TR: Representing Context and Word Disambiguation

[steve gallant]

The following TR is available:

          A Practical Approach for Representing Context
          And for Performing Word Sense Disambiguation
                      Using Neural Networks

                       Stephen I. Gallant

                            ABSTRACT

Representing and manipulating context information is one of the hardest
problems in natural language processing.  This paper proposes a method for
representing some context information so that the correct meaning for
a word in a sentence can be selected.  The approach is based upon
work by Waltz & Pollack, who emphasized neurally plausible sys-
tems.  By contrast this paper focuses upon computationally feasi-
ble methods applicable to full-scale natural language processing
systems.

There are two key elements:  a collection of context vectors
defined for every word used by a natural language processing sys-
tem, and a context algorithm that computes a dynamic
context vector at any position in a body of text.  Once the
dynamic context vector has been computed it is easy to choose
among competing meanings for a word. This choice of definitions
is essentially a neural network computation, and neural network
learning algorithms should be able to improve the system's
choices.

Although context vectors do not represent all context informa-
tion, their use should improve those full-scale systems that have
avoided context as being too difficult to deal with.  Good candi-
dates for full-scale context vector implementations are machine
translation systems and text retrieval systems.  A main goal of
this paper is to encourage such large scale implementations and
tests of context vector approaches.

A variety of interesting directions for research in natural
language processing and machine learning will be possible once a
full set of context vectors has been created.  In particular the
development of more powerful context algorithms will be an impor-
tant topic for future research.

-----------------

Copies are available by Email only.
To obtain a Latex copy, send mail to `sg at corwin.ccs.northeastern.edu'.

Please do not post to other lists.

Please be careful not to reply to the entire connectionist list!