[developers] a late follow-up from lisboa

Ann Copestake Ann.Copestake at cl.cam.ac.uk
Wed Oct 26 22:24:42 CEST 2005

Thanks very much for producing this.

minor quick comments:

>     distinction in the type system already (in my view, the LKB use of
>     the term `lexical' rule is potentially mis-leading; but documented).

documented, very long-standing and discussed in a number of papers ...

>     i (still) have yet to re-read the earlier discussion on this on the
>     `developers' list, but currently at least berthold seems affected
>     by the two systems not doing the same (and the LKB no longer doing
>     what it used to).  so, it seems as if a written specification for
>     the mechanism was needed, and then hopefully we can aim at getting
>     all processors to implement it alike.

a non-procedural written specification.  I produced an attempt at this which
you'll find on developers

re MAF, PIC etc - a detailed discussion has been going on about this
between Ben, Uli S and Bernd.  I feel that MAF is important for
interaction with the outside world, but that the most important thing
is that we agree we're going to support:

a) character position specification enabling stand-off annotation
b) ambiguity from tokenisation onwards 
c) feature structure output from tokenisation, especially for named entities

(whoops, I think I've forgotten one, anyway, one of the others will no
doubt remind me).  Anyway, MAF and PIC both support these
notationally.  I understand PET supports them internally and the LKB
now supports them all too internally (or will do so very shortly).  I
think a) has implications for the HoG issues that you raise.  However,
I would like to suggest that we leave Ben et al to report on this
after they've met in a couple of weeks in Saarbruecken rather than
discuss it further by email.


More information about the developers mailing list