Re: Semantic network DBMS [was: Re: Contexts (was Classes vs. Instances)]
Dave,
There are two issues that must be distinguished:
1. The logic.
2. The data structures that represent the logic.
Graphs and relations can represent *identical* information in
logically equivalent ways. The query language must express
whatever logical relationships the user is asking for in
any way the user happens to think about them. The computer
system must store the data in whatever structures are
appropriate for the data access patterns and algorithms.
> The impedance mismatch of graphs with SQL has performance
> limitations of sizing and speed, but they don't really matter
> if you're solving the problem.
All these issues were thoroughly analyzed and understood 30 years
ago, and the ideal solution is to let the computers and the humans
do what they do best: the people should be able to think about
their problems without worrying about the details of the data
structures, and the computers should optimize the data access
without bothering the users with the details of the underlying
storage structures.
> It seems to me that anything more complicated than a SQL based
> database only attracts the attention of the more elite tier
> of software engineers.
Of course. Those are the *only* people who should worry about how
to design appropriate algorithms for matching the users' queries,
rules, and updates to the underlying data structures. The people
who enter the data, the rules, and the queries should not have to
worry about or even be aware of the details of the representation.
Following is a note I sent to another email list, which discusses
a system that has many of these properties. These goals are
achievable, but they require some serious work by the "elite"
to design systems that are usable by everybody else. (And don't
hold your breath waiting for RDF and OWL to achieve these goals.)
John Sowa
_________________________________________________________________
Extract from a note sent in March 2006:
I would like to mention an example of a very large application
that has been able to support interoperability by doing
semi-automated extraction of ontologies from unstructured
documents. The system does have a rather simple upper level,
but the axioms required for interoperability are extracted
as needed -- they are not predefined by the ontology. See
the paragraph at the end of this note.
More info about it can be found on their web site, but the
descriptions are rather salesy and don't get into the technical
detail. The developer, Gerard Ellis, is writing an article
for publication, but it is not yet available.
I wouldn't claim that these techniques will obviate all uses
of an upper ontology, but they do support my point that the
lower levels of the ontology, which are highly task dependent,
are where the detailed axioms belong -- and those axioms can
often be derived with the assistance of users who have no
training in formal logic or even the usual IT techniques.
John Sowa
_______________________________________________________________
The largest application system that uses conceptual graphs is
Sonetto (IVIS Group 2006), which was designed by Gerard Ellis and
implemented with extended versions of earlier algorithms by
Levinson and Ellis (1992). A key innovation of Sonetto is its
semi-automated methods for extracting ontologies and business
rules from unstructured documents. The users who assist Sonetto
in the knowledge extraction process are familiar with the subject
matter, but they have no training in programming or knowledge
engineering. The Conceptual Graph Interchange Format (CGIF) is
the knowledge representation for ontologies, rules, and queries.
CGIF is also used to manage the schemas of documents and other
objects in the system and to represent the rules that translate
CGIF to XML and other formats.
References:
IVIS Group (2006) "Sonetto overview," http://www.sonetto.com/sonetto/
Levinson, Robert A., & Gerard Ellis (1992) "Multilevel
hierarchical retrieval," Knowledge Based Systems 5:3, pp. 233-244.
For CGIF, see the draft ISO standard for Common Logic:
http://cl.tamu.edu