Past, Present, Future of Data/Information Modeling

1
Past, Present, Future of Data/Information
Modeling

• Foster Distinguished Chair Professor
• Computer Science Dept.
• Louisiana State University
• Baton Rouge, LA 70803, USA
• pchen_at_lsu.edu
• http//www.csc.lsu.edu/chen

Peter P. Chen
2
Overview

• Historical Background
• How Entity-Relationship Model (ERM) was Developed
• Last Twenty-Five Years
• ER Conferences, IDEF, ANSI/IRDS, CASE
  Methodologies/tools
• The Present
• OO, Data Mining, UML
• The Future
• Discovering Links/Relationships
• Validity/Credibility of Data, Machine
  Learning/Reasoning
• Natural Languages and Data/Information Modeling
• Modeling for XML/Semantic Web
• Conclusions

3
The Needs of the DB Community in the Early 70s

• For Software/Hardware Vendors
• Integration of Various File and DB Formats
• Incorporating More Data Semantics
• For User Organizations
• A Unified Methodology for File and DB design for
  Various File and DBMSs
• Incorporating More Business Rules

4
How the ERM was Developed -- Right Place
at the Right Time (I)?

• I Got Ph.D. From Harvard in 1973
• Thesis Title Optimal File Allocation
• Worked for Honeywell from 73 to 74
• In a 10-person Architecture Team for Next
  Generation Distributed System
• Many Team Members Were DB Experts 20 Years
  Older
• Charles Bachman, Henry Leftkovitz,
• John Lyon

5
Right Place at the Right Time (II)?

• Joined MIT Management School Faculty in 74
• Interacted with User Organizations
• They wanted a unified modeling and design
  methodology
• Completed the ERM Paper
• Most other faculty members were busy implementing
  DBMS prototypes

6
Concepts of Entity and Relationship
Figure 2
7
An Example of ER Diagram
Figure 2
8
Theoretical Foundations of ER Model

• Set Theory
• Modern Algebra
• Logic
• Lattice Theory

9
Defining ER Concepts using Set
Figure 2
10
Relationship as an Ordered Tuple
Figure 2
11
First ER Paper Initial Reactions (I)

• Published in ACM Transactions on DBMS, Vol. 1,
  No.1, pp. 9-36, March 76

12
First ER Paper Initial Reactions (II)

• The Situation then
• Most People were in Religious War
• I was a New Kid on the Block
• The Advices I got
• Dropped the ER Model
• Joined one of the Religious Camps

13
The First Five Years (I)

• Persistence with the ER Model
• Continue to Write Papers on ERM
• Organized First ER Conference in 1979 at UCLA
• 2nd ER Conference Two Years Later
• Now, an Annual Conference
• November 2001 in Japan
• 2002 in Finland
• 2003 in Chicago
• 2004 in Shanghai, China

14
The First Five Years (II)

• Some Academic People Started to Develop
  Semantics-Richer Data Models
• Mike Hammer of EECS, MIT, now a guru in
  reverse-engineering
• US-AF ICAM/IDEF Project
• Served as a consultant
• The ERM Became the Basis of IDEF
• More Companies Started to Experiment ERM

15
Related Developments in Next 20 Years

• Codds RM/T Model added ER Concepts
• Bachmans Partnership Model, too.
• ANSI/IRDS Standard of Information Resource
  Directory Systems (IRDS) Adopted ERM
• CASE (Computer-Aided Software Engineering)
• First Major CASE Symposium in Atlanta, 1987,
  Keynote speaker
• IBM AD Cycle Based on ERM
• IBM DB2 Repository Mgr ERM
• Oracle Desinger/200 -- ERM

16
The Present Status of Data/Information Modeling
(I)

• ER Modeling is the most widely used methodology
  in the business DB application development world
• more than 85 of the FORTUNE 3,000 companies and
  major organizations are using it
• More advanced ER concepts are proposed and used
• UML, which is a specific language syntax,
  reinforces the ER concepts

17
The Present Status of Information Modeling (II)

• OO Modeling incorporates many concepts of ERM
• Object is an implementation concept
• Current OO methodologies deed more general
  concepts of relationship
• What is Data Mining?
• Discover hidden relationships,
• Discover the embedded ER Models

18
Future (I)

• Discovering Links/Relationships from Data in
  Various Sources (such as DARPAs EELD Program)
• Validity/Credibility Analysis and Integration of
  Data Machine Learning/Reasoning
• Natural Languages vs. Data/Info Modeling. ER
  Modeling Concepts are Similar to
• Chinese Character Composition Methods
• Ancient Egyptian Hieroglyph
• English Sentence Grammar Structure

19
The Future (II)

• ERM and the Fundamental Principles of Systems
  Architecture
• ER Model is closely related to
• XML
• Semantic Web

20
Information Validity/Credibility Analysis

• A Paper was published in InfoFusion 2001,
  Montreal
• Algorithm was developed
• Prototype developed
• Also, developed machine learning algorithm

21
ER Modeling English

• First presented the ideas (abstract of a paper)
  in 2nd ER Conference in Washington, D.C. 1981
• Paper was published in Information Sciences, 1983
• Adopted as a standard systems analysis technique
  by some large consulting firms
• Recently, OO Analysis re-discovered some of the
  basic concepts
• Also, the research community started to use,
  modify, or extend the concepts

22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
Chinese Characters as Models of Real World
Entities
Figure 2
26
Ancient Egyptian Hieroglyph
Figure 2
27
Various Components of XML

• XML has many components
• XML (language part)
• XSL
• DOM
• DTD
• XLink and XPointer
• RDF
• XML Schema
• etc.
• Not all are compatible with each other

28
What is RDF?

• Acronym for Resource Description Framework
• As a way to specify metadata
• Two parts
• Model and Syntax
• Schema
• The RDF Schema is not a W3C recommended
  specification, yet
• http//www.w3.org/TR/REC-rdf-syntax

29
W3C Pays Attention to ERM

• The Cambridge Communiqué (http//www.w3.org/TR/sch
  ema-arch) states
• RDF can be viewed as a member of the
  Entity-Relationship Model Family
• In several articles, Tim Berners-Lee discusses
  the similarity and differences between ERM and
  RDF.

30
RDF vs. ER Model

• RDF can be viewed as a version of binary ER model
  (but at a lower and more detailed level)
• RDFs dependence on sentence analysis is similar
  to a series of work done in the correspondence
  between the ER model and English (and several
  other natural languages).
• Reference Chen, P.P., Entity-Relationship
  Diagram and English Sentence Structure,
  Information Science, 1983, Academic Press.
• Major concepts
• Noun --gt Entity, Verb --gt Relationship
• Adjective --gt Attribute of Entity, Adverb --gt
  Attribute of Relationship
• Gerund --gt Relationship-converted Entity
• Etc.

31
Real World Modeling Fundamental Principles of
Systems Architecture

• Entity Lattice and Other Mathematical Structures
  and Operations
• Fundamental Principles of Systems Architecture
• Starting from Info System Architecture, and then
  extends to all kinds of systems
• Fundamental Questions on
• Representation/Understanding
• Operations
• Costs/Benefits/Optimization

32
(No Transcript)
33
Conclusions (I)

• ER Modeling was triggered by the needs
• Unifying data views from top-down and bottom-up
  perspectives
• For vendors user organizations
• Incorporating more sematics
• Entity and relationship are fundamental concepts
  for
• Data/Knowledge Representation
• Database design
• Software engineering
• Information system development
• And others (data mining, ...)

34
Conclusions (II)

• Future
• Discovering Missing/Intended/Un-intended
  Relationship from data
• Prediction of the Validity of data and data
  model Machine Learning/Reasonsing
• Natual Languages and ERM
• Multi-language Information Extraction and
  Understanding
• Culture-based Modeling Methodology
• Modeling Design of Web
• ?Theory of Web, Semantic Web
• Fundamental Principles of Systems Architecture

35
References

• ER and other Conferences
• ER2002 (Finland), ER2003 (Chicago), ER2004
  (Shanghai)
• http//www.conceptualmodeling.org
• Chens papers online
• http//www.csc.lsu.edu/chen
• XML Schema
• Primer http//www.w3c.org/TR/xmlschema-0/
• Structure http//www.w3c.org/TR/xmlschema-1/
• Data Types http//www.w3c.org/TR/xmlschema-2/
• XML XLink XPointer
• http//www.w3c.org/XML/Linking
• RDF
• http//www.w3c.org/RDF/