{"id":{"repo_id":"birmingham","oai_identifier":"oai:etheses.bham.ac.uk:11673"},"canonical_url":"https://search.ndltd.org/etd/birmingham/oai:etheses.bham.ac.uk:11673","repository":{"repo_id":"birmingham","name":"University of Birmingham","base_url":"https://etheses.bham.ac.uk/cgi/oai2"},"display":{"title":"Multistage feature-assisted deep learning and its application in fine-grained fake news detection","abstract":"The rapid increase of real-time news posted in social media has led to the emergence of fake news. Assessing the veracity of news claims requires an enormous labour of human fact-checkers and therefore automating the sub-tasks of fake news detection pipeline could help them identify false claims. Although the related literature addresses fake news detection tasks in a simple binary or a multiclass classification setting, challenges still remain. For instance, (1) this domain suffers from a lack of large scale datasets and a large proportion of the instances belongs to legitimate news which creates a class-imbalance problem, (2) the characteristics of fake news are not yet known in order to generate effective discriminative features (3) and the content of multiclass categories can be very similar which makes it hard for multiclass classifiers to capture the finer distinctions between them. The major focus of this thesis is to investigate novel models in Natural Language Processing (NLP) and Machine Learning (ML) that can help classify the veracity of a claim with respect to textual evidence into multiclass categories. Our first contribution is related to boosting the performance of multiclass stance detection. We show that using a feature-assisted neural model, aided with augmented training samples to deal with data imbalance, provides state-of-the-art performance on the FNC-1 dataset. The second contribution explores a way to improve stance detection, especially the minority categories, by proposing multistage classification approaches. We show a significant performance increase by breaking down the multiclass categories into different sub-stage feature-based and feature-assisted neural classifiers with category-specific features. Inspired by the multistage classification approaches, the final contribution proposes five-stage and three stage feature-assisted neural classifiers into multiclass fake statement detection. We conclude that sub-dividing the fine-grained task into multiple feature-specific classifier provides state-of-the-art performance.","abstract_html":"The rapid increase of real-time news posted in social media has led to the emergence of fake news. Assessing the veracity of news claims requires an enormous labour of human fact-checkers and therefore automating the sub-tasks of fake news detection pipeline could help them identify false claims. Although the related literature addresses fake news detection tasks in a simple binary or a multiclass classification setting, challenges still remain. For instance, (1) this domain suffers from a lack of large scale datasets and a large proportion of the instances belongs to legitimate news which creates a class-imbalance problem, (2) the characteristics of fake news are not yet known in order to generate effective discriminative features (3) and the content of multiclass categories can be very similar which makes it hard for multiclass classifiers to capture the finer distinctions between them. The major focus of this thesis is to investigate novel models in Natural Language Processing (NLP) and Machine Learning (ML) that can help classify the veracity of a claim with respect to textual evidence into multiclass categories. Our first contribution is related to boosting the performance of multiclass stance detection. We show that using a feature-assisted neural model, aided with augmented training samples to deal with data imbalance, provides state-of-the-art performance on the FNC-1 dataset. The second contribution explores a way to improve stance detection, especially the minority categories, by proposing multistage classification approaches. We show a significant performance increase by breaking down the multiclass categories into different sub-stage feature-based and feature-assisted neural classifiers with category-specific features. Inspired by the multistage classification approaches, the final contribution proposes five-stage and three stage feature-assisted neural classifiers into multiclass fake statement detection. We conclude that sub-dividing the fine-grained task into multiple feature-specific classifier provides state-of-the-art performance.","abstract_has_math":false,"creators":["Hassan, Fuad Mire"],"institution":"University of Birmingham","degree_name":"d_ph","degree_level":"d_ph","degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2021,"date_issued":"2021-07","date_published":"2021-07","updated_at":"2026-11-19T09:31:28Z","subjects":["P Philology. Linguistics","QA75 Electronic computers. Computer science","T Technology (General)"],"languages":["eng"],"rights":[],"rights_urls":[],"identifier_entries":[],"person_identifier_entries":[]},"links":{"outbound_url":null,"outbound_label":null,"outbound_source":null,"record_url":null,"record_label":null,"record_kind":null,"record_source":null,"download_url":"http://etheses.bham.ac.uk//id/eprint/11673/13/Hassan2021PhD.pdf","download_label":"Download PDF","download_source":"dc:identifier.uri","access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://etheses.bham.ac.uk/cgi/oai2?verb=GetRecord&metadataPrefix=uketd_dc&identifier=oai%3Aetheses.bham.ac.uk%3A11673","prefix":"uketd_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.sponsor","label":"Sponsor","values":["other"]},{"key":"dc:creator","label":"Author","values":["Hassan, Fuad Mire"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2021-07-21"]},{"key":"dc:date.issued","label":"Date","values":["2021-07"]},{"key":"dc:publisher.department","label":"Dc Publisher Department","values":["College of Engineering & Physical Sciences","School of Computer Science"]},{"key":"dc:publisher.institution","label":"Dc Publisher Institution","values":["University of Birmingham"]},{"key":"dc:relation.isreferencedby","label":"Dc Relation Isreferencedby","values":["http://etheses.bham.ac.uk//id/eprint/11673/"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"dc:type.qualificationlevel","label":"Dc Type Qualificationlevel","values":["d_ph"]},{"key":"dc:type.qualificationname","label":"Dc Type Qualificationname","values":["d_ph"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["P Philology. Linguistics","QA75 Electronic computers. Computer science","T Technology (General)"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.dcterms_iso639_2","label":"Language (ISO 639-2)","values":["eng"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["http://etheses.bham.ac.uk//id/eprint/11673/13/Hassan2021PhD.pdf"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["The rapid increase of real-time news posted in social media has led to the emergence of fake news. Assessing the veracity of news claims requires an enormous labour of human fact-checkers and therefore automating the sub-tasks of fake news detection pipeline could help them identify false claims. Although the related literature addresses fake news detection tasks in a simple binary or a multiclass classification setting, challenges still remain. For instance, (1) this domain suffers from a lack of large scale datasets and a large proportion of the instances belongs to legitimate news which creates a class-imbalance problem, (2) the characteristics of fake news are not yet known in order to generate effective discriminative features (3) and the content of multiclass categories can be very similar which makes it hard for multiclass classifiers to capture the finer distinctions between them. The major focus of this thesis is to investigate novel models in Natural Language Processing (NLP) and Machine Learning (ML) that can help classify the veracity of a claim with respect to textual evidence into multiclass categories. Our first contribution is related to boosting the performance of multiclass stance detection. We show that using a feature-assisted neural model, aided with augmented training samples to deal with data imbalance, provides state-of-the-art performance on the FNC-1 dataset. The second contribution explores a way to improve stance detection, especially the minority categories, by proposing multistage classification approaches. We show a significant performance increase by breaking down the multiclass categories into different sub-stage feature-based and feature-assisted neural classifiers with category-specific features. Inspired by the multistage classification approaches, the final contribution proposes five-stage and three stage feature-assisted neural classifiers into multiclass fake statement detection. We conclude that sub-dividing the fine-grained task into multiple feature-specific classifier provides state-of-the-art performance."]},{"key":"dc:format","label":"Dc Format","values":["text"]},{"key":"dcterms:accessrights","label":"Dcterms Accessrights","values":["public"]},{"key":"dc:title","label":"Title","values":["Multistage feature-assisted deep learning and its application in fine-grained fake news detection"]}]}],"canonical_facts":{"dc:contributor.sponsor":["other"],"dc:creator":["Hassan, Fuad Mire"],"dc:date":["2021-07-21"],"dc:date.issued":["2021-07"],"dc:description.abstract":["The rapid increase of real-time news posted in social media has led to the emergence of fake news. Assessing the veracity of news claims requires an enormous labour of human fact-checkers and therefore automating the sub-tasks of fake news detection pipeline could help them identify false claims. Although the related literature addresses fake news detection tasks in a simple binary or a multiclass classification setting, challenges still remain. For instance, (1) this domain suffers from a lack of large scale datasets and a large proportion of the instances belongs to legitimate news which creates a class-imbalance problem, (2) the characteristics of fake news are not yet known in order to generate effective discriminative features (3) and the content of multiclass categories can be very similar which makes it hard for multiclass classifiers to capture the finer distinctions between them. The major focus of this thesis is to investigate novel models in Natural Language Processing (NLP) and Machine Learning (ML) that can help classify the veracity of a claim with respect to textual evidence into multiclass categories. Our first contribution is related to boosting the performance of multiclass stance detection. We show that using a feature-assisted neural model, aided with augmented training samples to deal with data imbalance, provides state-of-the-art performance on the FNC-1 dataset. The second contribution explores a way to improve stance detection, especially the minority categories, by proposing multistage classification approaches. We show a significant performance increase by breaking down the multiclass categories into different sub-stage feature-based and feature-assisted neural classifiers with category-specific features. Inspired by the multistage classification approaches, the final contribution proposes five-stage and three stage feature-assisted neural classifiers into multiclass fake statement detection. We conclude that sub-dividing the fine-grained task into multiple feature-specific classifier provides state-of-the-art performance."],"dc:format":["text"],"dc:identifier.uri":["http://etheses.bham.ac.uk//id/eprint/11673/13/Hassan2021PhD.pdf"],"dc:language.dcterms_iso639_2":["eng"],"dc:publisher.department":["College of Engineering & Physical Sciences","School of Computer Science"],"dc:publisher.institution":["University of Birmingham"],"dc:relation.isreferencedby":["http://etheses.bham.ac.uk//id/eprint/11673/"],"dc:subject":["P Philology. Linguistics","QA75 Electronic computers. Computer science","T Technology (General)"],"dcterms:accessrights":["public"],"dc:title":["Multistage feature-assisted deep learning and its application in fine-grained fake news detection"],"dc:type":["Thesis"],"dc:type.qualificationlevel":["d_ph"],"dc:type.qualificationname":["d_ph"]},"updated_at":"2026-11-19T09:31:28Z"}