Beschreibung
Agreeably, the internet has increased access to information and socialization in todays digital age. Topic modelling such as Latent Dirichlet Allocation (LDA) and its variants has been applied in analyzing text-data from these sources, with only few in the analysis of Online Community Forums - a kind of microblog that displays contents as posts and also facilitates social interactions/reactions of users as well as interest such as comments and number of views received by posts. In this work, two LDA variant models namely the Correlated Topic Model (CTM) and Supervised LDA (sLDA) were used to analyze political posts from a Nigerian Online community forum (Nairaland.com) as both supervised and unsupervised learning tasks. The CTM, an unsupervised model variant of LDA, which is able to capture topics correlation via the use of Logistic Normal Distribution in modelling the topic proportion as opposed Dirichlet distribution used in LDA. The CTM was trained on a subset of the forum post, which uncovered latent topics which can serve as an overview of issues embedded in Nigeria political spectrum. Details in the book.
Autorenporträt
Mr Babatunde B. Adebayo carried out the research work under the supervision of Dr OlaOluwa S. Yaya of Computational Statistics and Economic & Financial Statistics Units, Department of Statistics, University of Ibadan, Nigeria. Dr Adebola K. Ojo is in Data Mining Unit, Department of Computer Science, University of Ibadan, Nigeria.
Herstellerkennzeichnung:
OmniScriptum SRL
Str. Armeneasca 28/1, office 1
2012 Chisinau
MD
E-Mail: info@omniscriptum.com
