Part II: Structured design methods
and representations
(構造化設計手法と設計の表現)
2. Data-driven design notation: Data
Flow Diagrams
(データフロー図)
Data flow diagrams are a tool for program design, with the
following features:(データフロー図は、以下の特徴を持つ
設計ツールである)
It is suitable for abstract design, since it offers an intuitive and
comprehensible notation for describing architecture of program
systems.(抽象設計に適用する)
It is a structured methods, supporting top-down, modular, and
hierarchical designs.(構造化設計手法である)
It concentrates on data flows among processes rather than on
control flows among processes. (データの流れに重視する)
It is a data-driven approach to program design.(データ駆動の設
計)
Let us first have a look at a simple example:(事例)
The data flow diagram for evaluating the arithmetic
expression: (次の数式を計算するプロセスをデー
タフロー図で表現する)(a + b) * (c + d) + f
a
Add1
b
sum1
f
p
Multiply
sum2
c
Add2
d
res
Add3
Synonyms of Data Flow Diagrams:(データフロー図の同意
語)
Bubble chart(バブル図)
DFD (the abbreviation for Data Flow
Diagram)
Bubble diagram(バブル図)
Process model(プロセスモデル)
Function model(機能モデル)
Application domain:(適用なドメイン)
DFD is suitable for the design of operational systems in
which the functions of the systems are more important
and more complex than data that the system deals with.
For example, banking systems, customer service systems,
railway ticket systems, and so on. (操作がデータより重
要であるシステムの設計に有効である)
The contents to be introduced:(紹介する内容)
The components of DFDs(DFDの部品)
Decomposition of processes(プロセス分解)
Guidelines for drawing hierarchical DFDs(階層
的なDFDの構築の指針)
2.1 The components of DFDs
(DFDの部品)
There are four kinds of components in DFDs:
(DFDは四種類の部品を持つ)
Process(プロセス)
Data flow(データフロー又はデータの流れ)
Data store(データ倉庫)
Terminator(ターミネーター)
2.1.1 The process
(プロセス)
A process realizes a transformation from inputs to outputs.
It represents an operation, and once executed it performs
some function in terms of consuming its input and
generating its output. (プロセスは入力により出力を生
成する操作である)
Synonyms of process:(プロセスの同意語)
Bubble(バブル)
Function(機能)
Transformation(変換)
Operation(操作)
The graphical representation:(図で表す)
Name
For example,
Arrange
Lectures
Important: processes are meaningful
only when they model a transformation
of data flows. That is, without connecting
to data flows, processes are not useful.
Register
Projects
(データフローの変換を表すプロセスだけが意味がある)
Guidelines for naming processes: (プロセス名の付け方)
A process can be named in one of the following two ways:
Verb-object phase, such as Arrange Lectures, Register
Projects. (動詞ー目的語の語句)
The name of an organization or entity that processes data.
For example, Student Office, Addition, Subtraction. (組
織名又は実体名)
Important: the name of a process should indicate the
potential function of the process. This would help
improve the readability of DFDs. (プロセス名は、その
プロセスの機能を示すべきである)
2.1.2 The data flow
(データフロー)
A data flow describes a data item moving from one part to
another of the system.(データフルーはデータ項目の移
動を表すものである)
The graphical representation: (図で表す)
a
Where a is the name of this data flow.(aはそのデータフ
ロー名である)
Guidelines for naming data flows:(データフロー名
の付け方)
The name should normally be a noun phrase, for
example, student, book, etc.(名詞語句)
The name should indicate the type of the
information. For example, student, book. (データ
の情報を示すべきである)
The name is usually written in lower case letters.
Data flows are usually used together with
processes; that is, they are used to connect
processes, indicating that data items are moving
from one process to another. (データフローはプ
ロセスと一緒に使われる)For example,
Sort
Records
sorted-records
qualified-student-records
study-records
Divide
Records
failed-student-records
Give
Recommendations
records-withrecommendations
Important points about data flows
(データフルーについての要点 )
A data flow should carry one type of information. We
should not name a data flow as: student and professor
and administrator. (データフルーは一種類の情報を表
す)
A data flow can carry a composite data item. For
example, university-member, which may consists of the
three components: student, teacher, and administrator.
(データフローは複合なデータ項目の表示ができる)
Data flows show directions.(データフローは方向を示
す)
Data flows allow diverging and converging. (データフ
ローは分岐 又は収束することができる)
For example,
student-records
Faculty
Office
University
Student
Office
Professor
Diverging data flow
Converging data flows
Provide
Student
Data
student
Provide
Professor
Data
professor
administrator
Provide
Administrator
Data
Classify
Data
classified-data
Data flows do not answer many procedural
questions. For example, whether, when, and how
input data flows of a process cause an execution
of the process are uncertain. Whether, when, and
how the process produces its output data flows are
not defined. (データフローは手続き的情報を示
していない)
a
c
P
b
d
e
2.1.3 The store
(データ倉庫)
A store describes a single or a set of data items at rest.
(データ倉庫は動かない状態の一つ又は幾つかのデータ
項目を表す)
For example, a file, database can be modeled as a store.
The graphical representation: one of the following
representations can be used:(データ倉庫を図で表す)
representation (a) only gives the name of the store, while
(b) gives both the name and the number that is used to
reference the store in a DFD involving many stores.
name
(a)
n
name
(b)
The use of stores: (データ倉庫の使い方)
students
Faculty
Office
Professor
sorted-records
study-records
Important points about stores:(データ倉庫の要点)
Stores are connected to processes by data flows. (データフローを通
じてプロセスに結ぶ)
A data flow to a store from a process: it means a writing or updating of the
store.(プロセスからプロセスへのデータフローは更新の意味を表す)
A data flow from a store to a process: it means a read of the store by the
process. (データ倉庫からプロセスへのデータフローは倉庫の情報を読まれ
るの意味を表す)
The data flows connected to a store can either be
labeled or unlabeled; they have no substantial
difference in semantics.(データ倉庫とプロセスの間
のデータフルーに名前を付けるか付けないかが構
わない)
The store is passive; data does not travel through
the store unless a process explicitly asks for them.
(データ倉庫は受動的である)
2.1.4 The terminator
(ターミネーター)
A terminator is an external entity with which the
program system communicates. (ターミネーターは
プログラムシステムと交信する外部の実体である)
Typically, a terminator is a person or a group of people, an
outside organization, or government agency. (例えば、
人や組織や政府機関などはターミネーターと成れる)
The graphical representation:(図で表す)
Name
Example of terminator:
Ministry of Education
student-enrollment-data
enrollment-data
request-for-study-records
University
Information
System
Parents
available-study-records
A DFD involving terminators is called Context Diagram.
(ターミネーターが含まれるDFDは文脈図と呼ば
れる)
Important points about terminators:(ターミネーター
についての要点)
They are outside the program under design.(ター
ミネーターは設計されているプログラムの外のも
の)
The designer cannot change the contents of
terminators or the way they work.(設計者は、
ターミネーターの内容や働き方などを変えること
ができない)
Relations between terminators are not shown in
the DFD of the program system.(ターミネーター
らの間の関係はDFDに表されない)
Exercise 2.1
Design a DFD for an ATM(Automated Teller Machine)
system with the following functions:(以下の機能を持つ
ATMシステムの設計をDFDで表現しなさい)
“The ATM system uses four processes: Receive-Request, Check-
Password, Withdraw, and Show-Balance. (ATMシステムは四つ
のプロセスを使う。)
Process Receive-Request takes a request for either withdrawing or
showing the
balance, and sends this request to either process Withdraw or ShowBalance. (プロセスReceive-Request は、「引き出す」又は「残高
を見せる」という命令を受け、その命令をプロセスWithdraw 又
はShow-Balance へ送る)。
At the same time, it also sends an instruction to Check-Password to
check whether the input card and password are correct. (同時に、
プロセスCheck-Password へも入力されたカードのIDとpassword
を検査する指示を送る)
Check-Password then sends a confirmation to both ShowBalance and Withdraw.( Check-Password からShowBalance and Withdraw にIDとpasswordが確認されたの
メセッジを送る)
Withdraw requests an amount from the customer, and
ouptuts the amount of the money if it is within the range
of the account balance. ( Withdraw はcustomer に引き出
すの金額を要請、もしその金額が口座の残高より小さ
いなら、その金額の金を出力する)
To determine whether the requested amount is small
enough to be withdrawn, the system needs to keep the file
accounts that allows all the related processes to access.
(皆プロセスからアクセスできるデータ倉庫accountsが必
要である)
Show-Balance outputs the present balance. ( ShowBalanceは口座の残高を出力する)“
2.2 Decomposition of processes
(プロセスの分解)
Motivation:(動機)
Refine an abstract process into a more detailed DFD so that we
can understand how the input data flows of the process are
transformed into its output data flows.(プロセスをDFDに分解す
ることにより、そのプロセスがどのように自分の入力から出力へ
の変換を行ったということがもっと詳しく分かられる)
Reduce the complexity of the entire system by dividing it into
small, manageable processes.(システムの複雑性を減少する)
Support teamwork concurrently; that is, different people working
on the same system may design different processes at the same
time.(並行なチームワークを支える)
Example of decomposition:
the context diagram of a Personal
Expenses Management System is the top level DFD:
item-reg
reg-confim
single-item-d
Person
total-exp
all-item-d
item-d
Personal Expenses
Management System
total-amount
Output device
(screen, printer)
items-table
Full names for data flows:
item-reg = item-registration
single-item-d = single-item-data
toatl-exp = toatl-expense
all-item-d = all-item-data
reg-confirm = registration-confirmation
item-d = item-data
total-amount = total-amount
items-table = items-table
The decomposition of process Personal Expenses
Management System is the next lower level DFD,
as shown on the next slide.(プロセスPersonal
Expenses Management Systemの分解されたDFD
は次のスライドで表される)
1
shop-name
Input Item
Data
date
price
item-name
item-reg
2
Register
Item
com
3
Output
Confirmation
Message
reg-confirm
expenses-file
single-item-d
4
Input Item
Name
5
Search Item
Information
all-item-d
item-d
Sort Items
item-name
total-exp
6
items-list
7
Compute
Total
Expense
8
Obtain All
Items Data
total-amount
all-items-list
9
Make Item
Table
items-table
The decomposition of process Compute Total
Expense is the following DFD: (プロセスCompute
Total Expense は次のDFDに分解される)
expenses-file
7.2
total-exp
month
7.1
Provide Month
Find All
Related Items
7.3
items-list
Calculate Total
Expense
total-amount
2.3 Guidelines for drawing hierarchical
DFDs
(階層的DFDを描くガイドライン)
A complex program system is usually designed as a
hierarchical DFDs. Such a hierarchical DFDs is
constructed by means of decompositions of high
level processes. There are some guidelines for
drawing hierarchical DFDs that help to ensure the
quality of the hierarchical DFDs.
Consider the abstract description of a hierarchical
DFDs on the next slide. (次の階層的DFDらを考
えましょう)
b
a
T1
T2
The
System
c
T3
Figure 0 the context diagram
a
d
e
b
2
1
f
h
s
4
3
g
c
Figure 1 the decomposition of process The System
d
h
e
j
2.1
2.2
i
3.1
b
3.3
f
Figure 2
3.2
s
g
2.3
h
k
3.4
l
Figure 3
j
2.2.1
h
x
2.2.2
y
2.2.3
Figure 4
b
Guidelines
Use numbers to relate a process to its decomposition,
which is a lower level DFD. (数字を用いて、プロセスと
それの分解されたDFDを結び)For example, process 2
in Figure 1 is decomposed into the DFD in Figure 2, so
the processes occurring in this DFD are numbered as 2.1,
2.2, and 2.3.
Show stores in the decomposition of a process that is
connected to the stores. (データ倉庫にアクセスするプロ
セスの分解されたDFDにもそのデータ倉庫を適当に使
う)For example, since process 3 writes to store s, the
detail of writing to this store by the low level processes in
the decomposition of process 3 must be shown.
A hierarchical DFDs is usually constructed using topdown approach. (階層的DFDを一般にトップーダウン設
計により作成する)That is, the program system is first
abstracted into a top level context diagram, and then
developed further by decomposing processes. This
decomposition process continues until sufficient functions
of the system are designed. However, this does not mean
that top-down approach always works well. In some cases
“middle-out” approach can be more effective. That is, you
first start with designing the middle level DFDs which
you know the best, and then develop them by both
decomposing involved processes and synthesizing those
DFDs into high level DFDs.
The number of processes in a DFD of any level should be kept
around 6, which is usually comfortable to manage, according to
industrial experience. (一つレベルのCDFDの規模としては、約六
つのプロセスを含むのは普通である)
The structural consistency of the hierarchical DFDs must be
ensured. (階層的DFDには、構造の整合性を守ることが重要であ
る)The structural consistency requires: the data flows coming into
and going out of a process at one level must appear as the data
flows coming into and going out of an entire figure at the next
lower level that describes that process. (階層的DFDの構造の整合
性によると、プロセスの入力と出力データフローは、必ずそれの
分解されたDFDのプロセスに使われる)For example, the top level
process The System in Figure 0 of the abstract DFD hierarchy has
one input data flow a, and two output data flows b and c, so a is the
input data flow to the decomposed DFD in Figure 1 and both b and
c are its output data flows.
Input-only and output-only processes must be used
with care. (入力だけ又は出力だけを持つプロセ
スを設計するとき、十分に注意することが必要で
ある)An input-only process means that the
process does not produce any output data flow that
is useful within this system, while an output-only
process means that the process produces output
data flows without the need of any input data flow
(e.g., receive values from input devices, or create
data flows by initializing variables).
Exercise 2.2
Draw a hierarchical DFDs for the ATM system described in
Exercise 2.1. The way suggested to build the DFD
hierarchy is to take the
“middle-out” approach. That is, on the one hand,
synthesize the DFD for Exercise 2.1 into a context
diagram, and on the other hand, decompose some of the
processes in the DFD. For example, Withdraw can be
realized by two processes: Check-Amount and DispenseCash.
(練習問題2.1に設計されたATMシステムの階層的DFD
を描きなさい。そのため、 “middle-out” アプローチの利
用を薦め。例えば、プロセスWithdrawを二つのプロセス
Check-Amount と Dispense-Cashを含むDFDに分解され
る。また、プロセスWithdrawを含むDFDを一つの高いレ
ベルのプロセスに抽象できる。)
Small project
Design a “Personal Information Management System” using
DFDs. This system must contain the following functions:
(階層的なDFDを用いて、「個人情報管理システム」を設計しな
さい。そのシステムは次の機能を提供する。)
Personal Expenses Management(個人費用の管理)
Course Data Management(個人授業データの管理)
Books Management(本の情報の管理)
The system should be designed based on your own
information and functional requirements.
There is no unique answer to this problem. (自分の情報に
よりその情報管理システムを設計しなさい。)
ダウンロード

Program Design (プログラム設計)