Open AccessEstimating News Coverage Patterns using Latent Dirichlet Allocation (LDA)
Author(s) -
Naeem Ahmed Mahoto
Publication year - 2018
Publication title -
sukkur iba journal of emerging technologies
Language(s) - English
Resource type - Journals
eISSN - 2617-3115
pISSN - 2616-7069
DOI - 10.30537/sjet.v1i1.142
Subject(s) - latent dirichlet allocation , newspaper , topic model , schema (genetic algorithms) , computer science , probabilistic logic , context (archaeology) , information retrieval , artificial intelligence , advertising , geography , business , archaeology
The growing rate of unstructured textual data has made an open challenge for the knowledge discovery, which aims extracting desired information from large collection of data. This study presents a system to derive news coverage patterns with the help of probabilistic model – Latent Dirichlet Allocation. Pattern is an arrangement of words within collected data that more likely appear together in certain context. The news coverage patterns have been computed as number function of news articles comprising of such patterns. A prototype, as a proof, has been developed to estimate the news coverage patterns for a newspaper – The Dawn. Analyzing the news coverage patterns from different aspects has been carried out using multidimensional data model. Further, the extracted news coverage patterns are illustrated by visual graphs to yield in-depth understanding of the topics, which have been covered in the news. The results also assist in identification of schema related to newspaper and journalists’ articles.