Hate speech detection is not as easy as you may think: A closer look at model validation (extended version)

Arango, Aymé; Pérez, Jorge; Poblete, Bárbara

doi:10.1016/j.is.2020.101584

Cited by 98 publications

(176 citation statements)

References 11 publications

Supporting

Mentioning

171

Contrasting

Order By: Relevance

“…Besides English (Basile et al, 2019;Waseem and Hovy, 2016;Davidson et al, 2017;Founta et al, 2018;Qian et al, 2018), we notice a growing interest in the study of hate speech in other languages, such as Portuguese (Fortuna et al, 2019), Italian (Sanguinetti et al, 2018), German (Ross et al, 2016), Indonesian (Ibrohim and Budi, 2019), French (Ousidhoum et al, 2019), Dutch (Hee et al, 2015, and Arabic (Albadi et al, 2018;Mulki et al, 2019;Ousidhoum et al, 2019). Challenging questions being tackled in this area involve the way abusive language spreads online (Mathew et al, 2019), fast changing topics during data collection (Liu et al, 2019), user bias in publicly available datasets (Arango et al, 2019), bias in hate speech classification and different methods to reduce it (Park et al, 2018;Davidson et al, 2019;Kennedy et al, 2020).…”

Section: Related Workmentioning

confidence: 99%

“…Moreover, the account-based dataset of (Mulki et al, 2019), referred to as AR3 in Figures 3 and 4 shows more robustness towards keywords. Nevertheless, such a collection strategy may generate a linguistic bias that goes with the same stylistic features used by the targeted accounts, similarly to Waseem and Hovy (2016)'s user bias reported by Arango et al (2019).…”

Section: Robustness Of Keyword-based Selectionmentioning

confidence: 99%

“…Despite its scarcity in comparison to normal web content, Mathew et al (2019) demonstrated that it tends to reach large audiences faster due to dense connections between users who share such content. Hence, a search based on generic hate speech keywords or controversial hashtags may result in a set of social media posts generated by a limited number of users (Arango et al, 2019). This would lead to an inherent bias in hate speech datasets similar to other tasks involving social data (Olteanu et al, 2019) as opposed to a selection bias (Heckman, 1977) particular to hate speech data.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Comparative Evaluation of Label-Agnostic Selection Bias in Multilingual Hate Speech Datasets

Ousidhoum¹,

Song²,

Yeung³

2020

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)

View full text Add to dashboard Cite

Work on bias in hate speech typically aims to improve classification performance while relatively overlooking the quality of the data. We examine selection bias in hate speech in a language and label independent fashion. We first use topic models to discover latent semantics in eleven hate speech corpora, then, we present two bias evaluation metrics based on the semantic similarity between topics and search words frequently used to build corpora. We discuss the possibility of revising the data collection process by comparing datasets and analyzing contrastive case studies.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Robustness Of Keyword-based Selectionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Comparative Evaluation of Label-Agnostic Selection Bias in Multilingual Hate Speech Datasets

Ousidhoum¹,

Song²,

Yeung³

2020

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)

View full text Add to dashboard Cite

show abstract

“…Waseem [46] wrote about the annotator's influence on hate speech detection, showing expert annotations contributed to a better detection rate. Arango et al [1] analyzed a model validation problem of hate speech detection.…”

Section: Related Work 21 Hate Speech Detectionmentioning

confidence: 99%

“…of online social networks, hate speech is spreading faster and affecting a larger population than before in human history across the world 1 . Therefore, quickly and accurately identifying hate speech becomes crucial for keeping a harmonic and healthy online social environment, mitigating the possible conflicts, and protecting the diversity of our society.…”

mentioning

confidence: 99%

SWE2: SubWord Enriched and Significant Word Emphasized Framework for Hate Speech Detection

Mou

Lee

2020

Proceedings of the 29th ACM International Conference on Information &Amp; Knowledge Management

View full text Add to dashboard Cite

Hate speech detection on online social networks has become one of the emerging hot topics in recent years. With the broad spread and fast propagation speed across online social networks, hate speech makes significant impacts on society by increasing prejudice and hurting people. Therefore, there are aroused attention and concern from both industry and academia. In this paper, we address the hate speech problem and propose a novel hate speech detection framework called SWE2, which only relies on the content of messages and automatically identifies hate speech. In particular, our framework exploits both word-level semantic information and sub-word knowledge. It is intuitively persuasive and also practically performs well under a situation with/without character-level adversarial attack. Experimental results show that our proposed model achieves 0.975 accuracy and 0.953 macro F1, outperforming 7 state-of-the-art baselines under no adversarial attack. Our model robustly and significantly performed well under extreme adversarial attack (manipulation of 50% messages), achieving 0.967 accuracy and 0.934 macro F1. CCS CONCEPTS • Information systems → Information retrieval; • Computing methodologies → Natural language processing.

show abstract

A taxonomy, data set, and benchmark for detecting and classifying malevolent dialogue responses

Zhang

Ren

Rijke

2021

Asso for Info Science & Tech

View full text Add to dashboard Cite

Conversational interfaces are increasingly popular as a way of connecting people to information. With the increased generative capacity of corpus‐based conversational agents comes the need to classify and filter out malevolent responses that are inappropriate in terms of content and dialogue acts. Previous studies on the topic of detecting and classifying inappropriate content are mostly focused on a specific category of malevolence or on single sentences instead of an entire dialogue. We make three contributions to advance research on the malevolent dialogue response detection and classification (MDRDC) task. First, we define the task and present a hierarchical malevolent dialogue taxonomy. Second, we create a labeled multiturn dialogue data set and formulate the MDRDC task as a hierarchical classification task. Last, we apply state‐of‐the‐art text classification methods to the MDRDC task, and report on experiments aimed at assessing the performance of these approaches.

show abstract

Hate speech detection is not as easy as you may think: A closer look at model validation (extended version)

Cited by 98 publications

References 11 publications

Comparative Evaluation of Label-Agnostic Selection Bias in Multilingual Hate Speech Datasets

Comparative Evaluation of Label-Agnostic Selection Bias in Multilingual Hate Speech Datasets

SWE2: SubWord Enriched and Significant Word Emphasized Framework for Hate Speech Detection

A taxonomy, data set, and benchmark for detecting and classifying malevolent dialogue responses

Contact Info

Product

Resources

About