Design and Implementation of a Search and Management System for Annotation Data

Abstract

An annotation is an extra information to explain or comment on its original data. Annotation data is commonly used as a label for the original data in artificial intelligence applications. Annotation data is generated in various formats according to data annotation tools. Because data producers use different data annotation tools to create a dataset, data consumers have difficulties in using and managing dataset that has various annotation formats. To tackle this problem, we design and implement a system that integrates and manages data in various annotation formats. The proposed system stores data in various formats using one common schema and provides a function to automatically manage it when data is updated and search for annotation data between different formats.