We describe Boeing's NLP system, BLUE, comprising a pipeline of a parser, a logical form (LF) generator, an initial logic generator, and further processing modules. The initial logic generator produces logic whose structure closely mirrors the structure of the original text. The subsequent processing modules then perform, with somewhat limited scope, additional transformations to convert this into a more usable representation with respect to a specific target ontology, better able to support inference. Generating a semantic representation is challenging, due to the wide variety of semantic phenomena which can occur in text. We identify seventeen such phenomena which occurred in the STEP 2008 "shared task" texts, comment on BLUE's ability to handle them or otherwise, and discuss the more general question of what exactly constitutes a "semantic representation", arguing that a spectrum of interpretations exist.
[1]
Albert Camus,et al.
The Fall
,
2020,
Gospel Patterns in Literature.
[2]
Alison A. Carr-Chellman.
I have a problem!
,
1999
.
[3]
Peter Clark,et al.
A library of generic concepts for composing knowledge bases
,
2001,
K-CAP '01.
[4]
Peter Clark,et al.
Capturing and answering questions posed to a knowledge-based system
,
2007,
K-CAP '07.
[5]
David J. Israel,et al.
Reading to Learn: An Investigation into Language Understanding
,
2007,
AAAI Spring Symposium: Machine Reading.
[6]
Johan Bos.
Introduction to the Shared Task on Comparing Semantic Representations
,
2008,
STEP.