OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

math-optimize-discuss message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]

Subject: Data interface standard


In an earlier posting Alan King suggested that the TC's mission should 
"include standards for interfaces between modeling languages, solvers, 
and data."

The data issue is an interesting one. My guess is that most of us agree 
that the linear program (or nonlinear program) instance fed to a solver 
should be in an XML format.  However, what about the raw data that is 
used by a modeling language to create the instance? Most of the modeling 
languages I am familiar with are set or table based.  I think that all 
of the major databases, DB2, Oracle, and SQL Server now allow the user 
to query the database and get the result in XML format.  Microsoft has 
SpreadsheetML an XML dialect for spreadsheets.  Hence, it seems 
reasonable to get data over the wire in XML format as input to a 
modeling language or solver.  Persistent XML databases such as Tamino 
from Software AG and Xindice from Apache are native XML databases.  The 
result of an XQuery/XPath query on XML data is a sequence of nodes or 
atomic values.  This is very different than a table.

I would like to know if the modeling language people that subscribe to 
this list think the TC should address any issues relating to a standard 
interface between the modeling language and the raw data?  Or, by data 
will we always mean the problem instance as opposed to the raw data used 
to generate the problem instance?


Kipp Martin
Professor of Operations Research
    and Computing Technology
University of Chicago
Graduate School of Business
1101 East 58 Street
Chicago IL 60637

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]