LREC Parse Banking Tutorial

There will be an LFG Parse Banking tutorial at this year’s LREC in May given by Alexis Baird and Paul Meurer.

This tutorial will examine the annotation procedures used for Powerset’s LFG Parsebanking project. The tutorial will include in-depth exploration of the LFG Parsebanker developed at the University of Bergen and Uni Digital , including the process for loading data, strategies for annotating different genres of text, and extracting results.

The parsebanking tool relies on input from the LFG-based XLE parser and can support several different types of grammars. We will showcase our standard index grammar as well as our query grammar. The tutorial will include an in-depth look at how annotators select the correct parse using a series of decision points and then comment on the sentence depending on whether the correct parse was available. After demonstrating the process for loading data and annotation, we will explore more sophisticated tasks such as adding a subset of duplicate sentences to extract inter-annotator agreement metrics.

The tutorial will cover:

1. Introduction
   • Motivation for the parsebanking tool
   • Short overview of theoretical foundations
2. Overview of the tool
   • Loading data
   • Annotating sentences
   • Tool features
   • Generating disambiguated parses for sentences
   • Examination of potential obstacles and solutions
3. Break
4. Cooperative creation of a “mini corpus”
   • Participants will each work using the tool themselves
   • Depending on interest and availability of different grammars, “mini-corpus” may be multi-lingual
   • “Mini corpus” will be available as a resource after the tutorial