martins
Site Admin
Joined: 16 Dec 2009
Posts: 1481
Location: Geneva, Switzerland
|
 Posted: Wed Dec 08, 2010 7:22 am Post subject: RES: Questions/Remarks about the attributes : conflicts of |
 |
|
1. We have been trying to avoid representing conventions by means of attributes, and that’s why the set of convention attributes has been suppressed from the Attributes page. The UNL should be language-independent and, in that sense, typographical conventions (such as parentheses, brackets, quotes, etc.) do not make much sense, because different writing systems adopt different strategies for conveying this kind of information. In that sense, we have decided to represent not conventions but the semantic information conveyed by conventions (the fact, for instance, that information coming between parentheses indicate the “content” or an “explanation” of the preceding elements). There is, indeed, a conflict between the Attributes page and the ones exported from the tagset, because the latter have been stored in a database which is used to validate UNL documents and should comprise all existing attributes, including those listed in UNL2005. But we have split it now in order to avoid such problems.
2. The attribute @not indicates negation, which is a semantic notion normally associated both to degree and to polarity. Notice, please, that – differently from relations, which is organized in a taxonomy, i.e., a hierarchy where every branch is also a relation – attributes have been categorized for better interpretability only. They do not constitute an ontology, in the sense that the set of attributes is simply a list, except when explicitly indicated (i.e., in case of @pl, @def, @indef, which may be further specified).
---------------------------------------------
Ronaldo MARTINS
Language Resources Manager
UNDL Foundation
48, route de Chancy
CH-1213 – Geneva - Switzerland
+41 22 879 8090
http://www.undlfoundation.org
---------------------------------------------
|
|
FAQ 
Search 
Memberlist 
Usergroups 
Profile 
