Marisol: Media-Based Automated Classifier System for Policy Issue Frames in Philippine Online News

Date of Award

12-1-2023

Document Type

Thesis

Degree Name

Master of Science in Computer Science

First Advisor

Maria Regina Justina E. Estuar, PhD

Abstract

Media plays an important function in disseminating information to the public, with the capacity to shape public opinion. Media may use frames in presenting the news. Frames are the actual news presentation, derived from selecting various aspects of news [27]. The objective of this study is to develop a publicly available tool to inform users of policy issue frame/s in news articles. Using Philippine online news as an input, MARISOL classifies the policy issue frame used by the article. Philippine Frame Corpus (PFC) trained on policy issue frame classification in Philippine online news. Media Frame Corpus [18] and PFC were tested on the classification of policy issue frames using supervised learning methods including BERT, LSTM, GRU, ELECTRA, GPT-2, and RoBERTa to classify the policy issue frames. Using unsupervised methodology in classifying frames for PFC showed similarities to frames like the Health and Safety Frame, and Law and Order, Crime and Justice Frame. LIWC and MANOVA were employed on the dataset. Results showed statistical significance on the linguistic features within policy frames and between datasets. The RoBERTa model was the bestperforming model with an accuracy of 77.01% and was used in the deployment of MARISOL.

Link to Full Text

Share

COinS