Papers in Neuroprose Archive

[Trent Lange]

The following two reprints have been placed in the Neuroprose
Archives at Ohio State University:
 

======================================================================
       Lexical and Pragmatic Disambiguation and Reinterpretation
         	   in Connectionist Networks

                         Trent E. Lange

                Artificial Intelligence Laboratory
                    Computer Science Department
               University of California, Los Angeles

   Lexical and pragmatic ambiguity is a major source of uncertainty
  in natural language understanding.  Symbolic models can make high-
  level inferences necessary for understanding text, but handle ambiguity
  poorly, especially when later context requires a reinterpretation of
  the input.  Structured connectionist networks, on the other hand, can
  use their graded levels of activation to perform lexical disambiguation,
  but have trouble performing the variable bindings and inferencing
  necessary for language understanding.  We have previously described a
  structured spreading-activation model, ROBIN, which overcomes many of
  these problems and allows the massively-parallel application of a large
  class of general knowledge rules.  This paper describes how ROBIN uses
  these abilities and the contextual evidence from its semantic networks to
  disambiguate words and infer the most plausible plan/goal analysis of
  the input, while using the same mechanism to smoothly reinterpret the
  input if later context makes an alternative interpretation more likely.
  We present several experiments illustrating these abilities and comparing
  them to those of other connectionist models, and discuss several directions
  in which we are extending the model.


*
 Appears in International Journal of Man-Machine Studies, 36: 191-220. 1992.

======================================================================

                               REMIND:
    Retrieval From Episodic Memory by INferencing and Disambiguation

                         Trent E. Lange
                Artificial Intelligence Laboratory
                    Computer Science Department
               University of California, Los Angeles

		       Charles M. Wharton
                    Department of Psychology
               University of California, Los Angeles

   Most AI simulations have modeled memory retrieval separately from
  language understanding, even though both activities seem to use many of
  the same processes.  This paper describes REMIND (Retrieval from
  Episodic Memory through INferencing and Disambiguation), a structured
  spreading-activation model of integrated text comprehension and
  episodic reminding.  In REMIND, activation is spread through a semantic
  network that performs dynamic inferencing and disambiguation to infer a
  conceptual representation of an input cue.  Because stored episodes are
  associated with concepts used to understand them, the spreading-activation
  process also activates any memory episodes in the network that share
  features or knowledge structures with the cue.  After the cue's conceptual
  representation is formed, the network recalls the memory episode having
  the highest activation.  Since the inferences made from a cue often
  include actors' plans and goals only implied in a cue's text, REMIND is
  able to get abstract, analogical remindings that would not be possible
  without an integrated understanding and retrieval model.

*
 To appear in J. Barnden and K. Holyoak (Eds.), Advances in Connectionist
 and Neural Computation Theory, Volume II:  Analogical Connections.
 Norwood, NJ:  Ablex.

======================================================================

Both papers are broken into two (large) postscript files stored in
compressed tarfiles.  To obtain a copy via FTP (courtesy of Jordan Pollack):
 
unix% ftp archive.cis.ohio-state.edu (or 128.146.8.52)
Name: anonymous
Password: (type your E-mail address)
ftp> cd pub/neuroprose
ftp> binary
ftp> get lange.disambiguation.tar.Z
ftp> get lange.remind.tar.Z
ftp> quit
unix% zcat lange.disambiguation.tar.Z | tar -xvf -
unix% lpr -s -P<local-printer> lange.disambiguation1.ps
unix% lpr -s -P<local-printer> lange.disambiguation2.ps
unix% zcat lange.remind.tar.Z | tar -xvf -
unix% lpr -s -P<local-printer> lange.remind1.ps
unix% lpr -s -P<local-printer> lange.remind2.ps
 
Note that the -s option for lpr is needed for most printers because
of the large size of the uncompressed postscript files (~ 1 meg each).
Sorry, no hard copies available.

Trent Lange

Artificial Intelligence Laboratory
Computer Science Department
University of California, Los Angeles
Los Angeles, CA 90024
E-Mail Address: lange at cs.ucla.edu