Argument Selection And Coercion

SemEval-2010 Task 7: Argument Selection and Coercion
James Pustejovsky and Anna Rumshisky and Alex Plotnick
Dept. of Computer Science
Brandeis University
Waltham, MA, USA
Elisabetta Jezek
Dept. of Linguistics
University of Pavia
Pavia, Italy

Olga Batiukova
Dept. of Humanities
Carlos III University of Madrid
Madrid, Spain

Abstract
We describe the Argument Selection and
Coerciontask for the SemEval-2010 evaluation exercise. This task involves characterizing the type of compositional operation that exists between a predicate and
the arguments it selects. Specifically, the
goal is to identify whether the type that
a verb selects is satisfied directly by the
argument, or whether the argument must
change type to satisfy the verb typing. We
discuss the problem in detail,describe the
data preparation for the task, and analyze
the results of the submissions.

1

Introduction

In recent years, a number of annotation schemes
that encode semantic information have been developed and used to produce data sets for training
machine learning algorithms. Semantic markup
schemes that have focused on annotating entity
types and, more generally, word senses, havebeen extended to include semantic relationships
between sentence elements, such as the semantic role (or label) assigned to the argument by the
predicate (Palmer et al., 2005; Ruppenhofer et al.,
2006; Kipper, 2005; Burchardt et al., 2006; Subirats, 2004).
In this task, we take this one step further and
attempt to capture the “compositional history” of
the argument selection relative to thepredicate. In
particular, this task attempts to identify the operations of type adjustment induced by a predicate
over its arguments when they do not match its selectional properties. The task is defined as follows: for each argument of a predicate, identify
whether the entity in that argument position satisfies the type expected by the predicate. If not, then

Valeria Quochi
ILC-CNR
Pisa,Italy

identify how the entity in that position satisfies the
typing expected by the predicate; that is, identify
the source and target types in a type-shifting or coercion operation.
Consider the example below, where the verb report normally selects for a human in subject position, as in (1a). Notice, however, that through
a metonymic interpretation, this constraint can be
violated, asdemonstrated in (1b).
(1) a. John reported in late from Washington.
b. Washington reported in late.

Neither the surface annotation of entity extents
and types nor assigning semantic roles associated
with the predicate would reflect in this case a crucial point: namely, that in order for the typing
requirements of the predicate to be satisfied, a
type coercion or a metonymy (Hobbs et al., 1993;Pustejovsky, 1991; Nunberg, 1979; Egg, 2005)
has taken place.
The SemEval Metonymy task (Markert and Nissim, 2007) was a good attempt to annotate such
metonymic relations over a larger data set. This
task involved two types with their metonymic
variants: categories-for-locations (e.g., placefor-people) and categories-for-organizations (e.g.,
organization-for-members). One of the limitationsof this approach, however, is that while appropriate for these specialized metonymy relations, the
annotation specification and resulting corpus are
not an informative guide for extending the annotation of argument selection more broadly.
In fact, the metonymy example in (1) is an instance of a much more pervasive phenomenon of
type shifting and coercion in argument selection.
For example, in(2) below, the sense annotation
for the verb enjoy should arguably assign similar
values to both (2a) and (2b).

3

Figure 1: The MATTER Methodology
(2) a. Mary enjoyed drinking her beer.
b. Mary enjoyed her beer.

The consequence of this is that under current sense
and role annotation strategies, the mapping to a
syntactic realization for a given sense is made
more complex, and is...