Hi Alberto:

Thank you for the pointer to your workshop. I did not know about it.
But I had lunch with Hector Garcia-Molina and he told me about
activity in semi-structured data.

In programming the data is semi-structured
for two reasons:

1. for each behavior, some of the data is noise and should not be mentioned
   in the high-level behavior.
2. the data structures are changing frequently within constraints
   which can be viewed as semi-structured.

I think that graph-oriented query languages such as GraphLog have
a bright future.

-- Karl

>From mendel@cs.toronto.edu Wed Feb 25 12:51:29 1998
>From: Alberto Mendelzon <mendel@cs.toronto.edu>
>To: lieber@ccs.neu.edu (Karl Lieberherr)
>Date: 	Wed, 25 Feb 1998 12:50:46 -0500
>
>Karl Lieberherr writes:
>> 
>> 
>> What is common between GraphLog and Demeter is that they both achieve a
>> similar form of structure-shyness. By that I mean that the queries or 
>> programs can operate on a wide variety of different structures.
>> The programming language community uses the term polytypic or shape 
>> polymorphism to express something similar.
>> 
>
>Interesting.
>In the database community there has been great interest
>recently on what is called "semistructured data", where there is
>no schema or only a very loose schema to constrain the structure of
>data.  See e.g. a workshop in conjunction with the ACM SIGMOD/PODS
>conference in '97. (www.research.att.com/~suciu/workshop-papers.html)
>
>This caused a bit of a renaissance of interest in graph-oriented
>query languages such as Graphlog.
>
>-alberto.
>