Subject: Re: [uima] Type System Base Model v.4
Overall looks good. I have a comment about the TextAnnotation and TextRegionalReference definintion.
There are some tricky details regarding a standard definition of beginChar, endChar. What exactly do these values represent - Unicode character offsets? Note that Apache UIMA does not use Unicode character offsets, but rather "UTF-16 code unit" offsets - that is, each element is a 16-bit value like a char in a Java String. This has the advantage of allowing constant-time indexing (e.g. for getCoveredTex). Unfortunately not all Unicode characters can be represented in 16 bits, so sometimes a single Unicode character is split across two 16 bit values. Apache UIMA doesn't use true Unicode character offsets in that cases. This is especially inconvenient for anyone using UTF-8 rather than UTF-16 (for example, Perl annotators).
I'm not sure what the right answer it, but using "real" Unicode character offsets seems like it's a more appropriate use of the Unicode standard. Then, for Apache UIMA to be compliant does it just need to accept XMI CASes with Unicode offsets and covert them to its own internal representation? Or would the internal representation have to actually change (which would be ugly).
Advisory Software Engineer
UIMA Framework Lead Developer
IBM T.J. Watson Research Center
Hawthorne, NY, 10532
Tel: 914-784-7706, T/L: 863-7706
Karin Verspoor <firstname.lastname@example.org> 03/30/2007 01:50 PM
[uima] Type System Base Model v.4
Please find attached the reviewed document outlining
changes/proposals/action plan for the Type System Base Model.
Karin Verspoor, Computational Linguist
Knowledge and Information Systems Science team
Computer, Computation & Statistics division
email: email@example.com Mail: Los Alamos National Laboratory
phone: 505-667-5086 PO Box 1663, MS B256
fax: 505-667-1126 Los Alamos, NM 87545
[attachment "TypeSystemBaseModel_v4.doc" deleted by Adam Lally/Watson/IBM]