PROVOKE: Toxicity trigger detection in conversations from the top 100 subreddits

<p>Promoting healthy discourse on community-based online platforms like Reddit can be challenging, especially when conversations show ominous signs of toxicity. Therefore, in this study, we find the turning points (i.e., toxicity triggers) making conversations toxic. Before finding toxicity tr...

وصف كامل

محفوظ في:
التفاصيل البيبلوغرافية
المؤلف الرئيسي: Hind Almerekhi (7434776) (author)
مؤلفون آخرون: Haewoon Kwak (5747558) (author), Joni Salminen (7434770) (author), Bernard J. Jansen (7434779) (author)
منشور في: 2022
الموضوعات:
الوسوم: إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
_version_ 1864513518708981760
author Hind Almerekhi (7434776)
author2 Haewoon Kwak (5747558)
Joni Salminen (7434770)
Bernard J. Jansen (7434779)
author2_role author
author
author
author_facet Hind Almerekhi (7434776)
Haewoon Kwak (5747558)
Joni Salminen (7434770)
Bernard J. Jansen (7434779)
author_role author
dc.creator.none.fl_str_mv Hind Almerekhi (7434776)
Haewoon Kwak (5747558)
Joni Salminen (7434770)
Bernard J. Jansen (7434779)
dc.date.none.fl_str_mv 2022-10-01T00:00:00Z
dc.identifier.none.fl_str_mv 10.1016/j.dim.2022.100019
dc.relation.none.fl_str_mv https://figshare.com/articles/journal_contribution/PROVOKE_Toxicity_trigger_detection_in_conversations_from_the_top_100_subreddits/25662672
dc.rights.none.fl_str_mv CC BY 4.0
info:eu-repo/semantics/openAccess
dc.subject.none.fl_str_mv Information and computing sciences
Library and information studies
Online toxicity
Conversation threads
Reddit
Toxicity triggers
Neural networks
Social media
dc.title.none.fl_str_mv PROVOKE: Toxicity trigger detection in conversations from the top 100 subreddits
dc.type.none.fl_str_mv Text
Journal contribution
info:eu-repo/semantics/publishedVersion
text
contribution to journal
description <p>Promoting healthy discourse on community-based online platforms like Reddit can be challenging, especially when conversations show ominous signs of toxicity. Therefore, in this study, we find the turning points (i.e., toxicity triggers) making conversations toxic. Before finding toxicity triggers, we built and evaluated various machine learning models to detect toxicity from Reddit comments. Subsequently, we used our best-performing model, a fine-tuned Bidirectional Encoder Representations from Transformers (BERT) model that achieved an area under the receiver operating characteristic curve (AUC) score of 0.983 to detect toxicity. Next, we constructed conversation threads and used the toxicity prediction results to build a training set for detecting toxicity triggers. This procedure entailed using our large-scale dataset to refine toxicity triggers' definition and build a trigger detection dataset using 991,806 conversation threads from the top 100 communities on Reddit. Then, we extracted a set of sentiment shift, topical shift, and context-based features from the trigger detection dataset, using them to build a dual embedding biLSTM neural network that achieved an AUC score of 0.789. Our trigger detection dataset analysis showed that specific triggering keywords are common across all communities, like ‘racist’ and ‘women’. In contrast, other triggering keywords are specific to certain communities, like ‘overwatch’ in r/Games. Implications are that toxicity trigger detection algorithms can leverage generic approaches but must also tailor detections to specific communities.</p><h2>Other Information</h2> <p> Published in: Data and Information Management<br> License: <a href="http://creativecommons.org/licenses/by/4.0/" target="_blank">http://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1016/j.dim.2022.100019" target="_blank">https://dx.doi.org/10.1016/j.dim.2022.100019</a></p>
eu_rights_str_mv openAccess
id Manara2_a3596ecaafaf328d64bdd3d0e2895c4d
identifier_str_mv 10.1016/j.dim.2022.100019
network_acronym_str Manara2
network_name_str Manara2
oai_identifier_str oai:figshare.com:article/25662672
publishDate 2022
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling PROVOKE: Toxicity trigger detection in conversations from the top 100 subredditsHind Almerekhi (7434776)Haewoon Kwak (5747558)Joni Salminen (7434770)Bernard J. Jansen (7434779)Information and computing sciencesLibrary and information studiesOnline toxicityConversation threadsRedditToxicity triggersNeural networksSocial media<p>Promoting healthy discourse on community-based online platforms like Reddit can be challenging, especially when conversations show ominous signs of toxicity. Therefore, in this study, we find the turning points (i.e., toxicity triggers) making conversations toxic. Before finding toxicity triggers, we built and evaluated various machine learning models to detect toxicity from Reddit comments. Subsequently, we used our best-performing model, a fine-tuned Bidirectional Encoder Representations from Transformers (BERT) model that achieved an area under the receiver operating characteristic curve (AUC) score of 0.983 to detect toxicity. Next, we constructed conversation threads and used the toxicity prediction results to build a training set for detecting toxicity triggers. This procedure entailed using our large-scale dataset to refine toxicity triggers' definition and build a trigger detection dataset using 991,806 conversation threads from the top 100 communities on Reddit. Then, we extracted a set of sentiment shift, topical shift, and context-based features from the trigger detection dataset, using them to build a dual embedding biLSTM neural network that achieved an AUC score of 0.789. Our trigger detection dataset analysis showed that specific triggering keywords are common across all communities, like ‘racist’ and ‘women’. In contrast, other triggering keywords are specific to certain communities, like ‘overwatch’ in r/Games. Implications are that toxicity trigger detection algorithms can leverage generic approaches but must also tailor detections to specific communities.</p><h2>Other Information</h2> <p> Published in: Data and Information Management<br> License: <a href="http://creativecommons.org/licenses/by/4.0/" target="_blank">http://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1016/j.dim.2022.100019" target="_blank">https://dx.doi.org/10.1016/j.dim.2022.100019</a></p>2022-10-01T00:00:00ZTextJournal contributioninfo:eu-repo/semantics/publishedVersiontextcontribution to journal10.1016/j.dim.2022.100019https://figshare.com/articles/journal_contribution/PROVOKE_Toxicity_trigger_detection_in_conversations_from_the_top_100_subreddits/25662672CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/256626722022-10-01T00:00:00Z
spellingShingle PROVOKE: Toxicity trigger detection in conversations from the top 100 subreddits
Hind Almerekhi (7434776)
Information and computing sciences
Library and information studies
Online toxicity
Conversation threads
Reddit
Toxicity triggers
Neural networks
Social media
status_str publishedVersion
title PROVOKE: Toxicity trigger detection in conversations from the top 100 subreddits
title_full PROVOKE: Toxicity trigger detection in conversations from the top 100 subreddits
title_fullStr PROVOKE: Toxicity trigger detection in conversations from the top 100 subreddits
title_full_unstemmed PROVOKE: Toxicity trigger detection in conversations from the top 100 subreddits
title_short PROVOKE: Toxicity trigger detection in conversations from the top 100 subreddits
title_sort PROVOKE: Toxicity trigger detection in conversations from the top 100 subreddits
topic Information and computing sciences
Library and information studies
Online toxicity
Conversation threads
Reddit
Toxicity triggers
Neural networks
Social media