Oudg cross model datum access

David C. Wyld et al. (Eds) : ITCS, CST, JSE, SIP, ARIA, DMS - 2015
pp. 165–184, 2015. © CS & IT-CSCP 2015 DOI : 10.5121/csit.2015.50117
OUDG: CROSS MODEL DATUM ACCESS
WITH SEMANTIC PRESERVATION FOR
LEGACY DATABASES
Joseph Fong1
and Kenneth Wong2
1
Department of Computer Science, City University of Hong Kong, Hong Kong
1
csjfong@cityu.edu.hk,2
wting_yan@hotmail.com
ABSTRACT
Conventional databases are associated with a plurality of database models. Generally database
models are distinct and not interoperable. Data stored in a database under a particular
database model can be termed as “siloed data”. Accordingly, a DBMS associated with a
database silo, is generally not interoperable with another database management system
associated with another database sil. This can limit the exchange of information stored in a
database where those desiring to access the information are not employing a database
management system associated with the database model related to the information. The DBMS
of various data models have proliferated into many companies, and become their legacy
databases. There is a need to access these legacy databases using ODBC. An ODBC is for the
users to transform a legacy database into another legacy database. This paper offers an end
user’s tool of Open Universal Database Gateway(OUDG) to supplement ODBC by
transforming a source legacy database data into Flattened XML documents, and further
transform Flattened XML document into a target legacy database. The Flattened XML
document is a mixture of relational and XML data models, which is user friendly and is a data
standard on the Internet. The result of reengineering legacy databases into each other through
OUDG is information lossless by the preservation of their data semantics in terms of data
dependencies.
KEYWORDS
Open universal database gateway, Legacy databases, Flattened XML Documents, Data
semantics, Data dependencies, Open database connectivity
1. INTRODUCTION
The database management system (DBMS) of various data models have proliferated into many
companies and, over time, have become legacy databases within the companies. However, there
is a need to access these legacy databases, e.g., for mass information transmission associated with
e-commerce, etc. Legacy databases, e.g., conventional databases, can be associated with a
plurality of database models, e.g., database silos. These database silos can be distinct and fail to
interoperate without significant costs or loss of data or data semantic information. Siloed data,
e.g., data within a database model acts is typically only readily accessible or interoperable within
that database model and not with data stored in another database silo, can limit the exchange of
information where those desiring to access the information are not employing a related DBMS.

166 Computer Science & Information Technology (CS & IT)
Additionally, even where a database environment is relatively modern, it can be incompatible
with other relatively modern database silos. The plurality of database silos in itself can be an
impediment to sharing data among them. As an example, where a first company employs a first
database associated with a first database model, a second company employs a second data model
for their data, and a third company employs a third data model for their data, sharing of data
across the three data silos can be impractical or impossible. Where the first company purchase
the second company, incorporating the second company’s data can be problematic, e.g., it can
require rewriting the data into the first data model at the risk of losing data or semantics.
Alternatively, the first company can operate the two databases separately but are then internally
faced with the incongruences of the two databases, bear the costs associated with operating or
maintaining two separate databases, etc. Further, the first company, even with access to the first
and second databases, still can face serious challenges with sharing data with the third company.
The evolution of database technologies intend to meet different users requirements. For example,
the Hierarchical and Network (Codasyl) databases(NDB) are good for business computing on the
large mainframe computers. The user friendly relational databases(RDB) are good for end user
computing on personal computers. The object‐oriented databases(OODB) are good for
multi‐media computing on mini computers. The XML databases(XML DB) are good for Internet
computing on the mobile devices. These are first generation Hierarchical and Network databases,
second generation relational databases, and third generation Object-Oriented and XML databases.
Flattened XML documents
Flattened XML documents are generic representation of any legacy database instance in any
legacy database data model. Flattened XML document is a valid XML document which contains
a collection of elements of various types and each element defines its own set of properties. The
structure of the flattened XML document data file is a relational table structured XML document.
It has XML document syntax with relational table structure. It replaces primary key with ID, and
foreign key with IDREF as follows:
<?xml version="1.0">
<root>
<table1 ID="…" IDREF1="…" IDREF2="…" … IDREFN="…">
<attribute1>…</attribute1>
…
<attributeN>…</attributeN>
</table1>
…
<tableN ID="…" IDREF1="…" IDREF2="…" … IDREFN="…">
<attribute1>…</attribute1>
…
<attributeN>…</attributeN>
</tableN>
</root>
For each table, the name of the table (tableN) determines its type name and the name of property
(attributeN) determines its property name. Each table defines an ID type attribute that can
uniquely identify itself and there are optional multiple IDREF type attributes that can refer to the
ID in other tables in the same flattened XML document instance. Each property XML element
encloses a property value in a proper textual representation format. In order to ensure a flattened
XML document instance to be valid, there must be either an internal or an external DTD
document that defines the XML structures and attribute types, in particular for those ID and
IDREF type attributes.

Computer Science & Information Technology (CS & IT) 167
Open Universal Database Gateway
An open universal database gateway(OUDG) is a database middleware which provides more
flexibility for the users to access legacy databases in their own chosen data model. Users can
apply OUDG to transform legacy databases into flattened XML documents, and then further
transform them into user’s own familiar legacy database for access. Since XML is the data
standard on the Internet, it becomes information highway for user to access data.
The reason we choose flattened XML document is due to its openness for DBMS independence.
All other data models are DBMS dependent. For example, an Oracle database can only be
accessed by Oracle DBMS, and a MS SQL Server database can only be accessed by MS SQL
Server DBMS. Nevertheless, users can access flattened XML documents on the Internet by
Internet Explorer without programming. Therefore, an Oracle user can access an MS SQL Server
database by using OUDG transforming the MS SQL Server database into flattened XML
document, and then further transform flattened XML document to Oracle database.
Similarly, the reason we choose relational table structure for the flattened XML document is that
relational table structure has a strong mathematical foundation of relational algebra to implement
the constraints of major data semantics such as cardinality, isa and generalization to meet users’
data requirements by replacing primary keys and foreign keys into ID(s) and IDREF(s) in XML
schema.
The OUDG can transform legacy databases into flattened XML document, and then further
transform the flattened XML document into one of four target legacy databases: relational,
object-oriented, XML and network. The result is that OUDG allows users reengineer a source
legacy database into an equivalent target legacy database of user’s choice with data semantics
preservation.
This paper offers flattened XML documents as universal database medium for the interoperability
of all legacy databases that can be accessed by the users using their own familiar legacy database
language via OUDG. We consider hierarchical data model same as XML data model in this paper
because they are all in tree structure. All proprietary legacy data models can be united into
flattened XML document as universal database as shown in Figure 1.
Figure 1. Legacy Databases Interoperability via Flattened XML documents
The major difference between regular tree-based XML document and flattened XML document is
that the latter is more user friendly and efficient in database update since its database navigation
access path is much shorter than the former. The flattened XML document database navigation
access path is limited to 2 levels from root element to sibling elements while the regular XML
document database access path is in several levels and much longer in general.

Problems:
(1) Most legacy database systems are proprietary. Database vendors do not facilitate tools to
export their databases to other legacy databases. Thus, companies need to use ODBC to access
heterogeneous databases, which requires programming and much time effort.
(2) Most users cannot access all legacy databases because they do not know all legacy database
languages. They rely on ODBC, which is not easy to learn.
(3) It is difficult to convert legacy databases in different data models because the data conversion
of legacy database involves data models transformation.
Solution:
Through OUDG, users can use one database language access another legacy databases of
relational, object-oriented, network and XML. The operation is more reliable and speedy because
same data can be concurrently processed by legacy database and their equivalent flattened XML
document.
Academic merit:
The novelty is that it is feasible to replace ODBC by OUDG transforming legacy database into
flattened XML document for access. ODBC needs programming, but OUDG is an end user
software utility.
Industrial merit:
The application of flattened XML document is for information highway on the Internet for data
warehouse, decision support systems (Fong, Li & Huang, 2003). The benefits are information
sharing among users for database interoperability.
OUDG as supplement for ODBC
OUDG can supplement ODBC to access any legacy database by transforming(reengineering)
them into a flattened XML document for access as universal database which is defined as a
database interchangeable to all legacy databases.
At present, most database systems are proprietary. Each DBMS vendor has software tools which
convert other legacy databases into databases using their own DBMS(s), but not vice versa for
converting their own databases into a target legacy database. The result makes legacy databases
not open to each other. However, using OUDG, any legacy database can be transformed into any
other legacy database via flattened XML documents. The benefit is that data sharing and data
conversion among legacy databases becomes possible. The openness of legacy database is
necessary for such application such as data warehousing, data mining and big data.
Figure 2 shows the architecture of an open universal database gateway which transforms legacy
databases into each other with different data models via flattened XML document as a
replacement for open database connectivity.
Data Semantics preservation in legacy databases
Data semantics describe data definitions and data application for users’ data requirements, which
can be captured in the database conceptual schemas. The following are the data semantics which
can be preserved among the legacy conceptual schemas and their equivalent flattened XML
schema:
(a) Cardinality: 1:1, 1:n and m:n relationships set between two classes
A one-to-one relationship between set A and set B is defined as: For all a in A, there exists at
most one b in B such that a and b are related, and vice versa. The implementation of one-to-one
relationship is similar to one-to-many relationship.

A one-to-many relationship from set A to set B is defined as: for all a in A, there exists one or
more b in B such that a and b are related. For all b in B, there exists at most one a in A such that a
and b are related.
A many-to-many relationship between set A and set B is defined as: For all a in A, there exists
one or more b in B such that a and b are related. Similarly, for all b in B, there exists one or more
a in A such that a and b are related.
In relational schema, 1:n is constructed by foreign key on “many” side referring to primary key
on “one” side. It can also be implemented by association attribute of a class object on “one” side
points to another class objects on “many” side in object-oriented schema. It can also be
implemented by owner record occurrence on “one” side and member record occurrences on
“many” side in network schema. It can also be implemented by element occurrence with IDREF
on “many” side links with element occurrence with ID on “one” side in XML schema. As to m:n
cardinality, it can be implemented by two 1:n cardinalities with 2 “one” side classes link with 1
“many” side class.
(b) Isa relationship between a superclass and a subclass
The relationship A isa B is defined as: A is a special kind of B.
In relational schema, a subclass relation has same primary key as its superclass relation, and
refers it as a foreign key in isa relationship. In object-oriented schema, isa can be implemented by
a subclass inheriting its superclass’s OID and attributes. In Network schema, isa can be
implemented by an owner record that has same key as its member record in network schema via
SET linkage. In XML schema, isa can be implemented by an element links one-to-one occurrence
with its sub-element.
Figure 2. Architecture of OUDG with schema translation and data transformation
(c) Generalization is the relationship between a superclass and its subclasses.
Multiple isa relationships construct generalization with common superclass.
In relational schema, A is a special kind of B, and C is also a special kind of B, then A and C
subclasses can be generalized as B superclass. In relational schema, multiple subclass relations
and their superclass relation contain the same key, with subclass relations’ keys referring to
superclass key as foreign key in generalization. In object-oriented schema, multiple subclasses
objects contain the same OID as their superclass object in generalization. In network schema, one
owner record links with multiple member records through a SET in generalization. In XML
schema, multiple subclass elements and their superclass element are in 1:1 linkage with same key
attribute in generalization. Generalization can be implemented by multiple isa relationships with
multiple subclasses generalized into one superclass.
Initially, OUDG maps major data semantics of cardinality, isa, and generalization into each
legacy data model as shown in Table 1 which shows data semantics preservation in legacy data
models and Flattened XML document
The preservation of data semantics among legacy databases can be verified by the preservation of
their data dependencies as follows:
Definition of FD (functional dependency)

Given a relation R, attribute Y of R is functionally dependent on attribute X of R, i.e., FD: R.X
R.Y, iff each X-value in R has associated with it precisely one Y value in R. Attribute X and Y
may be composite.
Definition of ID (inclusion dependency)
ID: Y Z states that the set of values appearing in attribute Y must be a subset of the set of
values appearing in attribute Z.
Definition of MVD (multi-valued dependency)
Let R be a relation variable, and let A, B and C be the attributes of R. Then B is multi-dependent
on A if and only if in every legal value of R, the set of B values matching a given AC pair value
depends on the A value, and is independent of the C value.
In general, the mapping and the preservation of the data semantics of cardinality, isa, and
generalization among legacy databases schemas can be shown in Figure 3 as follows:
In one-to-many cardinality, for example, each child relation B tuple determines its parent relation
A tuple in relational schema; each member record B determines its owner record A in network
schema; each “many” side object B determines its associated “one” side object A in object-
oriented schema, and each sub-element B occurrence determines its element A occurrence in
XML schema.
In many-to-many cardinality, two one-to-many cardinality MVD(s) can construct a many-to-
many cardinality. For example, many tuples in relation B determine many tuples in relation A
and vice versa (many relation A tuples determine many relation B tuples); many records B
determine many records A. Therefore many elements B occurrence determine many elements A
occurrences, and vice versa.
Table 1 showing information related to data semantic preservation.
Data model
Data Semantic
Relational Object-
Oriented
Network XML Flattened
XML
1:n cardinality Many child
relations’
foreign keys
referring to
same parent
relation’s
primary key.
A class’s
association
attribute
refers to
another
class’s many
objects’
OID(s) as a
Stored OID.
An owner
record data
points to
many
member
records data
via SET
linkage.
An element
has many
sub-elements.
The IDREF(s)
of a “many”
side sibling
element’s data
refer to an ID
of “one” side
sibling
element data
under root
element.

m:n
cardinality
A
relationship
relation’s
composite
key are
foreign keys
referring to 2
other
relations’
primary keys.
A class’s
association
attribute
refers to
another
class’s many
objects’
OID(s) as an
Stored OID,
and vice
versa.
Two owner
records data
point to the
same
member
record data
via 2 SETs
linkages .
A sub-
element of an
element links
another
element
IDREF
referring to
the latter’s
ID. The 2
elements are
in m:n
cardinality.
An sibling
element data
has 2
IDREF(s)
referring to 2
other sibling
elements ID(s)
under root
element.
Is-a Subclass
relation’s
primary key is
a foreign key
referring to its
superclass
relation’s
primary key.
A subclass
inherit
OID(s),
attributes and
methods of its
superclass as
its OID plus
its own
attributes and
methods.
An owner
record data
links to a
member
record data in
1:1 with same
key.
An element
occurrence
links with a
sub-element
occurrence in
1:1 linkage.
The IDREF of
a subclass
sibling
element data
refers to the
ID of a
superclass
sibling
element. Both
elements has
same key
value under
root element.
Generalization 2 subclass
relations’
primary keys
are foreign
keys referring
to same
superclass
relation’s
primary keys.
Two
subclasses
inherit OID
and attributes
of their
identical
superclass as
their OID plus
their own
attributes.
An owner
record data
occurrence
points to two
member
records data
occurrence
with same
key.
An element
data
occurrence
links with two
sub-elements
data
occurrence in
1:1 linkages.
The IDREF(s)
of 2 subclass
sibling
elements data
occurrence
refer to an ID
of a superclass
sibling
element data
occurrence
with same key
value under
root element.
In isa relationship, for example, each B tuple is a subset of A tuple; each record B is a subset of A
record; each object B is a subset of object A; and each sub-lement B occurrences is a subset of
element A occurrence. In generalization, the data dependencies are similar to isa relationship,
except the pair of subclass B and C is a subset of superclass A.
The above data semantics can be preserved in flattened XML documents with sibling elements
only, linking with each other via IDREF and ID as shown in Figure 4.

Relational Schema
Relation A (A1, A2)
Relation B (B1, B2, *A1)
A A1
A2
B B1
B2
Relational conceptual
schema in EER model
R
1
m
FD: B A
B2
b21
*A1
a11
B1
b11
Relations
R1
R2
A1
a11
A2
a21
a12 a22
b22 a12b21
A A1
A2
Set AB
B B1
B2
Network conceptual
schema in Network Graph
FD: B A
Network Schema
Record Name is A
A1 Character
A2 Character
Record Name is B
B1 Character
B2 Character
Set AB
Owner is A
Member is B
B2
b21
B1
b11
Records
A
B
A1
a11
A2
a21
a12 a22
b22b12
(a) one-to-many cardinality(a) one-to-many cardinality
Object-Oriented
Conceptual schema in UML
A
A1, A2
B
B1, B2
1
m
FD: B A
Object-Oriented Schema
Class A
Attribute A1 Char
Attribute A2 Char
Attribute A_B set (B)
End
Class B
Attribute B1 Char
Attribute B2 Char
Attribute B_A (A)
End
A2A1OIDA
a21#1
#2 a22a12
a11
Classes
B2B1OIDB
b21#3
#4 b22b12
b11
Stotred OID
#1
#2
Stored_OID
#3, #4
#3, #4
A
A1
A2
B B1
B2
XML conceptual schema in
DTD Graph
*
XML schema in DTD
<!ELEMENT A(B*)>
<!ATTLIST A1 CDATA #REQUIRED>
<!ELEMENT B EMPTY>
<!ATTLIST B1 CDATA #REQUIRED>
FD: B A
<A A1=”a11" A2=”a12">


</A>
<A A2=”a21" A2=”a22">


</A>
XML Docment
Figure 3a Data semantics preservation in legacy databases (1:n Cardinality)

(b) many-to-many cardinality(b) many-to-many cardinality
Network Schema
Record Name is A
A1 Character
A2 Character
Record Name is B
B1 Character
B2 Character
Record Name is AB
Set AAB
Owner is A
Member is AB
Set BAB
Owner is B
Member is AB
A
Set AAB
AB
B
B1
B2
Set BAB
Network conceptual
AB
MVD: A B
MVD: B A
B2
b21
B1
b11
Records
A
B
A1
a11
A2
a21
a12 a22
b22b12
A1
A2
Relational Schema
Relation A (A1, A2)
Relation B (B1, B2)
Relation AB (*A1, *B1)
A A1
A2
B
B1
B2
AB
n
m
schema in EER model
MVD: A B
MVD: B A
*B1
b11
B1
b11
*A1
a11
Relations
A
AB
A1
a11
A2
a21
a12 a22
b21
b21
a12
B1
b11
A1
a11
b21a12
B2
b21
b22
XML schema in DTD
<!ELEMENT A(AB*)>
<!ELEMENT AB EMPTY>
<!ATTLIST AB_iderf IDREF #REQUIRED>
<!ELEMENT B EMPTY>
<!ATTLIST B id ID CDATA #REQUIRED>
Object-Oriented Schema
Class A
Attribute A1 Char
Attribute A2 Char
Attribute A_B set (B)
End
Class B
Attribute B1 Char
Attribute B2 Char
Attribute B_A set (A)
Member is B
A
A1, A2
B
B1, B2
n
m
Object-Oriented
MVD: A B
MVD: B A
Classes
A2A1OIDA
a21#1
#2 a22a12
a11
B2B1OIDB
b21#3
#4 b22b12
b11
Stotred OID
#1, #2
#1, #2
Stored_OID
#3, #4
#3, #4
A
A1
A2
B
B1
B2
AB idref
id*
DTD Graph
MVD: A B
MVD: B A
XML Docment
<A A1=”a11", A2=”a21">
<AB idref=”1"></AB>
</A>
<A A1=”a12", A2=”a22">
<AB idref=”1"></AB>
</A>

Figure 3b Data semantics preservation in legacy databases (m:n Cardinality)

A
B A1
A3
A1
A2
schema in EER model
Relational Schema
Relation A(A1 , A2)
Relation B(*A1 , A3)
ID : B.A1 A.A1
A3
a31
*A1
a11
Relations
A
B
A1
a11
A2
a21
a12 a22
a32a21
B
Set
A A1
A2
A1
A3
Network conceptual
Network Schema
Record Name is A
A 1 Character
A 2 Character
Record Name is B
A 1 Character
A 3 Character
Set AB
Owner is A
Member is B
ID : B.A1 A.A1
A1
a11
Records
A
B
A1
a11
A2
a21
a12 a22
a21
A3
a31
a32
A
A1 , A2
B
A1 , A3
Object- Oriented
Object- Oriented Schema
Class A
Attribute A1 Char
Attribute A2 Char
End
Class B subclass of class A
Attribute A1 Char
Attribute A3 Char
End
ID : B.OIDA A.OIDA
A2A1OIDA
a21#1
#2 a22a12
a11
Classes
A1OIDB
#1
#2 a12
a11
A3
a31
a32
B
A A1
A2
A1
A3
DTD Graph
XML schema in DTD
<! ELEMENT A(B?)>
<! ATTLIST A1 # REQUIRED>
<! ELEMENT B EMPTY>
ID : B.A1 A.A1
< A A1=”a11 " A2=”a12"></A>

</A>
< A A1=”a11 " A2=”a22"></A>

</A>
XML document
Figure 3c Data semantics preservation in legacy databases (ISA relationship)

A
B
o
C
A1
A4
A1
A3
A1
A2
schema in EER model
Relational Schema
Relation A(A1 , A2)
Relation B(*A1 , A3)
Relation C(*A1 , A4)
A3
a31
*A1
a11
Relations
A
B
A1
a11
A2
a21
a12 a22
a32a21
C
*A1 A4
a11 a41
a21 a42
B
Set AB
C
A A1
A2
A1
A3
Network conceptual
Network Schema
Record Name is A
A 1 Character
A 2 Character
Record Name is B
A 1 Character
A 3 Character
Record Name is C
A 1 Character
A 4 Character
Set AB
Owner is A
Member is B
Set AC
Owner is A
Member is C
Set AC
RecordsA
B
A1
a11
A2
a21
a12 a22
C
A4
a41
a42
A1
a11
a12
*A1 A3
a11 a31
a21 a32
ID : B.A1 A.A1
ID : C.A1 A.A1
ID : B.A1 A.A1
ID : C.A1 A.A1
B1 B A A1
idref
id
Flattened XML conceptual
schema in DTD Graph
ABC
idref
id
Flattened XML Document schema in DTD
<! ELEMENT ROOT (A , AB , B)>
<! ELEMENT A EMPTY >
<! ATTLIST A id ID # REQUIRED >
<! ATTLIST A A 1 CDATA # REQUIRED >
<! ELEMENT B EMPTY >
<! ATTLIST B id ID # REQUIRED >
<! ATTLIST B B 1 CDATA # REQUIRED >
<! ELEMENT AB EMPTY >
<! ATTLIST AB idref 1 IDREF # REQUIRED >
<! ATTLIST AB idref 2 IDREF # REQUIRED >
<! ATTLIST AB C CDATA # REQUIRED >
(b ) many-to- many cardinality
MVD : A.id B.id
MVD :B.id A.id
Flattened XML Document Data
< ROOT >
< A A1="a11 " id ="1"></A>

<A B C="c11 " idref1="1" idref2="2”></AB>
<A B C="c12 " idref1="2" idref2="1”></AB>
</ ROOT >
B A A1
idref id
schema in DTD Graph
<! ELEMENT ROOT (A , B)>
<! ATTLIST B idref IDREF # REQUIRED >
<! ATTLIST B B 1 CDATA # REQUIRED >
FD : B.iderf A.id
(a ) one-to- many cardinality
B1
< ROOT >
< A A1="a11 " id ="1"></A>


</ ROOT >
Figure 4a Data semantics preservation in flattened XML documents (1:n & m:n cardinalities)

< ROOT>
< A A1="a11 " id="A1.1"></A>

</ ROOT>
B A
idref id
schema in DTD Graph
(c ) isa relationship
A1A1
<! ELEMENT ROOT (A , B)>
<! ATTLIST B A 1 CDATA # REQUIRED >
ID : B.idref A.id
A1A C A1
idref
schema in DTD Graph
B
idref
id
(d ) generalization
A1
< ROOT>
<A A1="a11 " id ="1"></A>
<A A1="a12 " id ="2"></A>

<C A1="a11 " idref=2"></C>
</ ROOT>
<! ELEMENT ROOT (A,B,C)>
<! ATTLIST A A1 CDATA # REQUIRED >
<! ATTLIST B A1 CDATA # REQUIRED >
<! ELEMENT C EMPTY >
<! ATTLIST C idref IDREF # REQUIRED >
<! ATTLIST C A1 CDATA # REQUIRED >
ID : B.idref A.id
ID : C.idref A.id
Figure 4b Data semantics preservation in flattened XML documents
(ISA, generalization)
2. RELATED WORK
On data transformation
Shoshani, A.[1] defined logical level approach data conversion by using source and target
schemas to perform data type conversion instead of physical data type conversion. He provided a
general methodology of using logical level approach of downloading source legacy database into
sequential file and uploading them into target legacy database for data transformation.
Lum et al [2] showed how to construct data conversion languages SDDL and TDL to extract and
restrict data from source legacy database into target legacy database.
The above two paper differ from this paper such that they apply sequential file as medium for
legacy databases exchange, but this paper applies flattened XML document as medium for legacy
databases exchange.
Fong and Bloor [3] described mapping navigational semantics of the network schema into a
relational schema before converting data from network database to relational database.

Fong [4] presented a methodology of transforming object-oriented database objects into
Relational database by using SQL Insert statements.
Fong and Shiu [5] designed a Semantic Export Markup Language as a data conversion language
to export component of relational database into XML database.
Fong et al.[6] applied logical level approach for data materialization between relational database
and object-oriented database using sequential file as medium.
On Heterogeneous database
Given huge investment for a company put into heterogeneous databases, it is difficult for
company convert them into homogeneous databases for new applications. Therefore, researchers
have come up with a solution of universal databases that can be accessed as homogeneous
databases by the user [7] . For instance, we can provide an relational interface to non‐relational
database such as Hierarchical, Network, Object‐Oriented and XML [8] .
Hsiao & Kamel [9] offered a solution of multiple-models-and-languages-to-multiple-models-and–
languages mapping to access heterogeneous databases.
Their papers propose a universal database for Hierarchical, Network and Relational databases
while this paper proposes a universal database for Network, Relational, Object-Oriented and
XML databases.
On Universal database
Fong et al. [10] applied universal database system to access universal data warehousing for the
integration of both relational databases and object-oriented databases with star schema and OLAP
functions.
Silverston & Graziano [11] used a universal data model in a diagram to design the conceptual
schema of different legacy data models of any legacy database.
The above papers differ from this paper such that the above paper proposes a universal data
model diagram for universal database conceptual schema while this paper proposes using DTD
Graph in Figure 3 and 4 as conceptual schema for universal database.
On Homogeneous database
Sellis, Lin & Raschid [12] presented a solution to decompose and store the condition elements in
the antecedents of rules such as those used in production rule-based systems in homogeneous
databases environment using relational data model.
This paper differs from the above paper such that it uses flattened XML document environment
for universal database.
On schema translation
Funderburk et al. [13] proposed DTD Graph as XML conceptual schema which is identical to
DTD, but in a graph format.
On Cloud Database
Derrick Harris [14] defines cloud database as databases in virtual machines.
This paper plans to include cloud computing for further research in future.
On Flattened XML document
Fong et al. [15] converted an XML document into Relational database by transforming XML
document into flattened XML document with relational table structure by Extensible Stylesheet
Language Transformation.
Compared with the above references, this paper has 3 uniqueness:
(1) Cover more data model
All other database research paper only involve 2 or 3 data models in the universal database. This
paper involves 4 data models such as Network, relational, object-oriented and XML.
(2) Use cloud platform

The reference papers do not use cloud platform to implement universal database. This paper
performs prototype in cloud platform.
(3) Flattened XML document as middleware
This paper applies flattened XML document as medium to transform legacy databases among
each other which is not done by other research papers.
Above all, this paper extends the work of universal database into an “open” universal database
gateway such that the universal database is not limited to a particular DBMS, but can be any
legacy database of user’s choice. Similarly, OUDG is more user friendly than ODBC because it
requires less programming effort.
3. METHODOLOGY OF OUDG AS ODBC SUPPLEMENT
This paper proposes OUDG as a database middleware to access legacy databases via flattened
XML documents as an universal database as follows:
First Legacy databases Phase 1: Second Flattened XML documents (universal database)
Phase 2:Third Legacy databases
The major functions of OUDG are:
Phase I: Transform first legacy databases into flattened XML documents
Any one of the four first legacy database can be transformed into the flattened XML document as
follows:
Case 1: Transform first legacy relational databases into second flattened XML documents
Firstly, we perform the preprocess of mapping relational schema into flattened XML schema.
Secondly, we perform their correspondent data conversion. The input is a relational database and
the output is an flattened XML document. The system will read relational table according to the
legacy relational schema. In one-to-many data semantic, it will post parent and child relations into
2 table structured sibling XML elements linked with id and idref. In many-to-many data semantic,
it will post 2 relations and their relationship relation into 3 table structured XML sibling elements
linked with idref(s) and id(s). In isa data semantic, it will post superclass and subclass relations
into 2 table structured XML sibling elements linked with id and idref with the same key as shown
in Figure 3 and Figure 4.
Case 2: Transform first XML databases into second flattened XML documents
Firstly, we perform the preprocess of mapping XML schema into flattened XML schema.
Secondly, we perform their correspondent data conversion. The input is an XML database and the
output is a flattened XML document with relational table structure. The system will read XML
document according to the XML schema. In one-to-many data semantic, it will post element and
sub-element into 2 XML sibling elements linked with id and idref. In many-to-many data
semantic, it will post 3 elements linked with id(s) and idref(s) into 3 XML sibling elements linked
with id(s) and idref(s). In isa data semantic, it will post superclass and subclass elements into 2
XML sibling elements linked with id and idref with the same key as shown in Figure 3 and Figure
4.
Case 3: Transform first legacy Object Oriented database into second flattened XML document
Firstly, we perform the preprocess of mapping object-oriented schema into flattened XML
schema. Secondly, we perform their correspondent data conversion. The input is an OODB and
the output is a flattened XML document. The system will read OODB according to OODB
schema. In one-to-many data semantic, it will post object and set of associated objects into 2
XML sibling elements linked with id and idref. In man-to-many data semantic, it will post 2 sets
of associated objects with a common object into 3 XML sibling elements such that a sibling

element with 2 IDREF(s) referring 2 sibling elements with 2 ID(s)). In isa data semantic, it will
post superclass and subclass objects with same OID into 2 XML sibling elements linked with id
and idref with the same key as shown in Figure 3 and Figure 4.
Case 4: Transform first legacy Network databases into second flattened XML documents
Firstly, we perform the preprocess of mapping network schema into flattened XML schema.
Secondly, we perform their correspondent data conversion. The input is a Network
database(NDB) and the output is a table structured flattened XML document. The system will
read NDB according to NDB schema. In one-to-many data semantic, it will post owner and
member records into 2 XML sibling elements linked with id and idref. In many-to-many data
semantic, it will post 2 owners and 1 common member records into 3 XML sibling elements
linked with id(s) and idref(s). In isa data semantic, it will post an owner and a member records
into 2 XML sibling elements linked with id and idref with the same key as shown in Figure 3 and
Figure 4.
Phase II: Transform second flattened XML documents into third legacy databases
In step 2, we map the flattened XML schema into another legacy database schema, followed by
the data transformation of the flattened XML documents into a legacy database according to the
mapped legacy database schema. In this way, each source database data type can be read by the
legacy database schema. Therefore, there is no need for physical data type conversion in this
approach. Therefore, we can post the flattened XML document into a legacy database of
relational, object-oriented, network or XML.
Case 5: Transform second flattened XML documents into third relational databases
Firstly, we perform the preprocess of mapping flattened XML schema into relational schema.
Secondly, we perform their correspondent data conversion. The input is a flattened XML
document and the output is a relational database. The system will read flattened XML document
according to flattened XML document schema. In one-to-many data semantic, it will post 2 XML
sibling elements into parent and child relations. In many-to-many data semantic, it will post 3
XML sibling elements linked with id(s) and idref(s) into 2 parents and 1 child relations. In isa
data semantic, it will post 2 XML sibling elements into superclass relation and sub-class relation
as shown in Figure 3 and Figure 4.
Case 6: Transform second flattened XML documents into third object-oriented databases
Firstly, we perform the preprocess of mapping flattened XML schema into object-oriented
schema. Secondly, we perform their correspondent data conversion. The input is a flattened XML
document and the output is an object-oriented database. The system will read flattened XML
document according to flattened XML document schema. In one-to-many data semantic, it will
post 2 XML sibling elements into 2 associated objects with OID and Stored OID. In many-to-
many data semantic, it will post 3 XML sibling elements linked with id(s) and idref(s) into 3
associated objects. In isa data semantic, it will post 2 XML sibling elements into 2 superclass and
sub-class objects as shown in Figure 3 and Figure 4.
Case 7 Transform second flattened XML documents into third network databases:
Firstly, we perform the preprocess of mapping flattened XML schema into network schema.
Secondly, we perform their correspondent data conversion.
The input is a flattened XML document and the output is a network database. The system will
read flattened XML document according to flattened XML document schema. In one-to-many
data semantic, it will post 2 XML sibling elements into 2 owner and member records. In many-to-
many data semantic, it will post 3 XML sibling elements linked with id(s) and idref(s) into 2
owners linked with 1 member record with the same key. In isa data semantic, it will post 2 XML

sibling elements into 2 owner and member record with the same key as shown in Figure 3 and
Figure 4.
Case 8: Transform second flattened XML documents into third legacy XML databases
Firstly, we perform the preprocess of mapping flattened XML schema into XML schema.
Secondly, we perform their correspondent data conversion.
The input is a flattened XML document and the output is an XML document. The system will
read flattened XML documents according to flattened XML documents schema. In one-to-many
data semantic, it will post 2 XML sibling elements into 2 XML element and sub-elements. In
many-to-many data semantic, it will post 3 XML sibling elements linked with id(s) and idref(s)
into 2 pairs of XML elements linked with same sub-element. In isa data semantic, it will post 2
XML sibling elements with the same key into XML element and sub-elements with the same key
as shown in Figure 3 and Figure 4.
4. CASE STUDY
A logistic system records the customer shipment information including which orders are being
packed and what the packing information is. Based on the XML schema below, there are three
intermediate independent entities: PL_INFORMAION recording the general information of the
shipment, PL_LINE_INFORMATION storing the packing information ― particularly
information about the BOXES ― and ORDER_INFORMATION storing the information of
orders such as the product information. A many-to-many relationship between
ORDER_INFORMATION and PL_LINE_DETAIL must be resolved early in the modeling
process to eliminate repeating information when representing PL_INFORMATION or
ORDER_INFORMATION (MySQL 2013). The strategy for resolving many-to-many
relationship[s] is to replace the relationship with two one-to-many cardinality with an association
entity and then relate the two original entities to the association entity. As a result, these two one-
to-many relationships are between PL_LINE_INFORMATION and PL_LINE_DETAIL, and
between ORDER_INFORMATION and PL_LINE_DETAIL. Similarly, the ORDER_INFOR
MATION can be divided into BulkOrder and CustomerOrder in generalization as shown in
Figure 6.
In Figure 6, there are six relations in relational database. Each table has its primary key and
foreign key. Their data dependencies are such that each foreign key determines its referred
primary key in functional dependency (FD) in one-to-many cardinality with foreign key on the
“many” side, and subclass foreign key is a subset of its referred primary key in inclusion
dependency (ID). For relations in many-to-many cardinality, their primary keys are in multi-
valued dependencies(MVD) to each other as follows:
FD: PL_Line_Information.PL_Information_Seqno PL_Information.PL_Information_Seqno
ID: Bulk_Order.BulkOrder.Order_Number Order_Information.Order_Number
ID: TailorMadeOrder.Order_Number Order_Information.Order_Number
MVD: PL_Line_Information.PL_Information_Seqno Order_Information.Order_Number
MVD: Order_Information.Order_Number PL_Line_Information.PL_Information_Seqno
In Figure 7, the relations are transformed into flattened XML document. The input relational
conceptual schema is Extended Entity Relationship model(Chen, 1976). We map input relational
schema into an flattened XML OUDG schema with relational structure in two levels tree. Notice
that the second level elements (under root elements) are linked together using idref referring to id,
which is similar to foreign key referring to primary key. There are seven elements. The second
level elements has id(s) and/or idref(s). Their data dependencies are such that each idref

determines its referred id in FD for one-to-many cardinality, and each subclass idref is a subset of
its superclass id in ID. For elements in many-to-many cardinality, their id(s) are in MVD as
follows:
FD: idref1 id1
ID: idref3 id3
MVD: id2 id3
MVD: id3 id2
In Figure 8, the flattened XML document is transformed into XML document. Elements
Pl_information and Pl_line_information are in element and sub-element 1:n association. Elements
Pl_line_information and Order_information are in m:n association through element
Pl_line_detail linked by pairs of idref referring to id. Elements Order_information and
Bulk_Order are in isa association. Elements Order_information and TailorMadeOrder are also in
isa association. Their data dependencies are such that each sub-element can determine its element
in FD. Each sub-class key is a subset of its superclass key in ID. Two one-to-many cardinality
with the same element on the “many” side is equivalent to a many-to-many cardinality of the two
“one” side elements in MVD as follows:
ID: BulkOrder.TechnicalOrderNo Order_Information.Order_Seqno
ID: TailorMadeOrder.CustomerOrderNo Order_Information.Order_Seqno
MVD: PL_Line_Information.PL_Information_Seqno Order_Information.Order_Seqno
MVD: Order_Information.Order_Seqno PL_Line_Information.PL_Information_Seqno
In Figure 9, the flattened XML document is transformed into Object-Oriented database. It shows
the mapping of flattened XML schema into UML as object-oriented conceptual schema. There
are six classes. Each class has its OID (object identity), which is similar to primary key in
relational schema, and Stored OID, which is similar to foreign key in relational schema. Their
data dependencies are such that each Stored OID key determines its referred OID in FD, and each
subclass OID is a subset of its superclass OID in ID. The class PL_Informationa nd class
PL_Line_Information are in 1:n association in FD. Classes PL_line_Informaton and class
Order_Information are in m:n association through class PL_Line_Detail. Subclass BulkOrder and
subclass TailorMadeOrder are in generalization under same superclass Order_Information in ID
as follows:
FD: PL_Line_Information.Stored_OID PL_Information.OID
ID: Bulk_Order.OID Order_Information.OID
ID: TailorMadeOrder.OID Order_Information.OID
MVD: PL_Line_Information.OID Order_Information.OID
MVD: Order_Information.OID PL_Line_Information.OID
In Figure 10, flattened XML document is transformed into Network database. Record
Pl_informations and record Order_information are under network DBMS as first records for
database navigation access path. The path can go from record Pl_information to
Pl_line_information in owner and member record in 1:n relationship in FD. Records
Pl_line_information (owner), Order_information(owner) and Pl_line_detail (member) are in flex
structure such that records Pl_line_information and Order_information they are in m:n
relationship in MVD. Records Order_information and BulkOrder are in isa relationship since they
have same key value in ID. Similarly, records Order_information and TailorMadeOrder are in isa
relationship due to same key value in ID. The set records are pointers only. Their data
dependencies are as follows:
ID: Bulk_Order.TechnicalOrderNo Order_Information.OrderSeqno
ID: TailorMadeOrder.CustomerOrderNo Order_Information.OrderSeqno
MVD: PL_Line_Information.PL_Information_Seqno Order_Information.OrderSeqno

MVD: Order_Information.OrderSeqno PL_Line_Information.PL_Information_Seqno
Papers in this format must not exceed twenty (20) pages in length. Papers should be submitted to
the secretary AIRCC. Papers for initial consideration may be submitted in either .doc or .pdf
format. Final, camera-ready versions should take into account referees’ suggested amendments.
PL_INFORMATION
PL_LINE_INFORMATION
PL_LINE_DETAIL
ORDER_INFORMATION
BulkOrder TailorMadeOrder
Pl_information_seqno
Pl_line_information_seqno
Order_number
1..1
1:m
1:m 1:m
Order_number Order_number
MappedObject-Orientedconceptualschema

5. CONCLUSION
Since relational database is the most user friendly legacy database, and XML database is the most
portable database for information highway on the Internet. In this project, we propose a Flattened
XML database as universal database such that it is most user friendly and portable as a database
middleware for all legacy databases.
The uniqueness of this paper are:
(1) Openness of an universal database: The reason we choose flattened XML document is due to
its openness, and DBMS independence. All other data models are DBMS dependent.
Nevertheless, users can use OUDG to access any legacy database via flattened XML documents
on the Internet by Internet Explorer without programming.
(2) Recovery of legacy database: Since flattened XML document is a replicate legacy database, it
can be used to recover any legacy database whenever the production legacy database is down. As
a result, replicate XML document can be parallel processing with legacy database in non-stop
computing.
(3) Heterogeneous database integration for data warehousing: By transforming all in-house legacy
databases into one legacy database as the data cube, companies can use OUDG to integrated their
legacy databases into a data warehousing for decision support system.
(4) Portability of Flattened XML document as Universal database: The OUDG solution is not
limited to using a particular DBMS, but allows users of any legacy database access other legacy
database.
In summary, the proposed OUDG unites all legacy databases data models into flattened XML
schema. The portability of the proposed flattened XML document can be transferred into any
open platform. The data conversion methodology of this OUDG is to download the raw data of
source database into flattened XML document using source database schema, and upload the
flattened XML document into target database using translated target database schema, which is a
logical level approach, and which can avoid physical data type conversion. Therefore, the
methodology can transform any legacy database into any other legacy database. The reason of
using flattened XML document as medium is to reduce the number of programs for the data
conversion; otherwise, we need 4 * 4 = 16 programs, instead of the current 4 + 4 = 8 programs to
do the data conversion for the four legacy database models: relational, network, object-oriented
and XML. Above all, all legacy databases can be transformed into each other via flattened XML
documents for data access in the same way as computers connect to each other via computer
network for information retrieval.
REFERENCES
[1] Shoshani, A., (1975) “A Logical-Level Approach to Data Base Conversion”, ACM SGMOD
International Conference on Management of Data, pp.112-122.
[2] Lum, V.Y., Shu N.C. & Housel B.C. (1976) “A General Methodology for Data Conversion and
Restructuring”, IBM Journal of research and development, Volume 20, Issue 5, pp.483-497.
[3] Fong, J.& Bloor C. (1994) “Data Conversion Rules from Network to Relational Databases”,
Information and Software Technology, Volume. 36 No. 3, pp. 141-154.
[4] Fong, J. (1997) “Converting Relational to Object-Oriented Databases”, SIGMOD RECORD, Volume
26, Number 1, pp53-58.
[5] Fong, J. & Shiu, H. (2012) “An interpreter approach for exporting relational data into XML
documents with Structured Export Markup Language”, Journal of Database Management, volume 23,
issue 1.
[6] Fong, J., Pang, R., Fong, A., Pang, F. & Poon, K. (2003) “Concurrent data materialization for object-
relational database with semantic metadata”, International Journal of Software Engineering and
Knowledge Engineering, Volume 13, Number 3, pp.257-291.
[7] Fong, J. & Huang, S. (1999) “Architecture of a Universal Database: A Frame Model Approach”,
International Journal of Cooperative Information Systems, Volume 8, Number. 1, pp. 47-82.
[8] Fong, J. (1996) “Adding Relational Interface to Non-relational Database",IEEE Software, pp. 89-97.

[9] Hsiao, D. & Kamel, M. (1989) “Heterogeneous Databases: Proliferations, Issues, and Solutions”,
IEEE Transactions on Knowledge and Data Engineering, Voumn 1, No. Pp.45-62.
[10] Fong, J., Li, Q. & Huang, S. (2003) “Universal Data Warehousing Based on a Meta-Data Modeling
Approach”, International Journal of Cooperative Information Systems, Volume 12, Number 3,
pp.325-363.
[11] Silverston, L. & Graziano, K.(2013) www.360doc.com/content/08/0830/01/1032_1590731.shtml
[12] Sellis, T., Lin, C. & Raschid, L. (1993) “Coupling Production Systems and Database Systems: A
Homogeneous Approach”, IEEE TRANSACTIONS ON KNOWLEDGE AND DATA
ENGINEERING, VOL. 5, NO.
[13] Funderburk J. (2002) “XTABLES: Bridging Relational Technology and XML”, IBM Systems
Journal, Vol 41, No. 4, PP 616 –641.
[14] Harris, D. (2012) “cloud-databases-101-who-builds-em-and-what-they-do”, GIGAOM,
http://gigaom.com/cloud/cloud-databases-101-who-builds-em-and-what-they-do/
[15] Fong, J., Shiu, H. & Wong, J. (2009) “Methodology for data conversion from XML documents to
relations using Extensible Stylesheet Language Transformation”, International Journal of Software
Engineering and Knowledge Engineering,Volume 19, Number 2, pp. 249-281
AUTHORS
Dr Joseph Fong is an Associate Professor at City University of Hong Kong. He is a
fellow of Hong Kong Computer Society, the founder chairman of the Hong Kong
Computer Society Database Special Interest Group, and the honorable founder
chairman of Hong Kong Web Society and International Hybrid Learning Society. Fong
had worked in the industry in US for 11 years, and in Hong Kong as an academician
since 1988. His research interests are in database, data warehousing, data mining, XML
and eLearning. His above 100 publications include SCI Journals, Conferences, Patent
(US), books, and an authored text book on "Information Systems Reengineering,
Integration and normalization" 3rd edition by Springer in 2015. He had been program
manager of M.Sc. of Staffordshire University for a decade and teaches Data
warehousing and data mining, Data Engineering, and Database Systems. Dr. Fong is a former editorial
board member of International Journal of Web Information Systems.
Mr Wong Ting Yan, Kenneth. has been graduated from the first degree, Computer
Engineering in Electronic Engineering Department in City University of Hong Kong at
2002, and Master of Science in Information Engineering in Chinese University of Hong
Kong at 2008, and Degree of Master of Philosophy in Computer Science in City
University of Hong Kong in 2014. He has worked in several educational institutes for
almost 10 years, included primary school, secondary schools, and has experienced to
work as teaching assistant and research associate in Open university of Hong Kong and
City University of Hong Kong respectively.

Oudg cross model datum access

More Related Content

What's hot (18)

Viewers also liked (20)

Similar to Oudg cross model datum access (20)

Recently uploaded (20)

Oudg cross model datum access