MetaTroll: Few-shot Detection of State-Sponsored Trolls with Transformer Adapters

State-sponsored trolls are the main actors of influence campaigns on social media and automatic troll detection is important to combat misinformation at scale. Existing troll detection models are developed based on training data for known campaigns (e.g.\ the influence campaign by Russia's Inte...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2023-03
Hauptverfasser: Lin, Tian, Zhang, Xiuzhen, Jey Han Lau
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Lin, Tian
Zhang, Xiuzhen
Jey Han Lau
description State-sponsored trolls are the main actors of influence campaigns on social media and automatic troll detection is important to combat misinformation at scale. Existing troll detection models are developed based on training data for known campaigns (e.g.\ the influence campaign by Russia's Internet Research Agency on the 2016 US Election), and they fall short when dealing with {\em novel} campaigns with new targets. We propose MetaTroll, a text-based troll detection model based on the meta-learning framework that enables high portability and parameter-efficient adaptation to new campaigns using only a handful of labelled samples for few-shot transfer. We introduce \textit{campaign-specific} transformer adapters to MetaTroll to ``memorise'' campaign-specific knowledge so as to tackle catastrophic forgetting, where a model ``forgets'' how to detect trolls from older campaigns due to continual adaptation. Our experiments demonstrate that MetaTroll substantially outperforms baselines and state-of-the-art few-shot text classification models. Lastly, we explore simple approaches to extend MetaTroll to multilingual and multimodal detection. Source code for MetaTroll is available at: https://github.com/ltian678/metatroll-code.git.
doi_str_mv 10.48550/arxiv.2303.07354
format Article
fullrecord <record><control><sourceid>proquest_arxiv</sourceid><recordid>TN_cdi_arxiv_primary_2303_07354</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2787015463</sourcerecordid><originalsourceid>FETCH-LOGICAL-a523-1ba57802acb807c3434cd7234a4aef3ec7bd6bb0371cac9b765d6c2977439a5f3</originalsourceid><addsrcrecordid>eNotj1FLwzAUhYMgOOZ-gE8GfG5Nc5Om9W1Mp8JEZH0vt2nKOrqmJpnTf2_tfDrn4eNwPkJuEhaLTEp2j-67_Yo5MIiZAikuyIwDJFEmOL8iC-_3jDGeKi4lzMjHmwlYONt1D3RtTpHf2UAfTTA6tLantqHbgMFE28H23jpT0wn29NSG3dix9411B-PossYhGOevyWWDnTeL_5yTYv1UrF6izfvz62q5iVByiJIKpcoYR11lTGkQIHStOAgUaBowWlV1WlUMVKJR55VKZZ1qnislIEfZwJzcnmcn33Jw7QHdT_nnXU7eI3F3JgZnP4_Gh3Jvj64fP5VcZYolUqQAv9qIW0M</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2787015463</pqid></control><display><type>article</type><title>MetaTroll: Few-shot Detection of State-Sponsored Trolls with Transformer Adapters</title><source>arXiv.org</source><source>Free E- Journals</source><creator>Lin, Tian ; Zhang, Xiuzhen ; Jey Han Lau</creator><creatorcontrib>Lin, Tian ; Zhang, Xiuzhen ; Jey Han Lau</creatorcontrib><description>State-sponsored trolls are the main actors of influence campaigns on social media and automatic troll detection is important to combat misinformation at scale. Existing troll detection models are developed based on training data for known campaigns (e.g.\ the influence campaign by Russia's Internet Research Agency on the 2016 US Election), and they fall short when dealing with {\em novel} campaigns with new targets. We propose MetaTroll, a text-based troll detection model based on the meta-learning framework that enables high portability and parameter-efficient adaptation to new campaigns using only a handful of labelled samples for few-shot transfer. We introduce \textit{campaign-specific} transformer adapters to MetaTroll to ``memorise'' campaign-specific knowledge so as to tackle catastrophic forgetting, where a model ``forgets'' how to detect trolls from older campaigns due to continual adaptation. Our experiments demonstrate that MetaTroll substantially outperforms baselines and state-of-the-art few-shot text classification models. Lastly, we explore simple approaches to extend MetaTroll to multilingual and multimodal detection. Source code for MetaTroll is available at: https://github.com/ltian678/metatroll-code.git.</description><identifier>EISSN: 2331-8422</identifier><identifier>DOI: 10.48550/arxiv.2303.07354</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Adaptation ; Adapters ; Computer Science - Computation and Language ; Elections ; Source code ; Transformers</subject><ispartof>arXiv.org, 2023-03</ispartof><rights>2023. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><rights>http://creativecommons.org/licenses/by/4.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,780,881,27902</link.rule.ids><backlink>$$Uhttps://doi.org/10.1145/3543507.3583417$$DView published paper (Access to full text may be restricted)$$Hfree_for_read</backlink><backlink>$$Uhttps://doi.org/10.48550/arXiv.2303.07354$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Lin, Tian</creatorcontrib><creatorcontrib>Zhang, Xiuzhen</creatorcontrib><creatorcontrib>Jey Han Lau</creatorcontrib><title>MetaTroll: Few-shot Detection of State-Sponsored Trolls with Transformer Adapters</title><title>arXiv.org</title><description>State-sponsored trolls are the main actors of influence campaigns on social media and automatic troll detection is important to combat misinformation at scale. Existing troll detection models are developed based on training data for known campaigns (e.g.\ the influence campaign by Russia's Internet Research Agency on the 2016 US Election), and they fall short when dealing with {\em novel} campaigns with new targets. We propose MetaTroll, a text-based troll detection model based on the meta-learning framework that enables high portability and parameter-efficient adaptation to new campaigns using only a handful of labelled samples for few-shot transfer. We introduce \textit{campaign-specific} transformer adapters to MetaTroll to ``memorise'' campaign-specific knowledge so as to tackle catastrophic forgetting, where a model ``forgets'' how to detect trolls from older campaigns due to continual adaptation. Our experiments demonstrate that MetaTroll substantially outperforms baselines and state-of-the-art few-shot text classification models. Lastly, we explore simple approaches to extend MetaTroll to multilingual and multimodal detection. Source code for MetaTroll is available at: https://github.com/ltian678/metatroll-code.git.</description><subject>Adaptation</subject><subject>Adapters</subject><subject>Computer Science - Computation and Language</subject><subject>Elections</subject><subject>Source code</subject><subject>Transformers</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>BENPR</sourceid><sourceid>GOX</sourceid><recordid>eNotj1FLwzAUhYMgOOZ-gE8GfG5Nc5Om9W1Mp8JEZH0vt2nKOrqmJpnTf2_tfDrn4eNwPkJuEhaLTEp2j-67_Yo5MIiZAikuyIwDJFEmOL8iC-_3jDGeKi4lzMjHmwlYONt1D3RtTpHf2UAfTTA6tLantqHbgMFE28H23jpT0wn29NSG3dix9411B-PossYhGOevyWWDnTeL_5yTYv1UrF6izfvz62q5iVByiJIKpcoYR11lTGkQIHStOAgUaBowWlV1WlUMVKJR55VKZZ1qnislIEfZwJzcnmcn33Jw7QHdT_nnXU7eI3F3JgZnP4_Gh3Jvj64fP5VcZYolUqQAv9qIW0M</recordid><startdate>20230313</startdate><enddate>20230313</enddate><creator>Lin, Tian</creator><creator>Zhang, Xiuzhen</creator><creator>Jey Han Lau</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20230313</creationdate><title>MetaTroll: Few-shot Detection of State-Sponsored Trolls with Transformer Adapters</title><author>Lin, Tian ; Zhang, Xiuzhen ; Jey Han Lau</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a523-1ba57802acb807c3434cd7234a4aef3ec7bd6bb0371cac9b765d6c2977439a5f3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Adaptation</topic><topic>Adapters</topic><topic>Computer Science - Computation and Language</topic><topic>Elections</topic><topic>Source code</topic><topic>Transformers</topic><toplevel>online_resources</toplevel><creatorcontrib>Lin, Tian</creatorcontrib><creatorcontrib>Zhang, Xiuzhen</creatorcontrib><creatorcontrib>Jey Han Lau</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni Edition)</collection><collection>ProQuest Central UK/Ireland</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central Korea</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection><collection>arXiv Computer Science</collection><collection>arXiv.org</collection><jtitle>arXiv.org</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Lin, Tian</au><au>Zhang, Xiuzhen</au><au>Jey Han Lau</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>MetaTroll: Few-shot Detection of State-Sponsored Trolls with Transformer Adapters</atitle><jtitle>arXiv.org</jtitle><date>2023-03-13</date><risdate>2023</risdate><eissn>2331-8422</eissn><abstract>State-sponsored trolls are the main actors of influence campaigns on social media and automatic troll detection is important to combat misinformation at scale. Existing troll detection models are developed based on training data for known campaigns (e.g.\ the influence campaign by Russia's Internet Research Agency on the 2016 US Election), and they fall short when dealing with {\em novel} campaigns with new targets. We propose MetaTroll, a text-based troll detection model based on the meta-learning framework that enables high portability and parameter-efficient adaptation to new campaigns using only a handful of labelled samples for few-shot transfer. We introduce \textit{campaign-specific} transformer adapters to MetaTroll to ``memorise'' campaign-specific knowledge so as to tackle catastrophic forgetting, where a model ``forgets'' how to detect trolls from older campaigns due to continual adaptation. Our experiments demonstrate that MetaTroll substantially outperforms baselines and state-of-the-art few-shot text classification models. Lastly, we explore simple approaches to extend MetaTroll to multilingual and multimodal detection. Source code for MetaTroll is available at: https://github.com/ltian678/metatroll-code.git.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><doi>10.48550/arxiv.2303.07354</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2023-03
issn 2331-8422
language eng
recordid cdi_arxiv_primary_2303_07354
source arXiv.org; Free E- Journals
subjects Adaptation
Adapters
Computer Science - Computation and Language
Elections
Source code
Transformers
title MetaTroll: Few-shot Detection of State-Sponsored Trolls with Transformer Adapters
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-31T13%3A58%3A36IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_arxiv&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=MetaTroll:%20Few-shot%20Detection%20of%20State-Sponsored%20Trolls%20with%20Transformer%20Adapters&rft.jtitle=arXiv.org&rft.au=Lin,%20Tian&rft.date=2023-03-13&rft.eissn=2331-8422&rft_id=info:doi/10.48550/arxiv.2303.07354&rft_dat=%3Cproquest_arxiv%3E2787015463%3C/proquest_arxiv%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2787015463&rft_id=info:pmid/&rfr_iscdi=true