Show simple item record

dc.contributor.advisorHoare, Terrien
dc.contributor.authorPitale, Sagar Kishor
dc.date.accessioned2021-04-28T20:04:38Z
dc.date.available2021-04-28T20:04:38Z
dc.date.issued2020
dc.identifier.citationPitale, S.K. (2020). Topic modelling and theme discovery on Aylien News articles during COVID-19. Masters Thesis, Dublin Business School.en
dc.identifier.urihttps://esource.dbs.ie/handle/10788/4248
dc.description.abstractTopic modelling is increasingly important for the analysis of large volumes of unlabelled data necessitating scanning a collection of documents and identifying keywords and language usage patterns. It is a technique of unsupervised machine learning that enables clustering of similar word groups and expressions under topics as well as analyse individual topic content. The negative impacts of the pandemic have been reflected in the news media. This research applies topic modelling to the COVID-19 news articles from AYLIEN to identify key themes in the large volume of COVID-19 news articles. Topic modelling algorithms applied and compared include LDA, NMF, LSI and HDP. LDA showed interpretable topics with better topic coherence and identification of underlying themes including worldwide spread; workplace activity impact; lockdown implications; medical supply shortages; social and sport knock-on effects; and disease statistics. Results of applying the different algorithms are presented.en
dc.language.isoenen
dc.publisherDublin Business Schoolen
dc.rightsItems in eSource are protected by copyright. Previously published items are made available in accordance with the copyright policy of the publisher/copyright holder.en
dc.rights.urihttp://esource.dbs.ie/copyrighten
dc.subjectCOVID-19 Pandemic, 2020-en
dc.titleTopic modelling and theme discovery on Aylien News articles during COVID-19en
dc.typeThesisen
dc.rights.holderCopyright: The publisheren
dc.type.degreenameMSc in Data Analyticsen
dc.type.degreelevelMScen


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record