Maritess: A Machine Learning Approach to a Multi-Source Automated Response Using Intelligent Time and Event-Based Spatio-Temporal Service
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
The diversity of public users in X’s (formerly known as Twitter) platform and the instantaneous speed by which information is delivered and presented in a data stream makes it an informal and extremely rich data source for important events. While previous studies made use of and developed automated and dynamic event detection on social media platforms, none of the services are available for public use. MARITESS, a Multi-source Automated Response Using Intelligent Time and Event-based Spatio-Temporal Service is a web-based application that provides a summary of events given an input location, using X as its data source. After collecting tweets related to the user input location, the application utilizes an LDA topic model to generate an undefined number of topics from the dataset and cluster them accordingly. A summary is generated for each topic using an transformer-based text summarization method to automatically provide a headline for each topic. The topic summaries accompanied by representative posts are presented to the user and location data (if available), giving an overview of the events related to the location being discussed on X. Evaluation of the model was done by measuring CV coherence score, as well as by comparing its performance against BERTopic, a state of the art topic modeling technique. The final application is accessible on HuggingFace Spaces.
Recommended Citation
Montero, Marielle G., (2023). Maritess: A Machine Learning Approach to a Multi-Source Automated Response Using Intelligent Time and Event-Based Spatio-Temporal Service. Archīum.ATENEO.
https://archium.ateneo.edu/theses-dissertations/869
