Computational Lexica Work Group
 
            Minutes of Meeting Held in Cambridge, Massachusetts
                         31 October 1990
 
                           Robert Ingria
 
                    Document Number:  TEI AI6M1
 
The TEI working group on computational lexica had its initial meeting
on October 31, 1990, at BBN in Cambridge, Massachusetts.  Present were
Robert Ingria (chair), James Pustejovsky, and Susan Warwick.
 
At this meeting, one parameter on the nature of possible tag sets for
lexicons was decided upon.  (Note that the term ``lexicon'' here
denotes a machine-readable text containing lexical information, which
is principally intended for machine consumption.)  It is not necessary
for an interchange tag set for lexicons to preserve the physical
structure of the input text.  Unlike other texts that might be encoded
according to TEI guidelines, there is no original print form that
needs to be preserved.  Only the informational structure needs to be
preserved.
 
Given this background, we looked at a number of actual lexical
entries, principally from Machine Translation lexicons.  (Sample MT
entries had been gathered by Susan Warwick via EMail solicitation.)
Among the systems represented were Critter, Polygloss, METAL, ELU,
DLT, and The Utrecht Mimo system.  These systems differed along a
number of dimensions:
 
(1) Declarative vs. procedural representation.
 
(2) Full bilingual lexica for individual language pairs vs.
monolingual lexica for each language and separate transfer lexica, for
each language pair.
 
There were also brief examinations of entries from parsing lexicons,
and the two volumes of Igor Melcuk's Explanatory Combinatory
Dictionary for French which have appeared so far.
 
After examining the range of parameters on which these lexicons and
lexical databases vary, we came up with the following proposals.
 
(1) The members of the Working Group will each examine one of the
major lexicon types, by consulting actually existing lexicons, in a
feasability study to determine whether it is even possible to come up
with an interchange format for each individual lexicon type.  Tasks
were divide up as follows:
Ingria - will update his Grosseto survey on parsing lexicons and will
also look at generation lexicons.
Pustejovsky - will look at lexical knowledge bases, such as Miller's
Wordnet, Melcuk's ECD, and others, such as that currently being
developed at Brandeis.
Warwick - will continue her examination of MT lexicons.
 
(2) Since Warwick has been invited to attend the DARPA Spoken Language
Workshop in Pacific Grove, California, on February 19-22, 1991, that
seems like a reasonable time for a meeting at which the Working group
members will report on their conclusions.  (There may be one meeting
of all the members, if this is possible, or there may be a meeting of
Ingria and Pustejovsky in Boston, and one between Ingria and Warwick,
in Pacific Grove.  This is yet to be firmed up.)
 
(3) The feasability studies will determine what the group can usefully
do within the mandate of the TEI to propose standards for exchange of
lexical data.  If possible, we will propose an initial set of
guidelines.  To the greatest extent possible, we will try to be
consistent with the proposals of the dictionary working group, and
other working groups (such as syntax, morphology, etc.)
 
(4) We will have a follow-up meeting, probably in Europe, to examine
the individual draft proposals and to see if they can be combined into
a single set of guideline or if they should be separate.  Even if they
are separate tag sets, it is likely that identical or analogous pieces
of information will be given the same tagging.  Scheduling for this
meeting will follow the successful outcome of our February meeting(s).
 
(5) At the end of the first year of this working group, depending on
the outcome of the meetings in (2) and (4), we will form one or more
new working groups, to draft more complete draft specification for the
requisite tag set(s).
 
The following items were handed out to the participants:
 
By the Chair:
 
An SGML version of TEI document AI6 P1
A text version of TEI document AI6 P1
An excerpt from the minutes of the Tucson meeting dealing with a
	discussion of the goals of the lexicon working group
Discussion notes
 
By Susan Warwick-Armstrong:
 
Susan Warwick: ``Automated Lexical Resources in Europe: a survey''
 
Example MT entries:
 
heid.ex = Polygloss entries
metal.ex = Metal entries
aval.ex = ELU Transfer dictionary
dlt.ex = DLT examples
pim1.ex = MIMO system
pim2.ex = ditto
CRITTER examples