Models and Tools for Collaborative Annotation

TL;DR

This paper presents the Annotation Graph Toolkit (AGTK), a software collection that supports various modes of collaborative linguistic annotation through database integration and tools like TableTrans, enabling efficient and flexible annotation workflows.

Contribution

It introduces AGTK's database interface and API, demonstrating how existing tools can be extended for collaborative annotation with minimal additional programming.

Findings

Supports multiple collaborative annotation modes

Enables efficient access to large annotation databases

Simplifies extending annotation tools for collaboration

Abstract

The Annotation Graph Toolkit (AGTK) is a collection of software which facilitates development of linguistic annotation tools. AGTK provides a database interface which allows applications to use a database server for persistent storage. This paper discusses various modes of collaborative annotation and how they can be supported with tools built using AGTK and its database interface. We describe the relational database schema and API, and describe a version of the TableTrans tool which supports collaborative annotation. The remainder of the paper discusses a high-level query language for annotation graphs, along with optimizations, in support of expressive and efficient access to the annotations held on a large central server. The paper demonstrates that it is straightforward to support a variety of different levels of collaborative annotation with existing AGTK-based tools, with a minimum of additional programming effort.