￾work construction beyond the ego-network of each us… view at source ↗
Figure 9
Figure 9. Figure 9: Boxplots of (a) Followers, (b) Reciprocity, (c) Hubs, and (d) Eigenvectors. The data are divided into three quartiles (i.e., first, third, and median quartiles in the data set). The top and the bottom whiskers indicate the maximum and minimum values, respectively. We removed outliers from the plots. tional and behavioral state of victims depend on the power of their bullies, e.g., more negative emotional e… view at source ↗
Figure 10
Figure 10. Figure 10: Overview of the neural network setup for classification of abuse on Twitter. with random subsets of features during the classification pro￾cess. So, an important advantage of the Random Forest clas￾sifier is its ability in reducing overfitting by averaging sev￾eral trees during the model construction process. Additionally, Random Forests are quite efficient in terms of the time they need for training a mo… view at source ↗
Figure 11
Figure 11. Figure 11: CDF plots for the active, suspended, and deleted users for the: (a) Account age, (b) Followers, (c) Posts, (d) Lists, (e) Favorites, (f) Sentiment, (g) Adjectives, (h) Hashtags. active deleted suspended Baseline 65.71% 25.86% 8.43% Gamergate 71.86% 16.22% 11.29% NBA 78.61% 9.14% 12.25% BBCpay 79.79% 10.17% 10.05% [PITH_FULL_IMAGE:figures/full_fig_p025_11.png] view at source ↗
read the original abstract

Cyberbullying and cyberaggression are increasingly worrisome phenomena affecting people across all demographics. More than half of young social media users worldwide have been exposed to such prolonged and/or coordinated digital harassment. Victims can experience a wide range of emotions, with negative consequences such as embarrassment, depression, isolation from other community members, which embed the risk to lead to even more critical consequences, such as suicide attempts. In this work, we take the first concrete steps to understand the characteristics of abusive behavior in Twitter, one of today's largest social media platforms. We analyze 1.2 million users and 2.1 million tweets, comparing users participating in discussions around seemingly normal topics like the NBA, to those more likely to be hate-related, such as the Gamergate controversy, or the gender pay inequality at the BBC station. We also explore specific manifestations of abusive behavior, i.e., cyberbullying and cyberaggression, in one of the hate-related communities (Gamergate). We present a robust methodology to distinguish bullies and aggressors from normal Twitter users by considering text, user, and network-based attributes. Using various state-of-the-art machine learning algorithms, we classify these accounts with over 90% accuracy and AUC. Finally, we discuss the current status of Twitter user accounts marked as abusive by our methodology, and study the performance of potential mechanisms that can be used by Twitter to suspend users in the future.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper claims to analyze 1.2 million Twitter users and 2.1 million tweets from normal topics (NBA) versus hate-related topics (Gamergate, BBC gender pay), labeling the latter as proxies for bullies/aggressors. It presents a supervised ML methodology using text, user, and network features to classify these accounts with >90% accuracy and AUC, explores manifestations of abuse in Gamergate, and discusses potential Twitter suspension mechanisms.

Significance. If the proxy labeling reliably identifies abusive behavior rather than topic-specific patterns, the multi-feature classifier at this scale could support practical platform moderation tools. The combination of feature types and the focus on both bullying and aggression are potential strengths, but the lack of validation for the labeling limits the result's immediate significance and generalizability.

major comments (2)
  1. [Abstract] Abstract: the central claim of >90% accuracy and AUC for distinguishing bullies/aggressors is unsupported because the abstract (and visible evidence) supplies no information on the labeling procedure, feature definitions, cross-validation strategy, class imbalance handling, or baseline comparisons.
  2. [Abstract and methods] Data collection and labeling (implied in abstract and methods): assigning positive labels to users participating in Gamergate/BBC discussions as a proxy for cyberbullying/cyberaggression without independent ground-truth verification, manual annotation, or external validation is load-bearing for the supervised classification result; this risks the model learning topic-specific signals instead of abuse signals.
minor comments (2)
  1. [Abstract] Abstract: dataset sizes are stated without breakdown by topic, class distribution, or how the 2.1M tweets relate to the 1.2M users.
  2. [Discussion] The discussion of Twitter suspension mechanisms is mentioned but lacks quantitative evaluation or comparison to existing platform policies.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments. We address each major point below and propose targeted revisions where appropriate.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim of >90% accuracy and AUC for distinguishing bullies/aggressors is unsupported because the abstract (and visible evidence) supplies no information on the labeling procedure, feature definitions, cross-validation strategy, class imbalance handling, or baseline comparisons.

    Authors: The abstract is intentionally concise per journal guidelines and summarizes the key result; full details on labeling (topic-based proxy), feature sets (text/user/network), 10-fold cross-validation, class weighting for imbalance, and baseline comparisons (e.g., against text-only models) appear in Sections 3 and 4. We will revise the abstract to include one additional sentence outlining the multi-feature supervised approach and evaluation protocol. revision: partial

  2. Referee: [Abstract and methods] Data collection and labeling (implied in abstract and methods): assigning positive labels to users participating in Gamergate/BBC discussions as a proxy for cyberbullying/cyberaggression without independent ground-truth verification, manual annotation, or external validation is load-bearing for the supervised classification result; this risks the model learning topic-specific signals instead of abuse signals.

    Authors: We selected Gamergate and BBC gender-pay topics precisely because they are documented in prior literature as containing elevated rates of abusive behavior, providing a scalable proxy when manual annotation of 1.2 M accounts is infeasible. Network and user features were included alongside text to reduce reliance on topic vocabulary alone; we further validate the proxy by manually examining abuse manifestations within the Gamergate subset. We will add an explicit limitations paragraph discussing the proxy assumption and outlining how future work could obtain platform-provided labels for external validation. revision: partial

Circularity Check

0 steps flagged

No significant circularity; standard empirical ML pipeline

full rationale

The paper collects tweets from topic-based cohorts (Gamergate/BBC as positive labels, NBA as negative), extracts text/user/network features, and trains standard supervised classifiers to report accuracy/AUC. No equations, parameter fits, or self-citations are shown that reduce the classification performance to a definition, a renamed input, or a load-bearing prior result by the same authors. The derivation chain consists of independent data collection, feature engineering, and off-the-shelf ML, making the reported results self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claim rests on standard machine-learning assumptions about feature predictiveness and data representativeness rather than new mathematical derivations.

free parameters (1)
  • ML model hyperparameters
    State-of-the-art algorithms require tuning; values not reported in abstract.
axioms (2)
  • domain assumption Participation in Gamergate or BBC gender-pay discussions serves as a valid proxy label for cyberbullying or cyberaggression.
    Used to construct the abusive class for supervised learning.
  • domain assumption Text, user, and network attributes are sufficiently discriminative for the classification task.
    Foundation for the feature-based methodology.

pith-pipeline@v0.9.0 · 5816 in / 1431 out tokens · 32638 ms · 2026-05-24T18:24:06.930444+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

107 extracted references · 107 canonical work pages · 3 internal anchors

  1. [1]

    List of swear words & curse words, 2017

    AllSlang. List of swear words & curse words, 2017. https: //www.noswearing.com/dictionary

    work page 2017

  2. [2]

    A. A. Amleshwaram, N. Reddy, S. Yadav, G. Gu, and C. Yang. Cats: Characterizing automation of twitter spammers. In 2013 Fifth International Conference on Communication Sys- tems and Networks (COMSNETS), pages 1–10, Jan 2013

    work page 2013

  3. [3]

    I dated Zoe Quinn

    Anonymous. I dated Zoe Quinn. 4chan. https://archive.is/ qrS5Q

    work page

  4. [4]

    Zoe Quinn, prominent SJW and indie developer is a liar and a slut

    Anonymous. Zoe Quinn, prominent SJW and indie developer is a liar and a slut. 4chan. https://archive.is/QIjm3

    work page

  5. [5]

    Neural Machine Translation by Jointly Learning to Align and Translate

    D. Bahdanau, K. Cho, and Y . Bengio. Neural machine trans- lation by jointly learning to align and translate. arXiv preprint arXiv:1409.0473, 2014

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  6. [6]

    Bergsma, M

    S. Bergsma, M. Post, and D. Yarowsky. Stylometric analysis of scientific articles. In Proceedings of the 2012 Conference of the North American Chapter of the Association for Com- putational Linguistics: Human Language Technologies, pages 327–337. Association for Computational Linguistics, 2012

    work page 2012

  7. [7]

    Bhattacharya and S

    D. Bhattacharya and S. Ram. Rt news: An analysis of news agency ego networks in a microblogging environment. ACM Transactions on Management Information Systems, 6(3):11:1– 11:25, 2015

    work page 2015

  8. [8]

    Blackburn, R

    J. Blackburn, R. Simha, N. Kourtellis, X. Zuo, M. Ripeanu, J. Skvoretz, and A. Iamnitchi. Branded with a scarlet ”c”: cheaters in a gaming social network. In WWW, 2012

    work page 2012

  9. [9]

    D. M. Blei, A. Y . Ng, and M. I. Jordan. Latent dirichlet alloca- tion. Journal of machine Learning research, 3(Jan):993–1022, 2003

    work page 2003

  10. [10]

    Blondel, J

    V . Blondel, J. Guillaume, R. Lambiotte, and E. Lefebvre. The Louvain method for community detection in large networks. Statistical Mechanics: Theory and Experiment, 10, 2011

    work page 2011

  11. [11]

    Bruns and S

    A. Bruns and S. Stieglitz. Towards more systematic twitter analysis: metrics for tweeting activities. International Journal of Social Research Methodology, 16(2):91–108, 2013

    work page 2013

  12. [12]

    Chatzakou, N

    D. Chatzakou, N. Kourtellis, J. Blackburn, E. D. Cristofaro, G. Stringhini, and A. Vakali. Mean birds: Detecting aggression and bullying on twitter. In WebSci, 2017

    work page 2017

  13. [13]

    Chatzakou, N

    D. Chatzakou, N. Kourtellis, J. Blackburn, E. De Cristofaro, G. Stringhini, and A. Vakali. Hate is not Binary: Studying Abusive Behavior of #GamerGate on Twitter. In ACM Hyper- text, 2017

    work page 2017

  14. [14]

    Chatzakou, N

    D. Chatzakou, N. Kourtellis, J. Blackburn, E. De Cristofaro, G. Stringhini, and A. Vakali. Mean birds: Detecting aggression and bullying on twitter. In Proceedings of the 2017 ACM on web science conference, pages 13–22. ACM, 2017

    work page 2017

  15. [15]

    Chatzakou, N

    D. Chatzakou, N. Kourtellis, J. Blackburn, E. De Cristofaro, G. Stringhini, and A. Vakali. Measuring #GamerGate: A Tale of Hate, Sexism, and Bullying. In WWW CyberSafety Work- shop, 2017

    work page 2017

  16. [16]

    Chatzakou, V

    D. Chatzakou, V . Koutsonikola, A. Vakali, and K. Kafet- sios. Micro-blogging Content Analysis via Emotionally- Driven Clustering. In ACII, 2013

    work page 2013

  17. [17]

    Chatzakou, N

    D. Chatzakou, N. Passalis, and A. Vakali. Multispot: Spotting sentiments with semantic aware multilevel cascaded analysis. In DaWaK, volume 9263, pages 337–350. Springer, 2015

    work page 2015

  18. [18]

    Chatzakou and A

    D. Chatzakou and A. Vakali. Harvesting opinions and emo- tions from social media textual resources.Internet Computing, IEEE, 19(4):46–50, 2015

    work page 2015

  19. [19]

    Chatzakou, A

    D. Chatzakou, A. Vakali, and K. Kafetsios. Detecting variation of emotions in online activities. Expert Systems with Applica- tions, 89:318 – 332, 2017

    work page 2017

  20. [20]

    C. Chen, J. Zhang, X. Chen, Y . Xiang, and W. Zhou. 6 million spam tweets: A large ground truth for timely Twitter spam detection. In IEEE ICC, 2015

    work page 2015

  21. [21]

    Y . Chen, Y . Zhou, S. Zhu, and H. Xu. Detecting Offensive Lan- guage in Social Media to Protect Adolescent Online Safety. In PASSAT and SocialCom, 2012

    work page 2012

  22. [22]

    Corcoran, C

    L. Corcoran, C. M. Guckin, and G. Prentice. Cyberbullying or cyber aggression?: A review of existing definitions of cyber- based peer-to-peer aggression. Societies, 5(2), 2015

    work page 2015

  23. [23]

    https://cyberbullying.org/ summary-of-our-cyberbullying-research, November 2016

    Cyberbullying Research Center. https://cyberbullying.org/ summary-of-our-cyberbullying-research, November 2016

    work page 2016

  24. [24]

    https://cyberbullying.org/ facts, 2017

    Cyberbullying Research Center. https://cyberbullying.org/ facts, 2017

    work page 2017

  25. [25]

    Dadvar, D

    M. Dadvar, D. Trieschnigg, and F. Jong. Experts and machines against bullies: A hybrid approach to detect cyberbullies. In Canadian AI, 2014

    work page 2014

  26. [26]

    Davis and M

    J. Davis and M. Goadrich. The relationship between Precision- Recall and ROC curves. In Machine learning, 2006

    work page 2006

  27. [27]

    T. G. Dietterich. Ensemble Methods in Machine Learning. In Proceedings of the First International Workshop on Multiple Classifier Systems, 2000

    work page 2000

  28. [28]

    Dinakar, R

    K. Dinakar, R. Reichart, and H. Lieberman. Modeling the de- tection of Textual Cyberbullying. The Social Mobile Web, 11, 2011

    work page 2011

  29. [29]

    Djuric, J

    N. Djuric, J. Zhou, R. Morris, M. Grbovic, V . Radosavljevic, and N. Bhamidipati. Hate Speech Detection with Comment Embeddings. In WWW, 2015

    work page 2015

  30. [30]

    Ekman, W

    P. Ekman, W. V . Friesen, and P. Ellsworth. What emotion cat- egories or dimensions can observers judge from facial behav- ior? Emotion in the human face, 1982

    work page 1982

  31. [31]

    Fetterly, M

    D. Fetterly, M. Manasse, and M. Najork. On the evolution of clusters of near-duplicate web pages. volume 2, pages 228–

    work page

  32. [32]

    Institute of Electrical and Electronics Engineers, Inc., Oc- tober 2004

    work page 2004

  33. [33]

    Fetterly, M

    D. Fetterly, M. Manasse, and M. Najork. Detecting phrase- level duplication on the world wide web. In28th Annual Inter- national ACM SIGIR Conference on Research and Develop- ment in Information Retrieval (SIGIR) , Salvador, Brazil, Au- gust 2005. Association for Computing Machinery, Inc

    work page 2005

  34. [34]

    https://www.figure-eight.com/, 2019

    Figure Eight. https://www.figure-eight.com/, 2019

    work page 2019

  35. [35]

    Fox and W

    J. Fox and W. Y . Tang. Sexism in online video games: The role of conformity to masculine norms and social dominance orientation . Computers in Human Behavior, 33, 2014

    work page 2014

  36. [36]

    Friedman, D

    N. Friedman, D. Geiger, and M. Goldszmidt. Bayesian Net- work Classifiers. Mach. Learn., 29(2-3), 1997. 31

    work page 1997

  37. [37]

    Giatsoglou, D

    M. Giatsoglou, D. Chatzakou, N. Shah, C. Faloutsos, and A. Vakali. Reteeting Activity on Twitter: Signs of Deception. In PAKDD, 2015

    work page 2015

  38. [38]

    D. W. Grigg. Cyber-aggression: Definition and concept of cy- berbullying. Australian Journal of Guidance and Counselling, 20(02), 2010

    work page 2010

  39. [39]

    Guardian

    T. Guardian. Gary Lineker is BBC’s best- paid star and only one not to take pay cut. https://www.theguardian.com/world/2018/jul/11/ gary-lineker-bbc-best-paid-star-only-one-not-to-take-pay-cut, 2018

    work page 2018

  40. [40]

    Guberman and L

    J. Guberman and L. Hemphill. Challenges in Modifying Ex- isting Scales for Detecting Harassment in Individual Tweets. In System Sciences, 2017

    work page 2017

  41. [41]

    L. D. Hanish, B. Kochenderfer-Ladd, R. A. Fabes, C. L. Mar- tin, D. Denning, et al. Bullying among young children: The in- fluence of peers and teachers. Bullying in American schools: A social-ecological perspective on prevention and intervention , 2004

    work page 2004

  42. [42]

    Hastie, S

    T. Hastie, S. Rosset, J. Zhu, and H. Zou. Multi-class adaboost. Statistics and its Interface, 2(3):349–360, 2009

    work page 2009

  43. [43]

    https://www.hatebase.org/, 2017

    Hatebase database. https://www.hatebase.org/, 2017

    work page 2017

  44. [44]

    K. He, X. Zhang, S. Ren, and J. Sun. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition , pages 770–778, 2016

    work page 2016

  45. [45]

    A. Hern. Feminist critics of video games facing threats in ‘gamergate’ campaign. The Guardian, Oct

    work page

  46. [46]

    https://www.theguardian.com/technology/2014/oct/23/ felicia-days-public-details-online-gamergate

    work page 2014

  47. [47]

    Hidasi, M

    B. Hidasi, M. Quadrana, A. Karatzoglou, and D. Tikk. Parallel recurrent neural network architectures for feature-rich session- based recommendations. In Proceedings of the 10th ACM Conference on Recommender Systems, pages 241–248. ACM, 2016

    work page 2016

  48. [48]

    G. E. Hine, J. Onaolapo, E. De Cristofaro, N. Kourtellis, I. Leontiadis, R. Samaras, G. Stringhini, and J. Blackburn. A longitudinal measurement study of 4chan’s politically in- correct forum and its effect on the web. arXiv preprint arXiv:1610.03452, 2016

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  49. [49]

    Hosseinmardi, R

    H. Hosseinmardi, R. Han, Q. Lv, S. Mishra, and A. Ghasemi- anlangroodi. Towards understanding cyberbullying behavior in a semi-anonymous social network. In IEEE/ACM ASONAM, 2014

    work page 2014

  50. [50]

    Hosseinmardi, S

    H. Hosseinmardi, S. A. Mattson, R. I. Rafiq, R. Han, Q. Lv, and S. Mishra. Analyzing Labeled Cyberbullying Incidents on the Instagram Social Network. In In SocInfo, 2015

    work page 2015

  51. [51]

    F. Jin, E. Dougherty, P. Saraf, Y . Cao, and N. Ramakrishnan. Epidemiological Modeling of News and Rumors on Twitter. In SNAKDD, 2013

    work page 2013

  52. [52]

    J.-H. K. Estimating Classification Error Rate: Repeated Cross- validation, Repeated Hold-out and Bootstrap. Comput. Stat. Data Anal., 53(11), 2009

    work page 2009

  53. [53]

    Kayes, N

    I. Kayes, N. Kourtellis, D. Quercia, A. Iamnitchi, and F. Bonchi. The Social World of Content Abusers in Commu- nity Question Answering. In WWW, 2015

    work page 2015

  54. [54]

    https://keras.io/, 2017

    Keras. https://keras.io/, 2017

    work page 2017

  55. [55]

    A. Z. Khan, M. Atique, and V . Thakare. Combining lexicon- based and learning-based methods for twitter sentiment analy- sis. International Journal of Electronics, Communication and Soft Computing Science & Engineering (IJECSCSE), page 89, 2015

    work page 2015

  56. [56]

    Kira and L

    K. Kira and L. A. Rendell. A Practical Approach to Feature Selection. In 9th International Workshop on Machine Learn- ing, 1992

    work page 1992

  57. [57]

    J. M. Kleinberg. Hubs, Authorities, and Communities. ACM Computing Surveys, 31(4es), 1999

    work page 1999

  58. [58]

    Twitter says it’s punishing 10 times more users for being abusive than it was a year ago

    Kurt Wagner. Twitter says it’s punishing 10 times more users for being abusive than it was a year ago. https://www.vox.com/2017/7/20/15999636/ twitter-safety-abuse-update-suspensions-increase, Jul 2017

    work page 2017

  59. [59]

    H. Kwak, J. Blackburn, and S. Han. Exploring Cyberbully- ing and Other Toxic Behavior in Team Competition Online Games. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 2015

    work page 2015

  60. [60]

    K. Lee, J. Caverlee, and S. Webb. Uncovering social spam- mers: Social honeypots + machine learning. In Proceedings of the 33rd International ACM SIGIR Conference on Research and Development in Information Retrieval , SIGIR ’10, pages 435–442, New York, NY , USA, 2010. ACM

    work page 2010

  61. [61]

    Massanari

    A. Massanari. #Gamergate and The Fappening: How Reddit’s algorithm, governance, and culture support toxic technocul- tures. New Media & Society, 2015

    work page 2015

  62. [62]

    McCord and M

    M. McCord and M. Chuah. Spam detection on twitter using traditional classifiers. In Autonomic and Trusted Computing , pages 175–186. Springer Berlin Heidelberg, 2011

    work page 2011

  63. [63]

    Efficient Estimation of Word Representations in Vector Space

    T. Mikolov, K. Chen, G. Corrado, and J. Dean. Efficient Es- timation of Word Representations in Vector Space. CoRR, abs/1301.3781, 2013

    work page internal anchor Pith review Pith/arXiv arXiv 2013

  64. [64]

    M. Miller. goo.gl/n1W6nt, Oct 2016

    work page 2016

  65. [65]

    T. E. Mortensen. Anger, Fear, and Games. Games and Culture, 2016

    work page 2016

  66. [66]

    Nahar, S

    V . Nahar, S. Unankard, X. Li, and C. Pang. Sentiment Analysis for Effective Detection of Cyber Bullying. In APWeb, 2012

    work page 2012

  67. [67]

    G. Navarro. A Guided Tour to Approximate String Matching. ACM Computing Surveys, 33(1), 2001

    work page 2001

  68. [68]

    Nilizadeh, F

    S. Nilizadeh, F. Labr `eche, A. Sedighian, A. Zand, J. Fernan- dez, C. Kruegel, G. Stringhini, and G. Vigna. Poised: Spotting twitter spam off the beaten paths. In Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security (CCS), 2017

    work page 2017

  69. [69]

    Nobata, J

    C. Nobata, J. Tetreault, A. Thomas, Y . Mehdad, and Y . Chang. Abusive Language Detection in Online User Content. In WWW, 2016

    work page 2016

  70. [70]

    O’Sullivan

    D. O’Sullivan. Bomb suspect threatened people on twitter, and twitter didn’t act. https://edition.cnn.com/2018/10/26/tech/ cesar-sayoc-twitter-response/index.html, Oct 2018

    work page 2018

  71. [71]

    Pennington, R

    J. Pennington, R. Socher, and C. D. Manning. Glove: Global vectors for word representation. In Empirical Methods in Nat- ural Language Processing (EMNLP), pages 1532–1543, 2014

    work page 2014

  72. [72]

    http://www.pewinternet.org/2017/07/ 11/online-harassment-2017/, 2014

    Pew Research Center. http://www.pewinternet.org/2017/07/ 11/online-harassment-2017/, 2014

    work page 2017

  73. [73]

    Pfeffer, T

    J. Pfeffer, T. Zorbach, and K. M. Carley. Understanding online firestorms: Negative word-of-mouth dynamics in social me- dia networks. Journal of Marketing Communications, 20(1-2), 2014

    work page 2014

  74. [74]

    Twitter tries new measures in crackdown on harassment

    Pham, Sherisse. Twitter tries new measures in crackdown on harassment. CNNtech, February

    work page

  75. [75]

    https://money.cnn.com/2017/02/07/technology/ twitter-combat-harassment-features/

    work page 2017

  76. [76]

    Pieschl, T

    S. Pieschl, T. Porsch, T. Kahl, and R. Klockenbusch. Relevant dimensions of cyberbullying - Results from two experimental studies . Journal of Applied Developmental Psychology, 34(5), 2013

    work page 2013

  77. [77]

    Plutchik

    R. Plutchik. A general psychoevolutionary theory of emotion. Theories of emotion, 1:3–31, 1980. 32

    work page 1980

  78. [78]

    A. PRESS. https://www.dailymail.co.uk/wires/ap/article- 3419263/venezuela-doctors-worried-official-silence-zika.htm. https://www.dailymail.co.uk/wires/ap/article-3419263/ Venezuela-doctors-worried-official-silence-Zika.htm, 2016

    work page 2016

  79. [79]

    J. Quinlan. Induction of Decision Trees. Machine Learning, 1(1), 1986

    work page 1986

  80. [80]

    E. Raff. Jsat: Java statistical analysis tool, a library for machine learning. Journal of Machine Learning Research , 18(23):1–5, 2017

    work page 2017

Showing first 80 references.