Recent Advances in a Feature-Rich Framework for Treebank Annotation

Proceedings of the 22nd International Conference on Computational Linguistics (Coling 2008), pages 673–680 Manchester, August 2008 Recent Advances in a Feature-rich Framework for Treebank Annotation Petr Pajas Charles Univ. in Prague, MFF ´UFAL Malostransk´e n´am. 25 118 00 Prague 1 – Czech Rep. pajas@ufal.ms.mff.cuni.cz Jan ˇStˇep´anek Charles Univ. in Prague, MFF ´UFAL Malostransk´e n´am. 25 118 00 Prague 1 – Czech Rep. stepanek@ufal.ms.mff.cuni.cz Abstract This paper presents recent advances in an established treebank annotation frame- work comprising of an abstract XML- based data format, fully customizable ed- itor of tree-based annotations, a toolkit for all kinds of automated data process- ing with support for cluster computing, and a work-in-progress database-driven search engine with a graphical user interface built into the tree editor. 1 Introduction Constructing a treebank is a complicated process. Among other things it requires a good choice of tools, varying from elementary data conversion scripts over annotation tools and tools for consis- tency checking, to tools used for semi-automatic treebank building (POS taggers, syntactic parsers). If no existing tool fits the needs, a new one has to be developed (or some existing tool adapted or ex- tended, which, however, seldom happens in prac- tice). The variety of tools that exist and emerged from various treebanking projects shows that there is no simple solution that would fit all. It is some- times a small missing feature or an incompati- ble data format that disqualifies certain otherwise well-established tools in the eyes of those who de- cide which tools to use for their annotation project. This paper presents an annotation framework that was from its very beginning designed to be extensible and independent of any particular anno- tation schema. While reflecting the feedback from several treebanking projects, it evolved into a set c⃝ 2008. Licensed under the Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported li- cense (http://creativecommons.org/licenses/by-nc-sa/3.0/). Some rights reserved. of generic tools that is open to all kinds of anno- tations that involve tree structures. By this paper we would like not only to promote this framework, but also show that due to its open nature, it may be easily extended to fit new requirements. The first three sections describe base components of the framework, an abstract data format, a versatile annotation tool for tree-oriented annotations, and a framework for automatic annotation processing; some of these components have been mentioned in earlier publications, but the framework has nei- ther been published in its integrity nor described in much detail. The last section describes a query engine that is a newest addition to the framework, first presented by this paper. 2 Data format The base data format selected for the present an- notation framework, both for data exchange and as a memory-model reference, is PML (Pajas and ˇStˇep´anek, 2006). PML is an abstract XML-based format intended to be generally applicable to all types of annotation purposes, and especially suit- able for multi-layered treebank annotations fol- lowing the stand-of principles. A notable fea- ture that distinguishes PML from other encoding schemes, like …

Content truncated to 500 words