Deep Representation Learning for Biomedical Text Mining

Simple item page
dc.contributor.advisorNajafirad, Peyman
dc.contributor.authorEbadi, Nima
dc.contributor.committeeMemberHuang, Yufei
dc.contributor.committeeMemberRao, Raghav
dc.contributor.committeeMemberJamshidi, Mo
dc.creator.orcidhttps://orcid.org/0000-0003-3479-2774
dc.date.accessioned2024-02-09T20:48:41Z
dc.date.available2024-02-09T20:48:41Z
dc.date.issued2021
dc.descriptionThis item is available only to currently enrolled UTSA students, faculty or staff. To download, navigate to Log In in the top right-hand corner of this screen, then select Log in with my UTSA ID.
dc.description.abstractThe rapidly evolving curation of biomedical publications has resulted in an information crisis, such that researchers and other individuals working in biomedical domain require specialized text mining tools to keep track of the ever-evolving literature landscape. With the advances in natural language processing (NLP) models, especially those that are based on deep representation learning, gaining valuable insights from large-scale biomedical corpora have become extremely popular among researchers. Inspired by the state-of-the-art models proposed in general NLP settings, many studies have tried to address the common biomedical natural language processing (BioNLP) tasks, such as named-entity recognition NER, information/relation extraction (IE/RE), and question answering. However, directly applying the advancements of general NLP field to biomedical text mining often yields unsatisfactory results due to scarcity of labeled biomedical documents, ambiguous relations of biomedical concepts–with other domains or with general English words–continuous distribution shift of word semantics, multiple synonyms across different domains, and reliability challenges of deep learning models in general. In this dissertation, we aim to address these challenges using deep representation learning algorithms customized for one or multiple common tasks in BioNLP domain. We study the under-investigate aspects of BioNLP problems such as interpretability and reliability in addition to challenges of scale, scarcity of labeled data, and changes of word distributions. This dissertation approach deep representation learning as a tool to address the requirements of the biomedical research community, rather than models that achieve competitive "accuracies". Furthermore, as the aforementioned challenges become more severe in the context of a pandemic, we constantly provide case studies of the current pandemic and provide exclusive solutions from which the literature could substantially benefit if such an emergency situation recur in the future (God forbid!).
dc.description.departmentElectrical and Computer Engineering
dc.format.extent136 pages
dc.format.mimetypeapplication/pdf
dc.identifier.isbn9798759960980
dc.identifier.urihttps://hdl.handle.net/20.500.12588/3292
dc.languageen
dc.subjectBiomedical Natural Language Processing
dc.subjectBiomedical Text Mining
dc.subjectDeep Learning
dc.subjectDeep Representation Learning
dc.subjectNatural Language Processing
dc.subjectText Mining
dc.subject.classificationElectrical engineering
dc.subject.classificationArtificial intelligence
dc.subject.classificationLinguistics
dc.titleDeep Representation Learning for Biomedical Text Mining
dc.typeThesis
dc.type.dcmiText
dcterms.accessRightspq_closed
thesis.degree.departmentElectrical and Computer Engineering
thesis.degree.grantorUniversity of Texas at San Antonio
thesis.degree.levelDoctoral
thesis.degree.nameDoctor of Philosophy

Files

Original bundle

Now showing 1 - 1 of 1
No Thumbnail Available
Name:
Ebadi_utsa_1283D_13549.pdf
Size:
5.46 MB
Format:
Adobe Portable Document Format
Download

Collections

Electronic Theses and Dissertations - UTSA Access Only