A Neighborhood Framework for Resource-Lean Content Flagging
<p dir="ltr">We propose a novel framework for cross- lingual content flagging with limited target- language data, which significantly outperforms prior work in terms of predictive performance. The framework is based on a nearest-neighbor architecture. It is a modern instantiation of...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | , , , , |
| Published: |
2022
|
| Subjects: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1864513506161721344 |
|---|---|
| author | Sheikh Muhammad Sarwar (19517701) |
| author2 | Dimitrina Zlatkova (19517704) Momchil Hardalov (18618397) Yoan Dinkov (19517707) Isabelle Augenstein (14013962) Preslav Nakov (17760905) |
| author2_role | author author author author author |
| author_facet | Sheikh Muhammad Sarwar (19517701) Dimitrina Zlatkova (19517704) Momchil Hardalov (18618397) Yoan Dinkov (19517707) Isabelle Augenstein (14013962) Preslav Nakov (17760905) |
| author_role | author |
| dc.creator.none.fl_str_mv | Sheikh Muhammad Sarwar (19517701) Dimitrina Zlatkova (19517704) Momchil Hardalov (18618397) Yoan Dinkov (19517707) Isabelle Augenstein (14013962) Preslav Nakov (17760905) |
| dc.date.none.fl_str_mv | 2022-05-04T03:00:00Z |
| dc.identifier.none.fl_str_mv | 10.1162/tacl_a_00472 |
| dc.relation.none.fl_str_mv | https://figshare.com/articles/journal_contribution/A_Neighborhood_Framework_for_Resource-Lean_Content_Flagging/26889445 |
| 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 Artificial intelligence Computer vision and multimedia computation Language, communication and culture Language studies Linguistics Cross-Lingual Content Flagging Limited Target-Language Data Encoding Schemes Abusive Language Detection Jigsaw Multilingual Dataset Model Adaptation Neighborhood-Based Approaches |
| dc.title.none.fl_str_mv | A Neighborhood Framework for Resource-Lean Content Flagging |
| dc.type.none.fl_str_mv | Text Journal contribution info:eu-repo/semantics/publishedVersion text contribution to journal |
| description | <p dir="ltr">We propose a novel framework for cross- lingual content flagging with limited target- language data, which significantly outperforms prior work in terms of predictive performance. The framework is based on a nearest-neighbor architecture. It is a modern instantiation of the vanilla k-nearest neighbor model, as we use Transformer representations in all its components. Our framework can adapt to new source- language instances, without the need to be retrained from scratch. Unlike prior work on neighborhood-based approaches, we encode the neighborhood information based on query– neighbor interactions. We propose two encoding schemes and we show their effectiveness using both qualitative and quantitative analysis. Our evaluation results on eight languages from two different datasets for abusive language detection show sizable improvements of up to 9.5 F1 points absolute (for Italian) over strong baselines. On average, we achieve 3.6 absolute F1 points of improvement for the three languages in the Jigsaw Multilingual dataset and 2.14 points for the WUL dataset.</p><h2>Other Information</h2><p dir="ltr">Published in: Transactions of the Association for Computational Linguistics<br>License: <a href="https://creativecommons.org/licenses/by/4.0/" target="_blank">https://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1162/tacl_a_00472" target="_blank">https://dx.doi.org/10.1162/tacl_a_00472</a></p> |
| eu_rights_str_mv | openAccess |
| id | Manara2_9530be59367bf37599664e76cfaa0063 |
| identifier_str_mv | 10.1162/tacl_a_00472 |
| network_acronym_str | Manara2 |
| network_name_str | Manara2 |
| oai_identifier_str | oai:figshare.com:article/26889445 |
| publishDate | 2022 |
| repository.mail.fl_str_mv | |
| repository.name.fl_str_mv | |
| repository_id_str | |
| rights_invalid_str_mv | CC BY 4.0 |
| spelling | A Neighborhood Framework for Resource-Lean Content FlaggingSheikh Muhammad Sarwar (19517701)Dimitrina Zlatkova (19517704)Momchil Hardalov (18618397)Yoan Dinkov (19517707)Isabelle Augenstein (14013962)Preslav Nakov (17760905)Information and computing sciencesArtificial intelligenceComputer vision and multimedia computationLanguage, communication and cultureLanguage studiesLinguisticsCross-Lingual Content FlaggingLimited Target-Language DataEncoding SchemesAbusive Language DetectionJigsaw Multilingual DatasetModel AdaptationNeighborhood-Based Approaches<p dir="ltr">We propose a novel framework for cross- lingual content flagging with limited target- language data, which significantly outperforms prior work in terms of predictive performance. The framework is based on a nearest-neighbor architecture. It is a modern instantiation of the vanilla k-nearest neighbor model, as we use Transformer representations in all its components. Our framework can adapt to new source- language instances, without the need to be retrained from scratch. Unlike prior work on neighborhood-based approaches, we encode the neighborhood information based on query– neighbor interactions. We propose two encoding schemes and we show their effectiveness using both qualitative and quantitative analysis. Our evaluation results on eight languages from two different datasets for abusive language detection show sizable improvements of up to 9.5 F1 points absolute (for Italian) over strong baselines. On average, we achieve 3.6 absolute F1 points of improvement for the three languages in the Jigsaw Multilingual dataset and 2.14 points for the WUL dataset.</p><h2>Other Information</h2><p dir="ltr">Published in: Transactions of the Association for Computational Linguistics<br>License: <a href="https://creativecommons.org/licenses/by/4.0/" target="_blank">https://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1162/tacl_a_00472" target="_blank">https://dx.doi.org/10.1162/tacl_a_00472</a></p>2022-05-04T03:00:00ZTextJournal contributioninfo:eu-repo/semantics/publishedVersiontextcontribution to journal10.1162/tacl_a_00472https://figshare.com/articles/journal_contribution/A_Neighborhood_Framework_for_Resource-Lean_Content_Flagging/26889445CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/268894452022-05-04T03:00:00Z |
| spellingShingle | A Neighborhood Framework for Resource-Lean Content Flagging Sheikh Muhammad Sarwar (19517701) Information and computing sciences Artificial intelligence Computer vision and multimedia computation Language, communication and culture Language studies Linguistics Cross-Lingual Content Flagging Limited Target-Language Data Encoding Schemes Abusive Language Detection Jigsaw Multilingual Dataset Model Adaptation Neighborhood-Based Approaches |
| status_str | publishedVersion |
| title | A Neighborhood Framework for Resource-Lean Content Flagging |
| title_full | A Neighborhood Framework for Resource-Lean Content Flagging |
| title_fullStr | A Neighborhood Framework for Resource-Lean Content Flagging |
| title_full_unstemmed | A Neighborhood Framework for Resource-Lean Content Flagging |
| title_short | A Neighborhood Framework for Resource-Lean Content Flagging |
| title_sort | A Neighborhood Framework for Resource-Lean Content Flagging |
| topic | Information and computing sciences Artificial intelligence Computer vision and multimedia computation Language, communication and culture Language studies Linguistics Cross-Lingual Content Flagging Limited Target-Language Data Encoding Schemes Abusive Language Detection Jigsaw Multilingual Dataset Model Adaptation Neighborhood-Based Approaches |