Developing Extensible Open Information Extraction with Adding Machine Learning Components

• Main Author: Adrian Saputra - NIM : 13513030 , Yoga
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/31741
• Institution: Institut Teknologi Bandung

Abstract—Open Information Extraction (Open IE) as a paradigm has been found since 2007. And in this time, there are many kinds of open IE system that exist. Basically, open IE as a process can be divided into three big part of tasks. Those are preprocess, extraction, and postprocess. An Open IE system that is modular and extensible (Owen, 2017) exists before. That system consists of components that control those three tasks to do an open information extraction. Using that system, we can do an open information extraction with our personal implementation as part of the system. However, as a modular open IE, that system is not able to keep modularity and extensibility when doing open IE with machine learning approach. This paper talks about developing that system so that it will be able to keep modularity and extensibility when doing open IE with machine learning approach. Analysis in existing Open IE systems with machine learning approach will be done in order to make the new system be able to do open IE with machine learning approach. As the result, new components will be added to that open IE system. Those are data process and classifier. In addition, some existing component will be modified. The new system is tested and show that open IE with machine learning approach is able to be applied in the system and users are able to add their personal implementation to every part of machine learning process.