Intro

Generative Lexicon (GL) is a theory of linguistic semantics
which focuses on the distributed nature of compositional-
ity in natural language (Pustejovsky, 1995). Unlike purely
verb-based approaches to compositionality, GL attempts to
spread the semantic load across all constituents of an utter-
ance. From the nature of word meaning to lexical creativity,
GL provides a different perspective on many of NLPs most
important questions. Hence, GL is not just a theory, but is
meant to be implemented as a component of the backbone
of larger NL systems (Pustejovsky and Boguraev, 1993).
One of the major complaints against GL has been that it is
too difficult to embed in such a system. At the heart of GL is
its network of qualia relations, and any true GL implemen-
tation would have to have a system of qualia-like structures.
However, creating such an ontology requires a prohibitive
investment of time for most researchers, and current GL
implementations place large constraints on any lexical de-
velopment project.

To help overcome this problem, we have developed a large
generative lexicon ontology and dictionary for use by the
general research community. This system, called the Bran-
deis Semantic Ontology (BSO), is intended to allow for
more widespread access to GL-based lexical resources and
help researchers in a variety of computational tasks. The
specification of the type system used in the BSO largely
follows that proposed by the SIMPLE specification (Busa
et al., 2001), which was adopted by the EU-sponsored SIM-
PLE project (Lenci et al., 2000).