Better Prevent than React: Deep Stratified Learning to Predict Hate Intensity of Twitter Reply Chains

Sahnan, Dhruv; Dahiya, Snehil; Goel, Vasu; Bandhakavi, Anil; Chakraborty, Tanmoy

doi:10.1109/icdm51629.2021.00066

Cited by 5 publications

(10 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Lin et al [38] classified the tweets into different levels of hatred and suggested the HEAR model track posts that are likely to cause hate speech. Closer to our work is DRAGNET [58], which is a deep stratified learning framework that predicts the hate intensity of a conversation thread based on what a root tweet can fetch through its subsequent replies. DRAGNET models the linear sequence for the conversation thread chain where the tweets are arranged chronologically.…”

Section: Hate Speech Detection In Social Mediamentioning

confidence: 99%

“…From the modeling standpoint, DRAGNET++ has advanced DRAGNET [58] on two aspects: DRAGNET++ exploits the tweet-level semantics information and the conversation-level structural information to improve hate intensity prediction. DRAGNET had leveraged sentiment features, calculated as each tweet's similarity in the conversation threads to the root tweet.…”

Section: Comparison With Dragnetmentioning

confidence: 99%

“…Lin et al [38] manually categorize Twitter conversation threads into different level of hatred, and proposed a deep learning model to forecast and classify tweets into the pre-defined hatred levels. Recently, we Sahnan et al [58] introduced the hate intensity prediction task and proposed DRAGNET, a deep stratified learning framework, which predicted the intensity of hatred that a root tweet can fetch through its subsequent replies. Nevertheless, DRAGNET assumes a linear sequence for the conversation thread in its model, in which tweets in a conversation threads are arranged in a chronological sequence.…”

Section: Introductionmentioning

confidence: 99%

“…Research Objectives. In this paper, we aim to fill this research gap and extend our earlier model, DRAGNET [58] by proposing DRAGNET++ that models multifaceted information to forecast the hate intensity of a Twitter conversation thread. Specifically, we define our problem statement as follows -given a root tweet and a few of its initial replies, can we predict the hate intensity of the subsequent replies of the tweet?…”

Section: Introductionmentioning

confidence: 99%

“…To model this problem, we first quantify hate intensities of tweets in the conversation thread and express them as a series of hate intensities, which essentially transforms the task into a time-series problem. At a high level, DRAGNET++ adopts a similar stratified learning framework as Manuscript submitted to ACM suggested in [58] to model the hate intensity profiles of Twitter conversation threads and categorizes them into clusters of varying hate intensity. In addition to the content sentiment and temporal information, DRAGNET++ also captures structural information of the conversation thread.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Predicting Hate Intensity of Twitter Conversation Threads

Meng¹,

Suresh²,

Lee³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Tweets are the most concise form of communication in online social media, wherein a single tweet has the potential to make or break the discourse of the conversation. Online hate speech is more accessible than ever, and stifling its propagation is of utmost importance for social media companies and users for congenial communication. Most of the research barring a recent few has focused on classifying an individual tweet regardless of the tweet thread/context leading up to that point. One of the classical approaches to curb hate speech is to adopt a reactive strategy after the hate speech postage. The ex-post facto strategy results in neglecting subtle posts that do not show the potential to instigate hate speech on their own but may portend in the subsequent discussion ensuing in the post's replies.In this paper, we propose DRAGNET++, which aims to predict the intensity of hatred that a tweet can bring in through its reply chain in the future. It uses the semantic and propagating structure of the tweet threads to maximize the contextual information leading up to and the fall of hate intensity at each subsequent tweet. We explore three publicly available Twitter datasets -Anti-Racism contains the reply tweets of a collection of social media discourse on racist remarks during US political and Covid-19 background;Anti-Social presents a dataset of 40 million tweets amidst the COVID-19 pandemic on anti-social behaviours with custom annotations; and Anti-Asian presents Twitter datasets collated based on anti-Asian behaviours during COVID-19 pandemic. All the curated datasets consist of structural graph information of the Tweet threads. We show that DRAGNET++ outperforms all the state-of-the-art baselines significantly. It beats the best baseline by an 11% margin on the Person correlation coefficient and a decrease of 25% on RMSE for the Anti-Racism dataset with a similar performance on the other two datasets.

show abstract

Section: Hate Speech Detection In Social Mediamentioning

confidence: 99%

Section: Comparison With Dragnetmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Predicting Hate Intensity of Twitter Conversation Threads

Meng¹,

Suresh²,

Lee³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Multi-step Online Hate Speech Detection and Classification Using Sentiment and Sarcasm Features

Maste,

Prabhu,

Shetty

et al. 2023

Lecture Notes in Networks and Systems

View full text Add to dashboard Cite

Proactively Reducing the Hate Intensity of Online Posts via Hate Speech Normalization

Masud

Bedi

Khan

et al. 2022

Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining

View full text Add to dashboard Cite

Although pre-trained large language models (PLMs) have achieved state-of-the-art on many NLP tasks, they lack understanding of subtle expressions of implicit hate speech. Such nuanced and implicit hate is often misclassified as non-hate. Various attempts have been made to enhance the detection of (implicit) hate content by augmenting external context or enforcing label separation via distance-based metrics. We combine these two approaches and introduce FiADD, a novel Focused Inferential Adaptive Density Discrimination framework. FiADD enhances the PLM finetuning pipeline by bringing the surface form of an implicit hate speech closer to its implied form while increasing the inter-cluster distance among various class labels. We test FiADD on three implicit hate datasets and observe significant improvement in the two-way and three-way hate classification tasks. We further experiment on the generalizability of FiADD on three other tasks, namely detecting sarcasm, irony, and stance, in which surface and implied forms differ, and observe similar performance improvement. We analyze the generated latent space to understand its evolution under FiADD, which corroborates the advantage of employing FiADD for implicit hate speech detection. 1

show abstract

Better Prevent than React: Deep Stratified Learning to Predict Hate Intensity of Twitter Reply Chains

Cited by 5 publications

References 24 publications

Predicting Hate Intensity of Twitter Conversation Threads

Predicting Hate Intensity of Twitter Conversation Threads

Multi-step Online Hate Speech Detection and Classification Using Sentiment and Sarcasm Features

Proactively Reducing the Hate Intensity of Online Posts via Hate Speech Normalization

Contact Info

Product

Resources

About