FLEXIBLE DATA SEARCHS USING CONDITION FORMULAS

Toshio Kodama

1

, Tosiyasu L. Kunii

2

and Yoichi Seki

3

1

Venture Business Department, Maeda Corporation, 2-10-26 Fujimi, Chiyoda-ku, Tokyo, Japan

2

Morpho, Inc., The University of Tokyo,7-3-1 Hongo, Bunkyo-ku, Tokyo, Japan

3

Software Consultant, 3-8-2 Sakaemachi, Hino City, Tokyo, Japan

Keywords: Cellular model, formula expression, topological space, cellular space, condition formula.

Abstract: Cyberworlds are distributed systems on the Web, and are constantly evolving like living things, creating

value. Currently, numerous Web business applications, such as cyberworld systems are being built, but in

the development of the systems, combinatorial explosion happens because schemas and application

programs must be modified whenever schemas change. We designed and implemented the logic of a

flexible data search function by employing a condition formula on the cellular data system. This is the

starting point to the implementation of the process graph theory, which makes a linear approach to

overcoming combinatorial explosion possible.

1 INTRODUCTION

The system of cyberworlds is a distributed system.

One of the features of cyberworlds is that data

dependencies are constantly changing in them.

Cyberworlds are more complicated and fluid than

any other previous worlds in human history and are

constantly evolving. For example, millions of users

manage their own blog information every day

through Web services on mobile phones, like SNS in

Japan, which is considered one of main elements of

cyberworlds. At the same time, user requirements

for cyberworlds also change and get more

complicated as cyberworlds change. If you analyze

data using the existing technology in business

application development, you have to modify the

schema design and application programs whenever

schemas or user requirements for output change.

That leads to combinatorial explosion, because user

requirements, and their combinations and schemas

must be specified clearly at the design stage in

general business application development. That is a

fundamental problem, so we have to reconsider

development from the data model level.

Is there a data model that can reflect the changes

in schemas and user requirements for output to

analyze data in cyberworlds? We believe that the

cellular model proposed by one of authors (T. L.

Kunii) is the most suitable model. The cellular

model based on the Incrementally Modular

Abstraction Hierarchy (IMAH) can model the

architecture and the changes of cyberworlds and real

worlds from a general level to a specific one,

preserving invariants while preventing combinatorial

explosion (T. L. Kunii and H. S. Kunii, 1999: 19-

21). From the viewpoint of IMAH, existing data

models are positioned as special cases. For example,

UML can model objects at levels below the

presentation level, and in the relational data model, a

relation is an object at the presentation level which

extends a cellular space because it has necessary

attributes in which a type is defined, while the

processing between relations is based on the set

theoretical level. In the object-oriented model, an

object is also the object in the presentation level,

which extends a cellular space, while the relation

between Class is the tree structure, which is a special

case of a topological space. An Object in XML is

considered a special case of a cellular space which

extends a topological space, because an attribute and

a value of it are expressed in the same tag format.

In our research, one of the authors (Y. Seki)

proposed an algebraic system called Formula

Expression as a development tool to realize the

cellular model. T. Kodama has actually implemented

CDS using Formula Expression (Toshio Kodama,

Tosiyasu L. Kunii and Yoichi Seki, 2006: 65-74). In

this paper, we have introduced a new concept of a

21

Kodama T., L. Kunii T. and Seki Y. (2008).

FLEXIBLE DATA SEARCHS USING CONDITION FORMULAS.

In Proceedings of the International Conference on e-Business, pages 21-28

DOI: 10.5220/0001906100210028

Copyright

c

SciTePress

condition formula and its processing maps into CDS.

A condition formula search is a very effective

measure when you want to analyze data in

cyberworlds without losing consistency in the entire

system, since you can search for the data you want

without changing application programs, if you

employ a condition formula search. In addition, we

put emphasis on practical use by taking up an

example. First, we explain the cellular model briefly

and add a new definition to Formula Expression.

(Section 2) Second, we design logical operation as a

condition formula generalizing search conditions of

users by Formula Expression, and design its

processing maps to process a condition formula to

each topological space (Section 3) and implement

them. (Section 4) We demonstrate the effectiveness

of CDS by developing a business application

system, thereby abbreviating the process of

designing and implementing most application

programs. (Section 5) It is a bidding results data

search system where the data of files, which

schemas differ, are inputted without designing

schemas. A more flexible data search is possible by

employing a condition formula search in the system.

2 THE CELLULAR MODEL AND

FORMULA EXPRESSION

The following list is the Incrementally Modular

Abstraction Hierarchy (IMAH) in the cellular model

to be used for defining the architecture of

cyberworlds and their modeling:

1. the homotopy (including fiber bundles) level

2. the set theoretical level

3. the topological space level

4. the adjunction space level (Fig 1)

5. the cellular space level

6. the presentation (including geometry) level

7. the view/projection level

Figure 1: An example of e-manufacturing on an adjunction

space level.

For a detailed explanation of each level, please refer

to our earlier paper.

Formula Expression in the alphabet is the result of

finite times application of the following (1)-(7).

(1) w (∈Σ) is Formula Expression

(2) unit element ε is Formula Expression

(3) zero element φ is Formula Expression

(4) when r and s are Formula Expression, addition

of r+s is also Formula Expression

(5) when r and s are Formula Expression,

multiplication of r×s is also Formula

Expression

(6) when r is Formula Expression, (r) is also

Formula Expression

(7) when r is Formula Expression, {r} is also

Formula Expression

(8) when r is Formula Expression, [r] is also

Formula Expression

In this paper, we have added the 3

rd

bracket [] of (8)

in the definition of Formula Expression. The

algebraic structure is the following.

[r]×(s+t) = [r]×s+[r]×t, (r+s)×[t] = r×[t]+s×[t]

In this way, if [] is added to a formula and

becomes the factor, it behaves like an identifier,

since [] is never removed by any map.

3 THE DEFINITION OF

LOGICAL OPERATION

3.1 A Condition Formula

If users can specify search conditions, data search

will become more functional when searching data

from data storage. Here, we introduce the function

for specifying conditions defining a condition

formula by Formula Expression into CDS. Let

propositions P, Q be sets which include characters p,

q respectively. The conjunction, disjunction and

negation of them in logical operation are defined by

Formula Expression as follows:

1) Conjunction

P∧Q = p×q

2) Disjunction

P∨Q = p+q

3) Negation

¬P = !p

ICE-B 2008 - International Conference on e-Business

22

A formula created from these is called a condition

formula. Here "!" is a special factor which means

negation. Recursivity by () in Formula Expression is

supported so that the recursive search condition of a

user is expressed by a condition formula. An

example is the following.

¬(P∨Q)∧((R∧S)∨(T∧U)) = !(p+q)(r×s+t×u)

3.2 A Quotient Acquisition Map

and a Remainder Acquisition Map

A quotient acquisition map f is a map that has a term

that includes a specified factor, and a remainder

acquisition map g is a map that has a term that

doesn’t include a specified factor. These two maps

are fundamental in processing a condition formula.

(3.3.) If you assume the entire set of terms to be A,

B and the entire set of factors to be C, f:A×C h B and

g:A×C h B. Arbitrary terms r, s, t, u, v, w, x, y (A)

and an arbitrary factor p (C) follow these rules:

f: r, p h φ (when r doesn’t include p)

f: r×p×s, p h r×p×s

f: r×(s+t×p×u+v)×w, p h r×t×p×u×w

f: r×{s+t×p×u+v}×w, p h r×t×p×u×w

f: r×[s+t×p×u+v]×w, p h r×[s+t×p×u+v]×w

g: r, p h r (when r doesn’t include p)

g: r×p×s, p h φ

g: r×(s+t×p×u+v)×w, p h r×(s+v)×w

g: r×{s+t×p×u+v}×w, p h r×{s+φ+v}×w

g: r×[s+t×p×u+v]×w, p h φ

If p is an identifier, f (or g) is usually repeated until p

is not enclosed in a bracket. Simple examples of

both maps are shown below.

f: a(b(c+d(e+f))+g)h, d h a×b×d(e+f)h

g: a(b(c+d(e+f))+g)h, d h a(b×c+g)h

3.3 A Condition Formula Processing

Map

A condition formula processing map h is a map that

gets a disjoint union of terms which satisfies a

condition formula from a formula. If you assume x

to be a formula and x

i

to be a term which consists of

x (namely

i

x

i

= x) and p, !p, p+q, p×q to be

condition formulas, the images of (x, p), (x, !p), (x,

p+q), (x, p×q), (x, !(p+q)), (x, !(p×q)) by h are the

following:

h(x, p) =

i

f(x

i

, p)

h(x, p×q) =

i

f(f(x

i

, p), q)

h(x, p+q) =

i

f(x

i

, p)+

i

f(g(x

i

, p), q)

h(x, !p) =

i

g(x

i

, p)

h(x, !(p+q)) =

i

g(g(x

i

, p), q)

h(x, !(p×q)) =

i

g(f(x

i

, p), q)+

i

g(x

i

, p)

Figure 2: Images from the condition formula processing

map.

Fig 2 is images of the map h. Here, f is a quotient

acquisition map and g is a remainder acquisition

map. It is obvious that any complicated condition

formula can be processed by the combinations of the

above four correspondences. A simple example is

shown below.

x=

"animal{color+size}(flesheating(bear{brown+big

}+monkey{brown+small}+orangutan{darkbrown

+big}+tiger{brown×black+big}+fox{brown×whi

te+small}+bear{black+big})+grasseating(horse{(

white+brown)+middle}+koala{brown+small}+g

oat{white+small}+hamster{white+verysmall}+p

anda{black×white+big}+zebra{black×white+mid

dle}+giraffe{yellow×black+verybig}+elephant{g

ray+verybig}+mouse{gray+verysmall}))"

FLEXIBLE DATA SEARCHS USING CONDITION FORMULAS

23

Output case 1.

User requirement: "information about a horse and a

zebra in x is required"

A condition formula = "horse+zebra"

h(x, horse+zebra)

=f(x, horse)+f(g(x, horse), zebra)

=animal{color+size}grasseating(horse{(white+bro

wn)+middle}+zebra{black×white+middle})

Output case 2.

User requirement: "information about animals

whose size is big or very big and grass-eating is

required"

A condition formula = "size (big+verybig) grasseat

ing"

h(x, size(big+verybig) grasseating)

=f(f(f(x, size), big+verybig), grasseating)

=f(f(f(x, size), big)+f(g(f(x, size), big), verybig),

grasseating)

=animal×size×grasseating(panda×big+giraffe×very

big+elephant×verybig)

4 IMPLEMENTATION

This system is a web application developed using

JSP and Tomcat 5.0 as a Web server. The client and

the server are the same machine. (OS: Windows XP;

CPU: Intel Core2 Duo, 3.00GHz; RAM: 3.23Gbyte;

HD: 240GB)

Fig 3 is the flowchart of the algorithm of a

quotient acquisition map which is the main function

of a condition formula search. Details are

abbreviated due to the restriction on the number of

pages. In this algorithm, the absolute position of the

specified factor by the function of the language and

the term including the factor are acquired first.

Next, the nearest brackets of the term are acquired

and because the term becomes a factor, a recursive

operation is done.

5 CASE STUDY: A BIDDING

RESULTS DATA SEARCH

SYSTEM

5.1 Outline

We have developed a business application system

using CDS for searching bidding results data for

public construction projects. Many of the data files

Figure 3: The flowchart of the algorithm for a quotient

acquisition map.

were downloaded in CSV format from the official

website of each bureau in the Ministry of Land,

Infrastructure, Transport and Tourism (MLIT,

http://www.mlit.go.jp/chotatsu/kekka/kekka.html) in

Japan. The feature of the files is that their schema

changes little from month to month or from bureau

to bureau. Once you convert the CSV data files to

formulas in CDS, you can unify them into a data

storage file (.txt) by the function of a disjoint union

+. After that, a user can search for the data she/he

wants from the data storage by creating a condition

formula. This system is actually being used in

Maeda Corp. which one of authors (T. Kodama)

belongs to.

5.2 The Space Design

We design a formula for the spaces as follows.

Σfile

i

×code[{Σattribute

i,j

}](Σk[{Σvalue

i,j,k

}])

file

i

: a factor which expresses a file name

attribute

i,j

: a term which expresses an attribute name

of file

i

ICE-B 2008 - International Conference on e-Business

24

value

i,j,k

: a term which expresses a value of an

attribute

i,j

Figure 4: The data structure of the bidding results data

search system.

5.3 Data Conversion and Data Input

In this subsection, we simplify the input data without

losing generality. Let the CSV data of Figure 5 be

bidding results from May, 2007 in the Tohoku

bureau. First, convert the downloaded CSV data to a

formula (formula 5.3-1) as a cell space and add it to

the data storage file.

Formula 5.3-1.

MayOf2007InTohoku×code[{bureau×name+project×name+bid×

date+contract×date+project×kind+bid×kind+bidding×company+

bidding×price}](1[{tohoku+A+12/05/2007+14/05/2007+general

×construction+general×bid+C1+1000000}]+2[{tohoku+A+12/05

/2007+14/05/2007+general×construction+general×bid+C2+1100

000}]+3[{tohoku+A+12/05/2007+14/05/2007+general×construct

ion+general×bid+C3+1100000}]+4[{tohoku+A+12/05/2007+14/

05/2007+general×construction+general×bid+C4+9800000}]+5[{

tohoku+A+12/05/2007+14/05/2007+general×construction+gener

al×bid+C5+1050000}]+6[{tohoku+A+12/05/2007+14/05/2007+

general×construction+general×bid+C6+1100000}]+7[{tohoku+B

+12/06/2007+13/06/2007+general×construction+general×biC×W

TO+C1+50000000}]+8[{tohoku+B+12/06/2007+13/06/2007+ge

neral×construction+general×bid×WTO+C2+51000000}]+9[{toh

oku+B+12/06/2007+13/06/2007+general×construction+general×

bid×WTO+C3+50500000}]+10[{tohoku+B+12/06/2007+13/06/2

007+general×construction+general×bid×WTO+C4+51500000}]

+11[{tohoku+B+12/06/2007+13/06/2007+general×construction+

general×bid×WTO+C5+49000000}]+12[{tohoku+B+12/06/2007

+13/06/2007+general×construction+general×bid×WTO+C6+500

00000}]+13[{tohoku+B+12/06/2007+13/06/2007+general×const

ruction+general×bid×WTO+C7+51000000}])

Figure 5: CSV data of bidding results from May 2007 in

the Tohoku bureau.

Next, add the CSV data from June in the Kanto

bureau, which schema is slightly different from that

in formula 5.3-1, convert it to a formula (formula

5.3-2) in the same way and add it to the data storage

file using + function.

Figure 6: CSV data of bidding results from June 2007 in

the Kanto bureau.

Formula 5.3-2.

(formula5.3-1)+MayOf2007InKanto×code[{bureau×name+proje

ct×name+place+bid×date+contract×date+delivery-month+projec

t×kind+bid×kind+bidding×company+bidding×price}](1[{kanto+

C+p1+15/06/2007+17/07/2007+5+general×construction+general

×bid×WTO+C1+800000000}]+2[{kanto+C+p1+15/06/2007+17/

07/2007+5+general×construction+general×bid×WTO+C2+7800

00000}]+3[{kanto+C+p1+15/06/2007+17/07/2007+5+general×c

onstruction+general×bid×WTO+C3+800000000}]+4[{kanto+C+

p1+15/06/2007+17/07/2007+5+general×construction+general×bi

d×WTO+C4+900000000}]+5[{kanto+C+p1+15/06/2007+17/07/

2007+5+general×construction+general×bid×WTO+C5+6800000

00}]+6[{kanto+C+p1+15/06/2007+17/07/2007+5+general×const

ruction+general×bid×WTO+C6+780000000}]+7[{kanto+C+p1+

15/06/2007+17/07/2007+5+general×construction+general×bid×

WTO+C1+820000000}]+8[{kanto+D+p2+16/06/2007+18/08/20

07+12+electric×facilities+general×bid+C2+60000000}]+9[{kant

o+D+p2+16/06/2007+18/08/2007+12+electric×facilities+general

×bid+C3+58000000}]+10[{kanto+D+p2+16/06/2007+18/08/200

7+12+electric×facilities+general×bid+C4+60000000}]+11[{kant

o+D+p2+16/06/2007+18/08/2007+12+electric×facilities+general

×bid+C5+59000000}]+12[{kanto+D+p2+16/06/2007+18/08/200

7+12+electric×facilities+general×bid+C6+61000000}]+13[{kant

o+D+p2+16/06/2007+18/08/2007+12+electric×facilities+general

×bid+C7+60000000}])

In this way, you can add data to the data storage

after converting it to a formula for a cell space using

FLEXIBLE DATA SEARCHS USING CONDITION FORMULAS

25

+ function. In doing this, you don’t need to consider

differences in schema at all.

In the same way, you can add data from another

organization, which schema is completely different

from others', as a formula for a cell space to the data

storage file.

5.4 Data Conversion and Data Input

When you search for data you want, you create

condition formulas according to requirements and

get an image of the formula in data storage by the

condition formula processing map h, you can get the

data you want. Examples and figures (Fig 7,8) are

shown below.

If you want to search for data for "construction

projects of Company C1 or C2 and for WTO (World

Trade Organization) property", you make the

condition formula "(C1+C2)WTO", and get the

image of formula 5.3-2 by the condition formula

processing map h.

h (formula 5.3-2, (C1+C2)WTO)

=

MayOf2007InTohoku×code[{bureau×name+project×name+bi

d×date+contract×date+project×kind+bid×kind+bidding×compan

y+bidding×price}](7[{tohoku+B+12/06/2007+13/06/2007+gener

al×construction+general×biC×WTO+C1+50000000}]+8[{tohok

u+B+12/06/2007+13/06/2007+general×construction+general×bi

d×WTO+C2+51000000}])+MayOf2007InKanto×code[{bureau×

name+project×name+place+bid×date+contract×date+delivery-m

onth+project×kind+bid×kind+bidding×company+bidding×price}

](1[{kanto+C+p1+15/06/2007+17/07/2007+5+general×construct

ion+general×bid×WTO+C1+800000000}]+2[{kanto+C+p1+15/

06/2007+17/07/2007+5+general×construction+general×bid×WT

O+C2+780000000}])

Next, if you want to search for data for

"construction projects in the Kanto bureau which are

not for WTO", you create the condition formula

"kanto×!WTO", and get the image of formula 5.3-2

by the map h.

Figure 7: The output result by the condition formula

"(C1+C2) WTO".

h (formula 5.3-2, kanto×!WTO)

=

MayOf2007InKanto×code[{bureau×name+project×name+plac

e+bid×date+contract×date+delivery-month+project×kind+bid×

kind+bidding×company+bidding×price}](8[{kanto+D+p2+16/0

6/2007+18/08/2007+12+electric×facilities+general×bid+C2+60

000000}]+9[{kanto+D+p2+16/06/2007+18/08/2007+12+electri

c×facilities+general×bid+C3+58000000}]+10[{kanto+D+p2+1

6/06/2007+18/08/2007+12+electric×facilities+general×bid+C4

+60000000}]+11[{kanto+D+p2+16/06/2007+18/08/2007+12+e

lectric×facilities+general×bid+C5+59000000}]+12[{kanto+D+

p2+16/06/2007+18/08/2007+12+electric×facilities+general×bid

+C6+61000000}]+13[{kanto+D+p2+16/06/2007+18/08/2007+

12+electric×facilities+general×bid+C7+60000000}])

Figure 8: The output result by the condition formula

"kanto×!WTO".

If you want to get attribute values by specifying an

attribute name, you remove "[]" once from the

formula and get the image by the quotient

acquisition map f. An example is shown.

If you want to search for data for "values of an

attribute of bid date in the Tohoku bureau", you get

the image by the composition map of f and h.

Assume that formula 5.3-2' is the formula after

removing all "[]" from the formula 5.3-2.

f (h (formula 5.3-2', kanto), project×name)

=MayOf2007InKanto×code{bid×date}(1{15/06/2007}+2{15/06

/2007}+3{15/06/2007}+4{15/06/2007}+5{15/06/2007}+6{1

5/06/2007}+7{15/06/2007}+8{16/06/2007}+9{16/06/2007}+

10{16/06/2007}+11{16/06/2007}+12{16/06/2007}+13{16/0

6/2007})

5.5 Considerations

When a business application system like the one

above is developed in the existing way, user

requirements are analyzed first. Next, the system,

schemas and application programs are designed

according to requirement analysis. Then,

implementation and testing are done. The

fundamental development process is changed if CDS

is used.

1. Schema Design and Data Input

It is almost impossible for a database designer to

design schema of this application system since

she/he cannot predict the changes in schema of

MLIT bidding results data. And whenever a new file

which schema is different from that already designed

appears, it is actually impossible to modify the

schema design and application programs or to

develop data conversion programs. If you employ

ICE-B 2008 - International Conference on e-Business

26

CDS in the development of this application system,

you don’t have to worry about the above problems.

This is because the concept of the disjoint union + of

the cellular model is supported in CDS, so that you

can add the data which schema are different to the

data storage one after another, if you only have to

convert the data to formulas of CDS.

2. Data Output

Data output design and application programs for

data output have to be done during application

system development, and they have to be modified

when there is a new user requirement for output

which was not expected in the user requirement

analysis. This can be costly. But if you use CDS in

the development, a user only has to create a

condition formula according to a user requirement

for output. This is because user requirements can be

generalized by condition formulas of CDS.

3. Processing Speed

Detailed benchmark tests have not been conducted

yet, but when we actually tried this system, the

output processing speeds of 500 records and 1,000

records from more than 200,000 records were 3.2s

and 6.7s respectively. This system is considered

practical for analyzing business data on a client PC.

6 RELATED WORKS

The distinctive features of our research are the

application of the concept of topological process,

which deals with a subset as an element, and that the

cellular space extends the topological space, as seen

in Section 2. Relational OWL as a method of data

and schema representation is useful when

representing the schema and data of a database

(Takashi Washio and Hiroshi Motoda, 2003: 59-68),

but it is limited to representation of an object that

has attributes. Our method can represent both

objects: one that has attributes as a cellular space

and one that doesn’t have them as a set or a

topological space. Many works applying other

models to XML schema have been done. The

motives of most of them are similar to ours. The

approach in (Giovanna Guerrini, Marco Mesiti,

Daniele Rossi, 2005: 39-44) aims at minimizing

document revalidation in an XML schema evolution,

based on a part of the graph theory. The X-Entity

model (Bernadette Farias Lósio, Ana Carolina

Salgado and Luciano do Rêgo GalvĐo, 2005: 39-44)

is an extension of the Entity Relationship (ER)

model and converts XML schema to a schema of the

ER model. In the approach of (N. Routledge, L. Bird

and A. Goodchild, 2002: 157-166), the conceptual

and logical levels are represented using a standard

UML class and the XML represents the physical

level. XUML (HongXing Liu, YanSheng Lu, Qing

Yang, 2006: 973-976) is a conceptual model for

XML schema, based on the UML2 standard. This

application research concerning XML schema is

needed because there are differences in the

expression capability of the data model between

XML and other models. On the other hand, objects

and their relations in XML schema and the above

models can be expressed consistently by CDS,

which is based on the cellular model. That is

because the tree structure, on which the XML model

is based, and the graph structure, on which the UML

and ER models are based, is special cases of a

topological structure mathematically. Entity in the

models can be expressed as the formula for a cellular

space in CDS. Moreover, the relation between

subsets, as we showed in 3.2, cannot in general be

expressed by XML. Although CDS and the existing

deductive database look alike apparently, the two are

completely different. The deductive database (Q.

Kong, G. Chen, 1995: 973-976) raises the

expression capability of the relational database

(RDB) by defining some rules. On the other hand,

CDS is a proposal for a new tool for data

management and has nothing to do with the RDB.

Plenty of CASE tools are currently available, but

they support system development according to

existing data models. The differences from CDS are

mainly that we apply a novel model, the cellular

model, for building CDS, and that the customer side

can confirm the output by changing formulas using

the defined maps after creating formulas as the input.

7 CONCLUSIONS AND FUTURE

WORKS

In this paper, we have developed a condition

formula search as an important function of CDS.

Using this function of CDS, you can search for data

you want from formulas as data storage by creating a

condition formula according to user requirements, so

that you don’t have to analyze user requirement for

output in typical business application development.

The point we should emphasize for future work is

that the search condition of a user as well as data for

input/output is expressed as a formula. This certainly

brings the system which is developing, including

user requirements recursively. This will be

connected to the implementation of a process graph

FLEXIBLE DATA SEARCHS USING CONDITION FORMULAS

27

(T. L. Kunii, 2003: 86-96). It is the next step where a

situation as a node is transferred to the next situation

selecting a path as an edge. Implementation has been

difficult up to the present time because there is no

tool to realize it, although one of authors (T.L.

Kunii) outlined the plan many years ago. The

appearance of Formula Expression will enable it in

the near future. If we implement the process graph

by developing CDS as future work, automation of

business application development will be done. We

believe that CDS brings great social impact,

changing existing development fundamentally. Our

research is still in its infancy, but it is progressing

every day. We are collaborating with companies and

universities worldwide.

REFERENCES

T. L. Kunii and H. S. Kunii, “A Cellular Model for

Information Systems on the Web - Integrating Local

and Global Information”, In Proc. of DANTE'99, IEEE

Computer Society Press, pp.19-24, 1999.

Bernadette Farias Lósio, Ana Carolina Salgado, Luciano

do Rêgo GalvĐo, “Conceptual modeling of XML

schemas”, In Proc. of WIDM'03, ACM Press, pp.102-

105, 2003.

Takashi Washio, Hiroshi Motoda, “State of the art of

graph-based data mining”, In ACM SIGKDD

Explorations Newsletter, ACM Press, pp.59-68, 2003.

David W. Embley, “Toward semantic understanding: an

approach based on information extraction ontologies”,

In Proc. of ADC'04, Australian Computer Society,

Inc., pp.3-12, 2004.

Cristian Pérez de Laborda, Stefan Conrad,

“Relational.OWL: a data and schema representation

format based on OWL”, In Proc. of APCCM '05,

Australian Computer Society, Inc., pp.89-96, 2005.

Nicholas Routledge, Linda Bird, Andrew Goodchild,

“UML and XML schema”, In Proc. of ADC'02,

Australian Computer Society, Inc., pp.157-166, 2002.

Sean K. Bechhofer, Jeremy J. Carroll, “Parsing owl dl:

trees or triples?”, In Proc. of WWW'04, ACM Press,

Inc., pp.266-275, 2004.

Giovanna Guerrini, Marco Mesiti, Daniele Rossi, “Impact

of XML schema evolution on valid documents”, In

Proc. of WIDM'05, ACM Press, pp.39-44, 2005.

Bernadette Farias Lósio, Ana Carolina Salgado, Luciano

do Rêgo GalvĐo, “Conceptual Modeling of XML

Schemas”, In Proc. of WIDM'03, ACM Press, pp.102-

105, 2003.

HongXing Liu, YanSheng Lu, Qing Yang, “XML

conceptual modeling with XUML”, In Proc. of

ICSE'06, ACM Press, pp.973-976, 2006.

Q. Kong, G. Chen, “On deductive databases with

incomplete information”, In ACM TOIS, pp.354-370,

1995.

Toshio Kodama, Tosiyasu L. Kunii, Yoichi Seki, “A New

Method for Developing Business Applications: The

Cellular Data System”, In Proc of CW'06, pp. 65-74,

IEEE Computer Society Press. http://www.mlit.go.jp/

chotatsu/kekka/kekka.html (22 Apr. 2008)

Tosiyasu L. Kunii, “What's Wrong with Wrapper

Approaches in Modeling Information System

Integration and Interoperability? “, In Proc of

DNIS2003, pp. 86-96, Lecture Notes in Computer

Science, Springer-Verlag.

ICE-B 2008 - International Conference on e-Business

28