Dear Paul and Nacho:

I am following the use of some of our Demeter ideas in new technologies 
and have a litle page on this at:

http://www.ccs.neu.edu/research/demeter/technology-transfer/in-main-stream.html

I noticed some interesting developments in the domain of your
thesis. Take a look at:

http://www.ccs.neu.edu/research/demeter/related-work/schema-learning/gionis.pdf
This is a SIGMOD 2000 paper from Bell Labs and Stanford.
I heard that they applied for a US patent.

This paper picks up the issue of deriving a class graph from 
object descriptions. The new terminology is:

object description	->	XML document
class graph		->	XML schema or document type definition (DTD)

The SIGMOD 2000 authors describe a system XTRACT that derives a DTD
from XML documents. 

There is one significant distinction between object descriptions and XML documents:
object descriptions give names to the parts while XML documents
do not.  But that might actually change as new XML
schema notations come along.

The Bell Labs/Stanford team explores similar issues as you do:
finding an appropriate generalization from the "examples",
minimizing the schema (they also have an NP-completeness result as
you do).

They use other techniques than you do (because they don't have the
part names).

It would be good if you could write a few paragraphs that compare 
your PhD thesis work with the new SIGMOD 2000 work.

It seems that your work is ahead of the SIGMOD 2000 work because you also
deal with incremental learning. 

Please compare the new paper with your works below.

A similar situation has also appeared with other theses. It is important
that we send a letter to Bell Labs and Stanford telling them 
that they missed some important related work that explored their
theme about 10 years ago.

Please let me know if your schedule is so busy that you
don't have time to look into this.

Best regards,
-- Karl

@ARTICLE{lieber-nacho-pberg:se-p,
AUTHOR = "Karl J. Lieberherr and Paul Bergstein and Ignacio {Silva-Lepe}",
TITLE = "From objects to classes: Algorithms for object-oriented design",
JOURNAL = "Software Engineering Journal",
YEAR =  "1991",
MONTH = "July",
VOLUME = 6,
NUMBER = 4,
PAGES = "205-228"
}

@INPROCEEDINGS{lieber-nacho-pberg:conf,
AUTHOR = "Karl J. Lieberherr and Paul Bergstein and Ignacio {Silva-Lepe}",
TITLE = "Abstraction of Object-Oriented Data Models",
BOOKTITLE = er,
YEAR = "1990",
ADDRESS = "Lausanne, Switzerland",
PAGES = "81-94",
EDITOR = "Hannu Kangassalo",
PUBLISHER = "Elsevier"
}

@INPROCEEDINGS{pberg-lieber:ecoop,
AUTHOR = "Paul Bergstein and Karl Lieberherr",
TITLE = "Incremental Class Dictionary Learning and Optimization",
BOOKTITLE = ecoop,
YEAR = "1991",
ADDRESS = "Geneva, Switzerland",
PAGES = "377-396",
PUBLISHER = "Springer Verlag Lecture Notes 512"
}

@PHDTHESIS{paul:thesis,
AUTHOR = "Paul Bergstein",
TITLE = "{Managing the Evolution of Object-Oriented Systems}",
SCHOOL = "Northeastern University",
YEAR = 1994,
NOTE = "151 pages"
}

@PHDTHESIS{nacho:thesis,
AUTHOR = "Ignacio {Silva-Lepe}",
TITLE  = "Techniques for Reverse-engineering and Re-engineering into the
Object-Oriented Paradigm",
SCHOOL = "Northeastern University",
YEAR   = "1994",
NOTE = "133 pages"
}

@InProceedings{Bergstein91,
author    = "Paul Bergstein",
title     = "Object-Preserving Class Transformations",
booktitle = "Proc. OOPSLA'91, {ACM} SIGPLAN Sixth Annual
Conference on
Object-Oriented Programming Systems,
Languages and Applications",
pages     = "299--313",
year      = 1991}