JP2006003939A - Software life cycle management method and software life cycle management device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a software life cycle management technique that can support the automation of software development, testing and maintenance by ensuring the interoperability of software components as a linear model. <P>SOLUTION: A topological space level processing block 130 designs programs stored in an object set storage part 190 as topological spaces. An attaching space level processing block 140 sets equivalence relations on the plurality of programs to design an attaching space in which program modules associated with the equivalence relations are attached together. A cellular space level processing block 150 represents the attributes of the modules in cells and designs the dimensions of the cells, and a representation level processing block 160 designs the representations of the cells. A view level processing block 170 designs a view for a software developer or maintainer. A homotopy level processing block 110 designs program changes caused by operations involved in software development, testing and maintenance as a homotopy. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a software life cycle management technique, and in particular, a software life cycle management apparatus and a software life cycle management method capable of maintaining data consistency among a plurality of different object sets, determining an interface, and other processes. About.

  The term “Software Engineering” is said to have been first used at the 1968 NATO Conference. The four items discussed in this meeting, software design method, project management, engineer education, software price and distribution, are still central issues in software engineering (see Non-Patent Document 1).

  Software lifecycle management consists of the following stages: requirements analysis / specification, design, implementation, testing, operation, and maintenance. If the initial requirements analysis is unsatisfactory and the customer's requirements are missed or misunderstood, the customer's requirements will not be correctly reflected in the subsequent system design and implementation. For this reason, it takes a lot of time and effort to maintain the software, such as improving the program or redesigning the system in the case of a large-scale change, in order to correct defects found in the test or operation stage. In the worst case, software that cannot withstand maintenance may be discarded, and development will start over. As described above, if time and effort are spent in the requirement analysis in the initial stage of the software development process, the cost is increased in the maintenance stage, and the total cost increases when the development process is viewed in total.

  One of the traditional software life cycle models is the waterfall model proposed in 1970. In the waterfall model, software products are considered to be developed sequentially from requirements analysis to system design, program design, program implementation, unit / combination testing, system testing, operation, and maintenance. Model development activities.

  The waterfall model is a linear development model, where one stage needs to be completed before starting the next stage, and no iterations across multiple stages are assumed. The waterfall model requires that all required specifications be clearly defined in the first stage and cannot cope with uncertainties. Even if customer requirements change after specification is determined or at the design stage of the program, the waterfall model cannot flexibly respond to new requirements. Since software is usually developed through many iterations through trial and error, a linear development model that does not assume repetition is not practical. Therefore, even though the waterfall model is used as a very large-scale software development model today, it is not generally used. The limitations of these well-known waterfall models have been pointed out in a survey on software engineering recently conducted in the United States (see Non-Patent Document 2).

Information system construction is accelerating from a single large-scale site to a multi-distributed large-scale site, as is typically seen in integrated systems associated with the integration of large banks, for example. However, the current software engineering and re-engineering technology, which is the basic technology, lacks an engineering model as a basic framework for dealing with this. The reason is that the integration method is lacking at the model level, so that the engineers in charge of each site design and construct each module group individually. For this reason, module group specifications are naturally different for each site. If the specifications of the module group are disclosed to each other, reconstruction is possible, but the number of reconstruction systems increases exponentially by the combination of the number of modules, and so-called “development of man-hours” occurs. Moreover, in reality, there are more cases where the specifications of the module group are not disclosed for the purpose of maintaining corporate confidentiality. In this case, although interface creation information between modules is disclosed, the number of interfaces also increases exponentially by the combination, so the aspect of “development of man-hours” becomes even more prominent.
By Shari Lawrence Priger, translated by Taisuke Horiuchi, "Software Engineering Theory and Practice", First Edition, Pearson Education, November 2001 Philip A. Laplante and Colin J. Neill, "The Demise of the Waterfall Model Is Imminent" and Other Urban Myths, ACM QUEUE (Feb. 2004), Vol.1, No.10, pp.10-15.

  Smooth software lifecycle from requirements analysis to system design, implementation, testing, and maintenance due to the lack of a theoretical engineering model to support efficient software development and maintenance The flow of software is slow, software development is delayed, and unforeseen situations are likely to occur at the operation stage, making it difficult to manage risk and increasing the cost of software maintenance.

  In Japan and overseas, relational database (RDB) models, ER models, XML with tags, UML with graphs, etc. have been studied and applied, but they are modeled as the theoretical basis for constructing large-scale distributed information systems. For example, the interface specification between modules is not a theoretical system that can be handled uniformly among different sites. Therefore, the above-mentioned “development of development man-hours” cannot be solved.

  In addition, due to the recent severe economic situation in each country, company integration is steadily increasing, but it is possible to freely perform information processing with consistency between different information systems, that is, to ensure interoperability of information systems. As a result, information system operation and maintenance costs are increasing.

  It is said that half to 2/3 of the cost of a large-scale system is maintenance cost, and maintenance is a major cost factor for companies that have introduced the system. In particular, when a malfunction occurs in an operating system, maintenance may become a problem related to the survival of the company, and it involves a great risk. The cost of making changes to the software is relatively low if the changes are made at the required specification decision stage or system or program design stage, but in the case of changes since the software was shipped and operated The cost is tremendous. Therefore, there is a need to find the required specifications and design defects at the earliest possible stage and reduce changes after operation. In addition, changes in the operation / maintenance stage may affect other software parts even if they are partly changed, and the change is complicated and requires careful work. Efficient software maintenance is extremely important in reducing the overall cost of software.

  The present invention has been made from this background, and its purpose is to ensure software interoperability as a linear model and support software development, testing and maintenance automation. It is to provide life cycle management technology.

  One embodiment of the present invention relates to a software life cycle management method. This method uses an incremental and modular abstract hierarchical model that guarantees linear interoperability between software component sets, and consistently extends the life cycle from software development to testing and maintenance. And manage.

  The “software component” is a component constituting software automatically generated by the apparatus, and includes data used by the software, a program constituting the software, and the like.

  Another aspect of the present invention relates to a software lifecycle management apparatus. This device has software development function blocks, test function blocks, and maintenance function blocks using an incremental and modular abstract hierarchical model that guarantees linear interoperability between software component sets. Cycles are managed consistently based on the abstract hierarchical model.

  According to this apparatus, an equivalence relationship is defined between a plurality of software component sets, and the equivalence relationship should be satisfied in stages in each abstract layer while inheriting the equivalence relationship between the abstract layers as an invariant. Specification can be designed. Through the aggregation level, phase space level, adhesion space level, cell space level, representation level, and view level, equivalence relations are guaranteed, and finally the software is generated in a mathematically consistent manner. In addition, each abstract layer is modularized, and even if software parts are added, deleted, and modified, the design can be incrementally changed while going up and down the abstract layer. Furthermore, by handling operations performed on the software component set as homotopy, the software component set can be shifted between the state before the change and the state after the change. While maintaining the software, development processes such as software componentization, reuse, and debugging can be supported.

  According to this device, with regard to software testing, it is tested whether the software generated by the development process conforms to detailed specifications in each abstract hierarchy, and one of them is determined depending on the problem found. The development process can be repeated by returning to the abstract hierarchy, and an efficient test process can be supported using the abstract hierarchy. In addition, regarding software maintenance, when there is a need to modify or extend the function of the operating software, the abstract layer is used by repeating the development process by returning to the abstract layer where the modification or function should be performed. Thus, an efficient maintenance process can be supported.

  It should be noted that any combination of the above-described constituent elements and a representation of the present invention converted between a method, apparatus, system, recording medium, computer program, data structure, etc. are also effective as an aspect of the present invention.

  According to the present invention, it is possible to guarantee the interoperability of an information system and to improve the efficiency of software life cycle management including software development, testing, and maintenance.

  Hereinafter, the present invention will be described based on preferred embodiments. First, an incrementally modular abstraction hierarchy proposed by the present inventor will be described, and then an information processing apparatus equipped with the incremental modular abstraction hierarchy will be described.

[1] Set-theoretic design First, we begin the design of the cyber world by defining a collection of objects to be constructed in cyberspace. In order to be able to perform automated operations using such a computer as an intelligent machine, each group must be a “set”. This is because computers are built as set-theoretic machines. Intuitively, the set X is a set of all objects x having the same property P (x). If this is represented by a symbol, X = {x | P (x)}. Any object included in the set is called a source or element. A set having no elements is called an empty set φ. A set is an open set if all elements are interior. Given the sets X and Y, the computer calculates the union X∪Y, the product or intersection X∩Y, the difference XY (also written x \ y), negation. ) Set theoretical operations like ¬X.

は、Ｕを含むＸのすべての閉部分集合の積である。言い換えれば、Ｕの閉包は、Ｕの外部（exterior）要素でないＸの要素である。Ｘの部分集合｛Ｕ｜Ｕ⊆Ｘ｝のすべてからなる集合をＸの巾（べき）集合といい、２^Ｘと表記する。これはまたＸの離散位相（discrete topology）とも呼ばれる。離散位相はサイバースペースを部分サイバースペースからなるものとして設計する上でたいへん有用である。なお、集合を要素とする集合（いわゆる「集合の集合」）を「集合族」（family of sets）ということがある。Ｘの巾集合は、集合Ｘのすべての部分集合を要素とする集合族である。 Let's start designing the cyberspace architecture from set X as the first cyberspace. Given all the elements u of an unknown cyberspace U, if it is confirmed that they are all elements of cyberspace X, the unknown cyberspace is designated as a subset of X, ie X It is called “partial cyberspace” and written as U⊆X. Whether it is a subset can be automatically checked by processing (∀u) (uεU → uεX). U closure

Is the product of all closed subsets of X including U. In other words, the closure of U is an element of X that is not an exterior element of U. A set consisting of all of the subsets {U | U⊆X} of ^X is referred to as a width (power) set of ^X and is expressed as 2X. This is also called the X discrete topology. Discrete phases are very useful in designing cyberspace as consisting of partial cyberspace. A set having a set as an element (a so-called “set of sets”) is sometimes referred to as a “family of sets”. The width set of X is a set family whose elements are all the subsets of the set X.

[2] Topological design From now on, we will begin to design cyberspace as the sum of X partial cyberspaces and their overlap. Cyberspace designed in this way is generally called phase (topology) space (X, T). Here is a T⊆2 ^X. The design of the phase space is automated with the following specifications:
1) X∈T and φ∈T;
2) For any subscript set J,
∀j∈J (U _j ∈T) → _∪j∈J U _{j ∈T} ;
3) U, VεT → U∩VεT.

Of the above design specifications, 1) is that the empty set and the entire set are elements of T, 2) is the element of T even if a number of T elements are added and summed, and 3) The product (common part) of the two elements also indicates that it is an element of T. The computer designs T from the width set 2 ^{X of X} to satisfy this design specification.

  T is called the phase (topology) of the phase space (X, T). Given two phases T1 and T2, which are T1⊂T2, T1 is said to be weaker or smaller than T2. Conversely, T2 is said to be stronger or larger than T1. T2 may be finer than T1, or T1 may be coarser than T2. As is apparent, the strongest phase is the discrete phase, ie, the width set, and the weakest phase is the empty set φ. For simplicity, if ambiguity does not occur, X is often used instead of (X, T) to represent phase space.

と書く。 How can we say that (X, T) and (Y, T ′) are equivalent when given two phase spaces (X, T) and (Y, T ′)? A criterion is provided here to automatically validate that the spaces are topologically equivalent using a computer. The two phase spaces (X, T) and (Y, T ′) are continuous with a continuous function f: (X, T) → (Y, T ′) and an inverse function thereof. Then, it is said to be topologically equivalent, ie, in-phase, in-phase, or homeomorphic. (X, T) is in phase with (Y, T ')

Write.

How can we prove the continuity of the function f? First, it is checked that ∀B∈T ′, f ⁻¹ (B) ∈T. Here, f ⁻¹ (B) means an inverse image of B by f. Next, it is examined that the following holds.
⇔ f ⁻¹ (B) where B is open is also open at X.

[3] Function When function f: X → Y is given, there are a total function and a partial function. For function f: X → Y, when ∀x∈X, ∃f (x), and only when that is the case, the function f is said to be a global function. The function f: X ′ → Y | X′⊇X is called a partial function. At this time, f (x) does not necessarily exist for every x∈X.

There are three basic types of relationships or mappings for global functions.
1. Injective or into
∀x, y∈X, x ≠ y⇒f (x) ≠ f (y)
In other words,
∀x, y∈X, f (x) = f (y) ⇒x = y
2. Surjective or onto
(∀y∈Y) (∃x∈X) [f (x) = y]
3. Bijective
Be shot and shot

[4] Equivalence relation For any binary relation R⊆X × X on set X, R is
1) If (∀x∈X) [xRx], it is reflective,
2) If (∀x, y∈X) [xRy⇒yRx], there is symmetry,
3) If (∀x, y, zεX) [[xRy⇒yRx] ⇒xRz], it is said that there is transitivity. The conditions 1) to 3) are referred to as reflection rule, symmetry rule, and transition rule, respectively.

  When R has reflectivity, symmetry, and transitivity, R is called an equivalence relation and is represented by “˜”.

これは集合Ｘが空でない互いに素の（すなわち共通要素をもたない）同値類に分割される（分解されるともいう）ことを意味する。 When xεX is given, a subset of X defined by x / ˜ = {yεX: x˜y} is called an equivalent class of x. Here, a class actually means a set, but it is traditionally called a class and is difficult to change now. A set X / ˜ of all equivalence classes is called a quotient space or equalization space of X.
X / ˜ = {x / ˜ε2 ^X | xεX} ⊆2 ^X
From the transition rule, the following holds for each x∈X and x / ˜ ≠ φ.

This means that the set X is divided into disjoint equivalence classes (that is, no common elements) that are not empty (also referred to as being decomposed).

  In general, when the equivalence relation is defined on the set X, a set in which all elements having the same value as a can be considered for an element a of X can be considered. This subset of X is an equivalence class having a as a representative element, and is represented as [a]. Here, no matter which element is included in the equivalence class, the equivalence class having it as a representative element becomes the same set. In other words, equivalence classes do not depend on how to take representative elements. Since equivalence classes satisfy the reflection law, symmetry law, and transition law, each equivalence class is not empty, a belongs to the equivalence class of a, and different equivalence classes have no common elements. Thus, if you collect all equivalence classes, they do not cross each other, and the total union is equal to X. That is, the set X is divided into disjoint unions of equivalence classes. This direct sum division is called classification related to the equivalence relation R of the set X, and the set of all equivalence classes is called a quotient set by the equivalence relation of the set X. A product whose phase is defined in a quotient set is called a quotient space as described above.

  A simple example will be described. If X is a set of rational integers and the relation R between the elements x and y of X is “x is even,” this is an equivalence relation. If the set X is classified by the equivalence relation R, the set X is divided into a direct sum of an odd equivalence class and an even equivalence class.

  In Euclidean geometry, when a set of figures is given, the congruence relation is an equivalence relation, and the whole figure set is divided into direct sums of subsets of congruent figures as a quotient space. The similarity relationship is also an equivalence relationship, and the entire figure set is divided into a direct sum of a subset of similar figures as a quotient space. The congruence relation and the similarity relation are examples of affine transformation. Symmetric relationships in group theory divide the entire set of figures into a direct sum of a subset of symmetrical figures as a quotient space.

Note that in electronic commerce, “being an electronic product” is an equivalence relationship, but “electronic transaction” is a partially ordered relationship. The binary relation S on the set X is a partial order relation when the following three conditions are satisfied.
1) (∀x∈X) [xSx] (reflection rule)
2) (∀x, y∈X) [xSy, ySx⇒x = y] (Anti-Symmetry)
3) (∀x, y, zεX) [[xSy⇒ySz] ⇒xSz] (transition rule)

  A set having a partial order relationship is called a partially ordered set, and is often referred to as a poset. A comparison is said to be possible when the two elements x, y of the set are either xSy or ySx. A partially ordered set that can compare any two elements is called a totally ordered set.

  The relationship between electronic transactions is a semi-order relationship because the seller-buyer relationship is asymmetric and does not satisfy the symmetry rule but satisfies the anti-symmetric rule. On the other hand, the relationship of being an electronic product is symmetric and equivalent because the electronic product for the seller is also the electronic product for the buyer.

  You can define strengths for equivalence relationships. For two equivalence relations R and S in one set X, when xRy⇒xSy, R is stronger than S and S is weaker than R. The classification by R is finer than the classification by S, and the classification by S is coarser than the classification by R. The strongest equivalence relation is an equivalence relation of “same thing”, and the weakest equivalence relation is an equivalence relation in which any two elements of X are equivalent.

[5] Commercial space (equalization space)
Let X be the phase space. A quotient map (often an equalization map (frequently), where f is a surjective and continuous mapping, and each point x∈X maps to an equivalence class x / ~ ∈X / ~ that is a subset containing x It is also called identification map).
f: X → X / ˜
Here, as already explained, “the mapping f: X → Y is surjective”
(∀y∈Y) (∃x∈X) [f (x) = y]
Means.

For a subset X ⁰ ⊆X of X,
⇔ f ⁻¹ (X ⁰ ) | y∈A where X ⁰ is open is open at X (this means that f is continuous)
X / ˜ is called a quotient space (or equalization space) by a quotient map (or equalization map) f. There is a reason why the quotient space is also called equalization space. This is because, as already described, the quotient space is obtained by identifying the equivalence class x / ˜∈X / ˜ that is each element with the point x∈X included in x / ˜.

と表す。逆に集合Ｃは互いに素である２つの集合Ａと集合Ｂに直和分解されるということができる。 [6] Adhesive space Generally, when there is no element common to the set A and the set B, that is, when AφB = φ, it is said that A and B are mutually disjoint. The union set C of the set A and the set B which are relatively prime is called a disjoint union (also called “exclusive OR”),

It expresses. Conversely, it can be said that the set C is directly summed into two sets A and B which are relatively prime.

は、接着写像（attaching map）ｆによってＹをＸに接着（attach）することによって得られる接着空間（attaching space）である。この接着写像ｆは、各点ｙ∈Ｙ_０｜Ｙ_０⊂Ｙをその像ｆ（ｙ）∈Ｘと同一視することにより同値関係を規定するものである。 Now, starting from the phase space X, another phase space Y is bonded thereto. FIG. 1 is a diagram illustrating a state in which an adhesive space is generated from two mutually prime phase spaces X and Y.

Is an attaching space obtained by attaching Y to X by an attaching map f. This adhesion map f defines an equivalence relation by equating each point yεY ₀ | Y ₀ ⊂Y with its image f (y) εX.

は商空間の一例であり、

である。この場合における等化写像（identification map）ｇは、

である。すなわち、等化写像ｇは、互いに素である２つの位相空間Ｘ、Ｙの直和

を、接着写像ｆによって規定される同値関係による同値類の直和

に写像する。 The adhesion map f is a continuous map of f: Y ₀ → X. However, Y ₀ ⊂Y. Thus, the adhesive space

Is an example of a quotient space,

It is. The identification map g in this case is

It is. That is, the equalization map g is a direct sum of two phase spaces X and Y that are relatively prime.

Is the direct sum of equivalence classes according to the equivalence relation defined by the adhesion map f.

To map.

である。ここで
ｉ：Ｘ→Ｙは、包含（inclusion）である。すなわち
∀ｘ∈Ｘ，ｉ（ｘ）＝ｘ
である。 [7] Restriction and inclusion For any function g: Y → Z, the restriction of g to X (X⊆Y) is

It is. Here, i: X → Y is an inclusion. That is, ∀x∈X, i (x) = x
It is.

[8] Extension and retraction of continuous map For phase space X, Y and subspace A⊂X, continuous map f: X → Y is mapped from A to X satisfying f | A: A → Y | A continuous extension (or simply extension). An extension is a partial function.

The restriction r is a continuous extension of identity map 1 _A : A → A that satisfies r: X → A. Therefore, r | A = 1 _A.

を満たすレトラクション（retraction）ｒ：Ｘ→Ａがあるならば、ＡをＸの変形レトラクト（deformation retract）と呼び、

で表す。
もし、Ａがただ一つの点Ａ＝｛ａ｝⊂Ｘであるならば、Ａはレトラクト可能(retractable)といい、

で表す。 For A⊂X,

If there is a retraction r: X → A that satisfies A, then A is called the deformation retract of X,

Represented by
If A is a single point A = {a} ⊂X, then A is said to be retractable,

Represented by

[9] Homotopy Homotopy is an example of an extension. Let X, Y be a phase space, f, g: X → Y be a continuous map, and I = [0, 1]. The homotopy is designed as follows.
H: X × I → Y
However, for t∈I,
When t = 0, H = f
When t = 1, H = g
Holds. At this time, f is said to be a homotopic to g.

Homotopy is an extension of continuous mapping,
H | X × {0} = fi ₀
H | X × {1} = gi ₁
Where i ₀ = X × {0} → X
i ₁ = X × {1} → X

である、すなわち同じホモトピー型をもつように設計する。それは、次の条件が満たされるように設計することでなされる。 Now, the design of the phase space at the homotopy level will be described. The two phase spaces X and Y are homotopically equivalent

That is, it is designed to have the same homotopy type. This is done by designing so that the following conditions are satisfied.

である。ここで１_Ｘおよび１_Ｙは恒等写像、すなわち１_Ｘ：Ｘ→Ｘ、１_Ｙ：Ｙ→Ｙである。 For two successive maps f: X → Y and h: Y → X,

It is. Here, 1 _X and 1 _Y are identity maps, that is, 1 _X : X → X, 1 _Y : Y → Y.

  Although homotopy equivalence is an example of equivalence relation, homotopy equivalence is a broader concept than topology equivalence. Homotopy equivalence can identify changes in information that are no longer topologically equivalent after change. Information changes with various operations and processes, but the operations and processes are specified by homotopy, and changes in information are verified by homotopy equivalence. In fact, from the abstract point of view of invariance, homotopy equivalence is more abstract than set-theoretical equivalence. This is because when a given set is changed by adding or deleting elements, the set can be made homotopy equivalent by saving the operation procedure of addition or deletion and the added or deleted elements. In addition, by performing the cell decomposition described later so as to have the same homotopy value, it is possible to reverse the cell decomposition and return to the state before the decomposition.

[10] Cell structure space (cell space) design A cell (cell) is a phase that is topologically equivalent (ie, in-phase) to a closed sphere (called a closed n-cell) of any dimension (for example, n dimensions, where n is a natural number). Space X.

と表記する。 Closed n-cell

Is written.

と表記する（しばしばｅ^ｎとも書かれる）。 Open n-cell

It referred to as (often written as ^{e n).}

は、ｎ次元閉球体（closed n-dimensional ball）である。ここで

はｎ次元の実数である。 Closed n cell

Is a closed n-dimensional ball. here

Is an n-dimensional real number.

は、ｎ次元開球体（open n-dimensional ball）であり、ｎ次元閉球体

の内部である。

は、閉のｎセルの境界であり、これは（ｎ−１）次元球面（sphere）Ｓ^ｎ−１である。 Open n cell

Is an n-dimensional open ball (n-dimensional ball)

Inside.

Is a closed n-cell boundary, which is an (n−1) -dimensional sphere S ⁿ⁻¹ .

は、連続関数

であり、次を満たす。

For phase space X, characteristic map

Is a continuous function

And satisfy the following.

は、開のｎセルであり、

は、閉のｎセルである。

Is an open n-cell,

Is a closed n-cell.

From the phase space X, a finite or infinite sequence {X ^p | X ^p ⊆X, pεZ} of cells X ^p that are subspaces of X, indexed by the integer set Z, can be constructed. This is called filtration and satisfies the following conditions. X ^p is the covering of X, ie X = ∪ _pεZ X ^p , and X ^p−1 is the subspace of X ^p , ie X ⁰ ⊆X ¹ ⊆X ² ⊆ ... ⊆ X ^{p− 1} ⊆X ^p ⊆... ⊆X (this is called a skeleton). The maximum p-dimensional skeleton is called a p-skeleton.

C = {X ^p | X ^P ⊂X, pεZ} is referred to as cell decomposition of the phase space X, or division of the phase space X into a partial space X ^p that is a closed cell. (X, C) is called a CW-complex.

  When performing cell decomposition while preserving cell adhesion maps, the cell space can be turned into a reusable resource. Such stored and shared information is named a cellular database, and a system that manages the cell database is named a cellular database management system or cellular DBMS.

によって、Ｘ^ｐ−１にｐ次元閉球体の直和（disjoint union）

を接着することによって作られる。 More precisely, according to JHC Whitehead, "Algebraic Homotopy Theory", Proceedings of International Congress of Mathematics, II, Harvard University Press, pp. 354-357, 1950, when phase space X is given, skeleton X ⁰ The filtration X ^p having ⊆X ¹ ⊆X ² ⊆ ... ⊆X ^p-1 ⊆X ^p ⊆ ... ⊆X can be recursively configured as a phase space satisfying the following conditions.
(1) X ⁰ ⊂X is a subspace whose elements are 0-dimensional cells of X, and (2) X ^p is created from X ^{p−1 as} follows. That is, X ^p is a surjective and continuous map called an adhesive map.

Gives the pjoint union of p-dimensional spheres in Xp ^-1.

Made by gluing.

を求め、

における各点ｘ、すなわち

を、連続写像

（ただし、各インデックスｉに対してｘ〜ｆ_ｉ（ｘ）である）
によって、その像

と同一視（identify）することによって、Ｘ^ｐ−１から構成される。 In other words, ^{X p} is a direct sum

Seeking

Each point x in

A continuous mapping

(Where x to f _i (x) for each index i)
By that statue

Is identified by X ^p−1 .

と書くことができ、これは、接着空間の一例である。写像Ｆ_ｉはセル

の接着写像の一例である。 Thus, X ^p is a quotient space (quotient space) or equalization space (identification space),

This is an example of an adhesive space. Map _Fi is a cell

It is an example of the adhesion map.

Filtration space is a space that is homomotoly equivalent to filtration. The phase space X and the skeleton X ⁰ ⊆X ¹ ⊆X ² ⊆ ... ⊆X ^p-1 ⊆X ^p ⊆ ... ⊆X are collectively referred to as CW-space. As a cell complex, as described above, this is called a CW complex.

を等化写像（identification map）の一例として得る。

In this way, the mapping

Is obtained as an example of an identification map.

に対する特性写像（characteristic map）

は、

と書ける。 N cells

Characteristic map for

Is

Can be written.

と書ける。 Embedding X ^p-1 as a closed subspace of X ^p is

Can be written.

  If the CW space is diffeomorphic, this is equivalent to a manifold space.

  A space in which the attributes of an object constructed in the cyber space are expressed by such cells is called a cellular structured space or simply a cellular space. Cell modeling allows cell composition and cell decomposition while maintaining cell dimensions.

  An example of cell dimensions is given. For example, in the cyber world, an object having one attribute cannot be transferred from one attribute to another, so that its degree of freedom is 0, and therefore the cell dimension is 0. This object is represented as a point at the representation level. Here, an attribute is an independent set for designating characteristics and features inherent to an object.

  In an object having two attributes, since the transition from one attribute to the other is possible, the degree of freedom is 1 and the dimension of the cell is also 1. This object can be represented as a straight line at the representation level. Similarly, an object having 3 or 4 attributes has 2 or 3 degrees of freedom, respectively, and can be expressed as a curved surface or a sphere, respectively. In general, an object having n attributes has n-1 degrees of freedom and a dimension of n-1. This can be expressed as a (n−1) -dimensional sphere. The relational model expresses an object having n attributes as a relational schema, and generates an n-column table as an instance thereof. The relational model is based on the Cartesian product of sets, so it can be said to be an expression at the set theory level.

  The cell space is obtained by adding the concept of dimension to the bonding space. In the cell space design, the dimension of the attribute of the object in the bonding space is defined. In the bonding space that is the direct sum of the equivalence classes, when there is a mismatch in the attribute dimensions of the objects included in each equivalence class, a process of adjusting the dimension of the object attributes is performed.

[11] Representation design After the cell dimensions are designed at the cell space level, the representation of the cell is designed at the representation level. In the expression design, the expression format such as the data type and the number of digits of the attribute of the object is detailed. When there is an inconsistency in the expression format of the attribute of the object, a process for matching the expression format with the admissible representation is performed. For example, there is a difference in the data type or the number of digits in the inconsistency in the expression format of the attribute of the object, and the difference can be eliminated by data type conversion or digit number conversion.

[12] View design After the expression format of the cell attribute is detailed at the expression level, the view, that is, the range that can be viewed by the user, is determined. The user has a request to view only a portion of the set of objects of interest in an appropriate format. In addition, the range that the user can refer to must be limited for security reasons. Therefore, in the view design, a subset of objects that can be referred to for the user and a view that restricts the attributes of objects that can be browsed are determined.

[13] Incremental modular abstraction hierarchy The homotopy, set, phase space, adhesion space, cell space, representation, and view abstraction levels described so far form an incrementally modular abstraction hierarchy. To do.
1. Homotopy level Set theory level Phase space level 4. Adhesive space level Cell space level Expression level View level

  These hierarchies are modularized, allowing information to be manipulated incrementally in each hierarchy, and extremely useful for inheriting equivalence relations that indicate the invariant of the cyber world. It is.

  First, a target to be equivalent to a specific element of a plurality of object sets is selected. Next, the phase is defined in the object set, and the equivalence relation is defined in the phase space, whereby the plurality of object sets are designed as an adhesive space that is the direct sum of the equivalence classes. Furthermore, by performing cell decomposition into an equivalent cell and other cells in the cell space, the equivalence relationship is inherited to the cell space level.

  Furthermore, the equivalence relation is inherited up to the expression level by defining the expression form of the cell based on the acceptance expression, and finally the equivalence relation is inherited up to the view level by defining the user's view. If a reception expression level for defining the reception expression is provided as one instance of the expression level, objects can be made equivalent at the expression level.

  In this way, the incremental modular abstract hierarchical model inherits equivalence relations between abstract hierarchies while designing the specifications to be satisfied by the object in each hierarchy, so that information is added, deleted, and modified incrementally. Go back and forth between hierarchies and refine object specifications. In addition, the object designed in this way can be made into parts in a reusable form.

  In a set S that is a subset of the set X, when the union of all sets belonging to S is equal to X, S is called covering of the set X. When integrating n information systems, the “business integration decision” corresponds to a decision to create a set of n sets on an incremental modular abstract hierarchy. Here, the work may be from different institutions. The n sets generally have overlap. For example, the same employee may participate in multiple tasks. Business integration is implemented by inheriting these n sets of coverings between the incremental modular abstract hierarchies. In other words, the business integration is that the business coverage is generated at each of the homotopy level, the phase space level, the adhesion space level, the cell space level, the expression level, and the view level. Once the covering is generated, integration and manipulation between the n information systems is all done through the covering on an incremental modular abstraction hierarchy. As a result, the integration and operation between the n information systems is all linear. As described above, according to the incremental modular abstract hierarchy, linear interoperability can be realized by configuring a linear interface called covering between n information systems.

The phase space in incrementer Li modular abstraction hierarchy model is not limited to the Hausdorff space (Hausdorff space) i.e. T ₂ phase space. In the T ₂ phase space, _two different points can be separated by an open set according to the separation axiom, but even in the T ₀ phase space and the T ₁ phase space where two different points cannot be separated by an open set. Note that space can be designed. Therefore, the abstract hierarchical model is very versatile.

  The bonding space abstracts the characteristics common to leading commercial information systems used by major companies and public institutions so as to be equivalent between different information systems in the form of bonding spaces, and builds a model. As described above, the bonding space is a novel information model capable of linearly integrating different kinds of information systems, and greatly contributes to avoiding combined explosion of integrated work amount. The incremental modular abstraction hierarchy described above is used to automate the generation of linear interfaces after linear integration at the adhesive space level. According to this abstract hierarchy, equivalence relations are inherited down to the lower hierarchy, so it is possible to provide an interface to an existing information system and realize linear interoperability for executing integrated system-scale tasks. .

  For comparison, a relational database system will be described as an example. In a relational database system, a relation is a set of tuples. The relation changes over time as new tuples are inserted into the relation, existing taples are deleted, or attribute values of the tuples are modified. For example, “employee (employee number, name, affiliation, year of hire)” is considered as the relation. When a new employee joins the company, a tuple representing the employee must be inserted into the relation “Employee”. On the other hand, when a certain employee retires, the relation “employee” must delete the tuple representing the employee. Further, when the assignment destination of a certain employee changes due to the rearrangement, the value of the attribute “affiliation” must be corrected in the tuple of the employee.

  In this way, in relational data verses, relations change over time due to the insertion, deletion, and modification of tuples. However, relations have the inherent property of representing employee number, name, affiliation, and year of employment, which is invariant to time. Relations have a structure that does not change with time in this way, and this is called a relation schema.

  Since the relational database model is based on set theory, an equivalence relation cannot be defined at the level of the topological space. Therefore, if a plurality of different databases are to be integrated, it is necessary to redesign the database such as reassigning the ID, which causes a problem of man-hour explosion. On the other hand, in the incremental modular abstract hierarchical model, the equivalence relation defined by the superordinate concept can be reduced to the expression level, so that integration of a plurality of different information systems is possible by linear operation. That is, the interface of each element of the information system can be obtained by converting it to the bonding space only once. Interoperability is ensured by n conversions of n elements, and the number of interfaces is linear.

  As described above, since the conventional relational model does not model the equivalence relation with respect to the abstract hierarchical model of the present invention, the information cannot be handled in a unified manner. At best, another table showing the relationship between the tables is provided in the manner of applying a patch, and an interface is attempted, but this increases the number of interfaces exponentially. On the other hand, SAP and UML (Unified Modeling Language) based on ER (Entity Relation) model are all intuitive graph theory models, and the model itself cannot handle equivalence relations. Problems arise. This is the current state of social infrastructure information systems, and it is a big problem that the system does not have clear interoperability.

[14] Application examples [14.1] Web information modeling Normally, the work to be done by the Web information management system is very close interaction with human recognition through visualizing information using Web graphics. It is to manage information on the Web. The job of Web graphics is to project various images on a graphics screen for human understanding. The human understands the displayed image by linking the image displayed in the display space to a recognition object in the human recognition space. Often, even geometrically accurate representations can lead to human perception errors due to low-priority geometric shapes that are not normally essential information in the cyber world. For Web information management, Web graphics must be able to convey important messages on the screen so that humans can instantly recognize them at a rate that matches the changes in the cyber world. Hereinafter, a simple example will be described.

と表記することができる。一般化のために、ＸとＹを位相空間とする。ファンドＹ_０は金融取引会社Ｙの属性の一部であるから、Ｙ_０⊂Ｙが成り立つ。図２は、Ｗｅｂ上のエレクトロニックファイナンスのプロセスをＷｅｂグラフィックスにおいて図示したものである。次に、ファンドが取引のために特定された後、顧客Ｘは取引会社Ｙとどのようにして関係づけられるかを述べる。ここで提案するＷｅｂ情報モデルは、顧客Ｘと取引会社Ｙの関係を接着写像ｆによって正確に表現し、また、「ファンドＹ_０が取引のために特定されている」という状況を２つの互いに素である位相空間Ｘ（顧客）と位相空間Ｙ（金融取引会社）の接着空間として表現する。この接着空間は、顧客Ｘから始めて、金融取引会社Ｙを顧客Ｘに接着することにより得られる。ここで、ＹのＸへの接着は、連続写像ｆを介して、各点ｙ∈Ｙ_０｜Ｙ_０⊆Ｙをその像ｆ（ｙ）∈Ｘと同一視することによって行われ、その結果、ｘ〜ｆ（ｙ）｜∀ｙ∈Ｙ_０である。このように、〜で表記される同値は、Ｗｅｂ情報の接着空間モデルとして接着空間を作成するために、Ｗｅｂ情報モデリングにおいて中心的な役割を果たす。 [14.2] in Web information modeling Electronic Finance of electronic finance, find customer X is, when you are surfing the Web, the fund Y ₀ can be expected of the revenue that has been posted on the home page of the Web page of the financial transaction company Y Suppose. At this point, since the customer X is in the stage of “window shopping” on the Web, the customer X and the trading company Y have not yet shared the fund. Accordingly, X and Y are relatively prime and have no common elements. That is,

Can be expressed as: For generalization, let X and Y be phase spaces. Since fund Y ₀ is part of the attributes of financial transaction company Y, Y ₀ ⊂Y holds. FIG. 2 illustrates the electronic finance process on the Web in Web graphics. The following describes how customer X is associated with trading company Y after the fund is identified for trading. The Web information model proposed here accurately represents the relationship between the customer X and the trading company Y by the adhesion map f, and the situation that “the fund Y ₀ is specified for trading” It is expressed as a bonding space between the phase space X (customer) and the phase space Y (financial trading company). This bonding space is obtained by bonding the financial transaction company Y to the customer X starting from the customer X. Here, the adhesion of Y to X is performed by equating each point yεY ₀ | Y ₀ ⊆Y with its image f (y) εX via the continuous map f, so that x~f (y) | ∀y∈Y is _0. As described above, the equivalent value represented by ~ plays a central role in Web information modeling in order to create an adhesive space as an adhesive space model of Web information.

  The above bonded space model can be applied essentially equally in electronic manufacturing. Exactly the same technology is required to manage and display the electronic manufacturing process. In order for electronic manufacturing to respond quickly to market demands, it must be specified how components of various sizes can be assembled in a unified design. Web information management system and Web graphics for electronic finance are actually electronic manufacturing, considering that the electronic finance presented above is assembling customers and trading companies as components of electronic manufacturing. It becomes applicable to. If different technologies are used for different applications, the rapidly growing cyber world cannot be managed in a timely manner by a Web information management system or displayed by Web graphics.

[14.3] Web Information Modeling for Electronic Manufacturing Web information modeling for electronic manufacturing using an adhesive space model is quite simple. Basically, manufacturing on the Web called electronic manufacturing is modeled as Web information consisting of the following information relating to the electronic manufacturing process.
1) Product specifications 2) Assembly specifications 3) Parts purchase on the Web 4) Purchases on the assembly site on the Web

Each step is broken down into finer sub-steps as needed. For example, step 1 can be broken down into the following substeps.
1.1) Market research for products in the electronic market 1.2) Product requirements derived from market research 1.3) Product specifications to meet requirements

  The core part of the entire Web technology for electronic manufacturing is the modeling of products and assemblies on the Web as Web information modeling. FIG. 3 clearly illustrates the most basic assembly modeling using a simple assembly example of a chair with two components: a seat and a support. In electronic manufacturing as an advanced manufacturing, product components are modular and can be replaced to buy higher quality components, to make the most effective upgrades or to repair Has been.

  It is clear that electronic finance and electronic manufacturing share the same information modeling based on the equivalent value of the bonding space.

[15] Example FIG. 4 shows a configuration of the software development support apparatus 100 according to the embodiment. The object set storage unit 190 holds a set of objects used for software development. Objects used for software development, that is, software components are data structures and program modules. The software development support apparatus 100 refers to a collection of objects stored in the object collection storage unit 190. The software development support apparatus 100 may acquire and refer to an object set from an information system on the web without using the object set storage unit 190. Hereinafter, in the embodiment, a case where a predetermined correspondence or interface is defined between two object sets will be described for convenience of explanation. More generally, a plurality of object sets are referred to to refer to those object sets. This is just an example of defining an interface between them.

  The software development support apparatus 100 includes a homotopy level processing block 110, an aggregation level processing block 120, a phase space level processing block 130, an adhesion space level processing block 140, a cell space level processing block 150, an expression level processing block 160, and a view level. A processing block 170 is included to perform processing at each hierarchical level in the incremental modular abstract hierarchical model already described.

が確定する。 The set level processing block 120 appropriately performs a set operation on the set of objects stored in the object set storage unit 190 to specify two object sets to be designed. The phase space level processing block 130 defines the phases for the two object sets specified by the set level processing block 120, and the two object sets are designed as phase spaces X and Y, respectively. At this point, the two phase spaces X and Y are relatively prime and are the direct sum or exclusive OR of these phase spaces X and Y.

Is confirmed.

  As shown in FIG. 5 in detail, the bonding space level processing block 140 includes an equivalence relation setting unit 44, an equivalence relation holding unit 52, a corresponding element extraction unit 20, an adhesion mapping acquisition unit 22, and an equalization mapping acquisition unit 24. The procedure acquisition unit 26 and the procedure recording unit 28 are included.

  The equivalence relation setting unit 44 receives an input of equivalence relations to be applied to two object sets from the user, and automatically generates an equivalence relation based on the user input. In order for the user to easily set the equivalence relation, the equivalence relation holding unit 52 stores candidates for the equivalence relation in advance. Further, an interface may be provided in which the user can specify the elements of the object set and describe the equivalence relation.

  The equivalence relation set by the equivalence relation setting unit 44 (hereinafter also referred to as a target equivalence relation) is transmitted to the corresponding element extraction unit 20. The corresponding element extraction unit 20 extracts constituent elements (hereinafter also referred to as “target elements”) for which a correspondence relationship is to be defined in the two object sets based on the target equivalence relationship. At this time, the user may designate one of the target elements in one object set, and the corresponding element extraction unit 20 may extract the corresponding target element from the other object set based on the target equivalence relationship.

  The target elements found to correspond with each other by the corresponding element extraction unit 20 form one partial phase space, that is, an equivalence class, and with the theoretical guarantee of the adhesion space model, the adhesion space is obtained as an exclusive OR of the equivalence classes. It is formed. Since equivalence classes are classified based on invariants defined by equivalence relations, the object set can be objectively classified, and the classification results do not include individual differences by the system designer. When one target element is determined, the other target element is determined. As described above, the correspondence or the number of interfaces is linear, and interoperability is guaranteed.

と定まる。接着空間は、もとは共通部分をもたない位相空間ＸとＹの間に、接着写像ｆを媒介として対応関係を持たせた位相空間である。 When the objective equivalence relationship is determined, the relationship between the target elements is also determined uniquely. Therefore, the adhesion map acquisition unit 22 can specify the adhesion map f indicating the correspondence between the target elements. Once the adhesive map f is determined, the adhesive space is

Is determined. The adhesion space is a phase space in which a correspondence relationship is established between the phase spaces X and Y that originally have no common part, using the adhesion map f as a medium.

  The equalization map acquisition unit 24 acquires an equalization map g, which is a map to the adhesive space determined by the adhesive map f, from the space of the mere exclusive OR of the phase spaces X and Y. The equalization map g can actually be described by the phase space X, Y and the adhesion map f from its definition.

The procedure acquisition unit 26 receives the target equivalence relationship from the equivalence relationship setting unit 44, the target element from the corresponding element extraction unit 20, the adhesion map f from the adhesion map acquisition unit 22, and the equalization map g from the equalization map acquisition unit 24. The target element specified by the user and the target equivalence relationship (hereinafter also referred to as “attention procedure”) are identified as the result of a series of processing, using homotopy.
・ Phase space X, Y
-Objective equivalence relation-Adhesive mapping f
・ Equalization map g
And recorded in the procedure recording unit 28.

  The cell space level processing block 150 includes a cell junction 62, a cell dimension adjusting unit 64, and a correspondence table holding unit 66, as shown in detail in FIG. The cell space level processing block 150 inherits the specifications of a plurality of object sets designed as an adhesion space based on the objective equivalence relation in the adhesion space level processing block 140, and refines the object set specifications at the cell space level. To do. The cell structure space model is obtained by introducing a dimension into the adhesion space model, and specifications regarding the attribute ID of the object and the number of attributes are detailed at the cell space level.

  The cell joining unit 62 joins two target elements that are in a correspondence relationship between two object sets by a target equivalence relation, and expresses attributes of objects belonging to each object set by cells. When there is an inconsistency regarding the attribute dimension between the two corresponding target elements, the cell dimension adjusting unit 64 performs processing for eliminating the inconsistency. As the inconsistency regarding the dimension of the cell, there is a difference in the attribute ID or number of the target element. The cell dimension adjusting unit 64 adjusts attribute IDs and the number of attributes between the two target elements as necessary, and a table indicating the attribute correspondence between the two target elements. To store.

  The expression level processing block 160 includes an expression generation unit 72, an expression format adjustment unit 74, and a conversion table holding unit 76, as shown in detail in FIG. The expression level processing block 160 inherits the specification of the object set detailed at the cell space level in the cell space level processing block 150, and further refines the specification of the object set at the expression level. At the expression level, in order to determine the specific expression format of the cell, the specifications regarding the data type and the number of digits of the attribute of the object will be detailed.

  The expression generation unit 72 generates expression of attributes of objects belonging to each of the two object sets. The expression format adjustment unit 74 performs processing for adjusting the inconsistency when there is an inconsistency in the expression level between two target elements that are in a correspondence relationship due to the objective equivalence relation. As an inconsistency in the expression level, there is a difference in the data type and the number of digits of the attribute of the target element. Data types include Boolean values, signed integers, positive integers, real numbers, double precision real numbers, character types, etc., and the expression format adjustment unit 74 converts these data types to align the data types of the target elements. Also good. If the number of digits, such as the number of digits of a character string or the number of significant digits of a numeric value, is different even if the data type is the same, the expression format adjustment unit 74 adjusts the number of digits of the attribute between the two target elements to align them Also good. Other examples of inconsistencies include unit differences such as weighing and money, and one unit system is converted to the other unit system using a conversion formula.

  When adhering the phase space Y to the phase space X, the representation format adjusting unit 74 at the representation level, the data type and the number of digits of the attributes of the target element of the object set of the phase space X and the target element of the object set of the phase space Y And convert the data type and the number of digits to align with the attribute with higher accuracy. The same applies when adjusting the inconsistency of the attribute expression level among the target elements of three or more object sets, and the expression format adjusting unit 74 is the most common among the attribute expressions of the corresponding target elements of a plurality of object sets. A highly accurate attribute expression is selected, and the attribute expression of another target element with low accuracy is adjusted to match the attribute expression of the target element with high accuracy. As a result, it is possible to prevent the original data from being lost due to data type conversion or digit loss after the target elements having different precisions are bonded.

  The expression format adjustment unit 74 acts as an expression filter having an acceptance expression having the maximum number of characters and digits, and expressions that pass through the expression filter have the same value at the expression level. The expression format adjustment unit 74 stores, in the conversion table holding unit 76, a table describing conversion rules when adjustment of expression mismatch is made.

  The view level processing block 170 sets a view suitable for the user, and displays an object set on a display or the like according to the view.

  The homotopy level processing block 110 ensures that, when each processing block of the software development support apparatus 100 performs an operation on the object set, the change of the object set caused by the operation becomes the homotopy equivalence. The homotopy level processing block 110 includes a table indicating the correspondence between the attention procedure related to the generation of the adhesion space recorded in the procedure recording unit 28, the dimension mismatch of the cells stored in the correspondence table holding unit 66, and the conversion table holding unit 76. By using the table showing the conversion rules of the cell representation stored in the table, the object set can be returned to the state before the change by tracing back the change brought about by the respective operations.

  In this way, starting with the set level processing block 120, the equivalence to the phase space level processing block 130, the adhesion space level processing block 140, the cell space level processing block 150, the representation level processing block 160, and the view level processing block 170 By inheriting the relationship and refining the specifications of the two object sets step by step, the two object sets are output as equivalent values up to the expression level. In this way, the software development support apparatus 100 based on the incremental modular abstract hierarchical model guarantees the equivalence relation from the collective level to the phase space level, the adhesive space level, the cell space level, the representation level, and the view level. The object set that passed through the filter is output as an instance whose equivalence relation is mathematically guaranteed.

  If there are multiple components to be associated, by setting the objective equivalence relationship for each component, an adhesion map is defined in the same way, and the result of finally bonding the corresponding elements between multiple object sets is Is output. If there are n components that one object set is interested in, the number of interfaces is linear for each of the n components because the corresponding relationship with any component of the other object set is specified. The problem of man-hour explosion is solved.

  According to the present embodiment, if an equivalence relationship is set, the components of a plurality of object sets can be associated with each other. The components that do not correspond are left as they are, and the components having the equivalence relationship are always the same. Since it can have information and can be classified and associated with invariants as the starting point, data usability can be improved and the efficiency of software development can be increased in each stage.

[16] Specific Example A specific example of the software development support procedure by the software development support apparatus 100 having the above configuration will be described. When the software development support apparatus 100 develops software based on various program modules, the interface between the program modules is determined and the specification of data used in each program module is detailed. [16.1] describes the data design procedure, and [16.2] describes the program module design procedure.

[16.1] Data Design As an example of a plurality of object sets used in software, an object set 300 of a shoe shop X and an object set 200 of a shoe manufacturer Y are considered. These object sets are designed by the set level processing block 120 and stored in the object set storage unit 190. These object sets may be managed as a database. The phase space level processing block 130 designs the object set 300 of the shoe shop X as the phase space X and the object set 200 of the shoe manufacturer Y as the phase space Y. At this point, the two phase spaces X and Y are disjoint.

  The adhesion space level processing block 140 refers to the object set 300 of the shoe shop X stored in the object set storage unit 190 as the phase space X and the object set 200 of the shoe manufacturer Y as the phase space Y, and based on the equivalence relation. A bonding space is generated from the two phase spaces X and Y. FIG. 8 shows how two phase spaces are bonded by the bonding space level processing block 140.

Now, it purchases a specific athletic shoes Y ₀ to produce the shoe manufacturer Y is a shoe shop X, and sells. In this case, software developers, the object set of shoe shop X, want to treat as the same object of this exercise shoes Y ₀ on the program between the objects set of shoe manufacturer Y. The equivalence relation setting unit 44 sets the athletic shoe Y ₀ that is a target element to be noted between the shoe manufacturer Y and the phase space of the shoe shop X as having an equivalence relation.

The corresponding element extraction unit 20 extracts the athletic shoes f (Y ₀ ) (reference numeral 302) of the shoe shop X corresponding to the athletic shoes Y ₀ (reference numeral 202) of the shoe manufacturer Y. The adhesion map acquisition unit 22 acquires the adhesion map f: Y ₀ → X | Y ₀ ⊆Y based on this correspondence.

に写像する等化写像ｇを取得する。すなわち、等化写像取得部２４は、等化写像

を取得する。ここで、

であり、同図に示すように、運動靴ｆ（Ｙ_０）（符号４０２）において２つの位相空間Ｘ、Ｙが接着する。 The equalization map acquisition unit 24 calculates an exclusive logical sum of the two phase spaces of the shoe shop X and the shoe manufacturer Y by the adhesive map f acquired by the adhesive map acquisition unit 22.

An equalization map g that maps to is obtained. That is, the equalization map acquisition unit 24 performs the equalization map.

To get. here,

As shown in the figure, the two phase spaces X and Y are bonded to each other in the athletic shoe f (Y ₀ ) (reference numeral 402).

  FIG. 9 is a diagram for explaining the cell structure of the athletic shoe 202 in the phase space of the shoemaker Y. The athletic shoe 202 includes a shoe brand name 210, a company name 212, a type 214, a size 216, a color 218, a material 220, a suggested retail price 222, a manufacturing cost 224, and a country of origin 226 as cell attributes.

  FIG. 10 is a diagram illustrating the cell structure of the athletic shoe 302 in the phase space of the shoe shop X. The athletic shoe 302 includes shoe brand name 310, manufacturer name 312, type 314, size 316, color 318, material 320, selling price 322, and sales store 324 as cell attributes.

  The cell joint unit 62 of the cell space level processing block 150 joins the athletic shoe 202 in the phase space of the shoemaker Y shown in FIG. 9 and the athletic shoe 302 in the phase space of the shoe shop X shown in FIG. , And associate common attributes with each other. The shoe brand names 210 and 310, the types 214 and 314, the sizes 216 and 316, the colors 218 and 318, and the materials 220 and 320 are associated with each other because the attribute names match.

  The company name 212 of the athletic shoes 202 of the shoe manufacturer Y and the manufacturer name 312 of the athletic shoes 302 of the shoe shop X are the same elements, although the attribute names are different. Therefore, the cell dimension adjusting unit 64 of the cell space level processing block 150 associates the attribute name on the shoe manufacturer Y side with the attribute name on the shoe shop X side, and unifies the attribute name in the joint cell to “maker name”. The correspondence table holding unit 66 holds a table describing the relationship in which the inconsistency is adjusted and the “company name” on the shoe manufacturer Y side is associated with the “maker name” on the shoe shop X side.

  When the athletic shoes 202 of the shoe manufacturer Y and the athletic shoes 302 of the shoe shop X are joined at the cell level, there may be an attribute in one phase space and no corresponding attribute in the other phase space. For example, the athletic shoe 202 of the shoe maker Y has the attribute of the country of origin 226, but the athletic shoe 302 of the shoe shop X has no corresponding one. Conversely, the athletic shoe 302 of the shoe shop X has an attribute of the store 324, but the athletic shoe 202 of the shoe maker Y has no corresponding item. As described above, when associating attributes between two target elements, if the attribute a exists in the first phase space and the corresponding attribute does not exist in the second phase space, the first phase Processing is performed assuming that there is no attribute to be associated with the attribute a of the space, or processing for associating with the attribute a of the first phase space by adding an empty attribute b in the second phase space. For example, the cell dimension adjusting unit 64 processes the country of origin 226 of the athletic shoe 202 of the shoe manufacturer Y as having no attribute to be associated, while the store 324 of the athletic shoe 302 of the shoe shop X has an empty space. The attribute is added to the athletic shoes 202 of the shoe maker Y, and the process of associating with the store 324 of the athletic shoes 302 of the shoe shop X is performed to adjust the mismatch regarding the number of attributes.

  As another example of adjusting the number of attributes, if there is one attribute “name” in one phase space and two attributes “last name” and “first name” in the other phase space, Since both have the same attribute, the cell dimension adjusting unit 64 associates one attribute “name” with two attributes “last name” and “first name”, thereby mismatching the number of attributes. Can be adjusted.

  FIGS. 11A to 11C are diagrams for explaining the adjustment of mismatch at the expression level. The company name 212 on the shoe manufacturer Y side is represented by an M-digit character string as shown in FIG. On the other hand, the manufacturer name 312 on the shoe shop X side is represented by an L-digit character string as shown in FIG. Here, L <M. The expression format adjustment unit 74 of the expression level processing block 70 may determine the expression format of the manufacturer name 412 that is the attribute of the joined cell, as shown in FIG. .

  When there is a difference in the number of digits in the expression format of the attribute of the target element, expressing according to the larger number of digits is an example of “acceptance expression level”. The expression format adjustment unit 74 adjusts the inconsistency in the expression format of the attribute of the target element by using a predetermined reception expression level as a filter.

  FIGS. 12A to 12D are also diagrams for explaining the adjustment of mismatch at the expression level. As shown in FIG. 12A, the size 216 on the shoe manufacturer Y side is expressed by a real value in units of inches. On the other hand, the size 316 on the shoe shop X side is represented by a real value in units of cm, as shown in FIG.

  When the unit of cm is adopted as the design specification of the software, the expression format adjustment unit 74 of the expression level processing block 160 uses the expression format of the size 416 that is the attribute of the junction cell as shown in FIG. The attribute value of size 316 on the shoe shop X side is used in units of cm. At this time, the expression format adjustment unit 74 stores the conversion formula 516 shown in FIG. 12D in the conversion table holding unit 76 to maintain compatibility of attribute values. Matching to any unit system in this way is also an example of the accepted expression level.

  FIGS. 13A to 13D are also diagrams for explaining adjustment of mismatch at the expression level. The color 218 on the shoe manufacturer Y side is represented by an integer value as shown in FIG. On the other hand, the color 318 on the shoe shop X side is represented by a character string as shown in FIG. This is because the shoe manufacturer Y uses a numbering system for finely identifying the color of the shoe, while the shoe shop X describes the color of the shoe with a name such as “紺” or “purple” that is easy for the customer to understand. Because it does. This difference is caused by the design intention of the designer and the needs of customers in the systems of the shoe manufacturer Y and the shoe shop X.

  As a software design specification, when the color expression format is a character string, the expression format adjustment unit 74 of the expression level processing block 160 uses the color 418 that is the attribute of the junction cell as shown in FIG. Using the expression format as a character string, the attribute value of the color 318 on the shoe shop X side is used. At this time, as shown in FIG. 13D, the expression format adjusting unit 74 sets a conversion rule between an integer value that identifies the color on the shoe manufacturer Y side and a character string that identifies the color on the shoe shop X side. The indicated conversion table is created and stored in the conversion table holding unit 76 to maintain compatibility of attribute values. Such conversion to a character string representation format is also an example of a reception representation level.

  In the above example, the software developer specified an athletic shoe that has an equivalence relationship between the shoe shop X and the shoe manufacturer Y. However, the software developer specified an equivalence relationship between the shoe shop X and multiple shoe manufacturers Y. May be. For example, on the basis of the equivalence relationship “shoe made from leather”, the corresponding elements between the top space of the shoe shop X and the top space of each of a plurality of shoe manufacturers Y1, Y2,. May be extracted, an adhesion space may be generated, and object data design may be performed.

[16.2] Design of Program Module Considering a program as an example of an object set, a procedure for developing software based on a plurality of programs by the software development support apparatus 100 according to the present embodiment will be described.

  FIGS. 14A and 14B are diagrams illustrating a state where two programs X and Y are integrated. All of these programs perform image processing, but it is assumed that the program configurations are different because of different program developers. These programs all have image encoding and decoding functions. For example, these programs can encode and decode images of products sold and sold in electronic commerce.

  The set level processing block 120 designs programs X and Y consisting of a plurality of modules as a set, and the phase space level processing block 130 designs programs X and Y as a phase space.

The equivalence relation setting unit 44 of the adhesion space level processing block 140 sets two equivalence relations “having an encoding processing function” and “having a decoding processing function” for the modules constituting the programs X and Y. . Corresponding element extraction unit 20 identifies a corresponding module between programs X and Y based on the two equivalence relationships set by equivalence relationship setting unit 44, and adhesive mapping acquisition unit 22 creates a correspondence relationship based on each equivalence relationship. Adhesive maps f ₁ and f ₂ are obtained.

に分割され、一方、プログラムＸの位相空間は、符号化処理機能をもつモジュールでＹ_０に対応するｆ（Ｙ_０）と、圧縮処理機能をもつモジュールでＹ_１に対応するｆ（Ｙ_１）と、それ以外の機能をもつモジュールの直和

に分割される。 Thus, the phase space of the program Y is a module Y ₀ having encoding processing function, the module Y ₁ having the compression function, the direct sum of the modules with other functions

On the other hand, the phase space of the program X is f (Y ₀ ) corresponding to Y ₀ in the module having the encoding processing function and f (Y ₁ ) corresponding to Y ₁ in the module having the compression processing function. And the direct sum of modules with other functions

It is divided into.

に写像する等化写像ｇを取得する。 The equalization map acquisition unit 24 calculates the direct sum of the two phase spaces of the programs X and Y shown in FIG. 14A as an adhesive space shown in FIG.

An equalization map g that maps to is obtained.

  FIGS. 15A and 15B are diagrams illustrating the input / output interfaces of the two modules A and B associated with each other by the adhesive mapping. The variables used for input of module A are two integer types a and b, and the variable used for output is real type c. On the other hand, the variables used for the input of the module B are two of the real type s and t, and the variable used for the output is the real type u. Although there are differences in the names of input variables and output variables between the two modules A and B associated with each other in the adhesion space level processing block 140, the cell dimension adjustment unit 64 of the cell space level processing block 150 includes two modules A and B. , B so that the names of the input variables and output variables can be matched. Further, if there is a difference in the number of input variables and output variables between the two modules A and B, the cell dimension adjusting unit 64 can correspond to the number of input variables and output variables of the two modules A and B. adjust.

  The expression format adjustment unit 74 of the expression level processing block 160 adjusts the data type if there is a difference between the data types of the input variables and output variables of the two modules A and B. Since the input variable of module A is an integer type and the input variable of module B is a real number type, for example, the input variable of module A is converted into a real number type. When the variable is a character type, the expression format adjustment unit 74 adjusts the number of digits of the character string.

  16 to 19, a procedure for developing software by the software development support apparatus 100 in which modules or data are associated with each other based on the equivalence relation will be described.

FIGS. 16A and 16B are diagrams for explaining how a program module is updated to a new version. _Assume that there is a new version of module A _new and an old version of module A _old as shown in FIG. They may be designed by different developers. The new version of the module A _new is associated with the old version of the module A _old by the equivalence relation. Accordingly, the input variables s and t of the new version module A _new are associated with the input variables a and b of the _old module A _old , respectively, and the output variable u of the new version module A _new is changed to the old version. _{Is associated with} the output variable c of the module A _old .

In this way, by associating the old and new versions of the modules with the equivalence relation, the software development support apparatus 100 replaces the _old version of the module A _old with the new version of the module A _new as shown in FIG. Can be updated. Since the software development support apparatus 100 replaces modules so that they have the same homotopy value, the new version of the module A _new can be returned to the old version of the module A _old as necessary, so that the software can be debugged efficiently. Can be done well.

FIGS. 17A and 17B are diagrams for explaining how two program modules are synthesized. As shown in FIG. 17A, the module A has two submodules A ₁ and A ₂ , and the module B has two submodules B ₁ and B ₂ . The sub-module B _{1 of the} module B is associated with the sub-module A ₁ of the module A by the equivalence relation. Accordingly, the input variable s of the submodule B ₁ is associated with the input variable a of the submodule A ₁ .

As shown in FIG. 17B, the sub-module B _{1 of the} module B is identified with the sub-module A _{1 of the} module A by the adhesive mapping determined by the equivalence relation, so that the two modules A and B are combined, One program is generated. At this time, data type and the number of digits of the correspondence obtained submodule B ₁ and sub-module A ₁ of the input and output variables are adjusted with an adhesive mapping. Since the software development support apparatus 100 performs the synthesis of modules so as to have the same homotopy value, the original modules A and B can be separated and extracted from the synthesized program as necessary.

  As described above, the software development support apparatus 100 can synthesize a plurality of modules based on the equivalence relation, and can extract a module from the synthesized program. Usability can be increased.

  18A and 18B are diagrams for explaining how data is shared between two program modules. As shown in FIG. 18A, there are a module A that inputs two variables a and b and outputs a variable c, and a module B that inputs two variables s and t and outputs a variable u. Assume that the input variable s of B and the input variable a of the module A are associated with each other by an equivalence relation, and there is no correspondence relation between other variables.

  As shown in FIG. 18B, the input variable s of the module B is identified with the input variable a of the module A by the adhesive mapping determined by the equivalence relation, so that the input variable 1 between the two modules A and B is 1 Are shared. In this way, the software development support apparatus 100 can share data associated as equivalence classes between program modules, and can improve the usability of the data.

  FIGS. 19A and 19B are diagrams illustrating a state in which two program modules are combined. As shown in FIG. 19A, there are a module A that inputs two variables a and b and outputs a variable c, and a module B that inputs two variables s and t and outputs a variable u. Assume that the output variable u of B and the input variable a of module A are associated by an equivalence relationship, and there is no correspondence relationship between other variables.

  As shown in FIG. 19B, the output variable s of the module B is identified with the input variable a of the module A by the adhesive mapping determined by the equivalence relation, so that the module B and the module A are coupled in this order. One program is generated. At this time, the data type and the number of digits are adjusted between the output variable of module B and the input variable of module A. Since the programs are combined so as to have the same homotopy value, the modules A and B can be separated by releasing the connection as necessary. As described above, the software development support apparatus 100 can combine or separate a plurality of program modules based on the equivalence relation, and can improve the usability of the program modules.

  As described above, according to the software development support apparatus 100, since the development process such as composition, replacement, and connection of software parts can be performed by linear operation based on the equivalence relation, the number of software parts increases. However, software can be developed efficiently without explosion of development man-hours.

  In the above specific example, the input / output interface is taken as an example of the attribute that characterizes the specification of the program module, and the adjustment of inconsistencies in the dimension and expression format of the input / output interface between the corresponding program modules was explained. But this is just one example of an attribute. As an example of another attribute characterizing the specification of a program module, a variable referred to by the program module may be considered. In that case, inconsistencies in the dimension and expression format of the reference variable are adjusted between corresponding program modules. A function used by a program module may be considered as an attribute. In that case, inconsistencies such as the function dimension, that is, the function name and the number of functions, are adjusted between corresponding program modules. Furthermore, inconsistencies in the dimension and expression format of the arguments and return values of the function may be adjusted.

  The view level processing block 170 of the software development support apparatus 100 may design a different view for each software developer. For example, an advanced software developer sets a view that allows data manipulation inside a program module so that the design of internal variables of the program module can be changed. On the other hand, in order to allow only the design change of the external interface of the program module to beginner software developers, a view in which data manipulation inside the program module is impossible is set. In this way, the view may be varied depending on the level of developer's programming experience and skill. In addition, when a plurality of developers jointly develop software, the view may be different for each developer so that the design can be changed only in the program area that each developer is responsible for. As will be described later, when the software development support apparatus 100 is used in a maintenance process, the view level processing block 170 may design a different view for each software maintainer.

[17] Software lifecycle management Software development activities include requirements analysis, system design, program design, program implementation, unit / combination testing, system testing, acceptance testing, and operation / maintenance. Activities are managed as a life cycle. Requirements analysis, system design, program design, and program implementation are software development processes. A unit program and a combination test are performed on the software program modules generated in the development process, a system test is performed on the functional aspects of the entire system, and finally an acceptance test at the time of system shipment is performed. This is the test process. A system that passes the acceptance test is operated and the system is maintained during operation. This is a maintenance process.

  A software life cycle management apparatus 600 shown in FIG. 20 manages the life cycle of software from the development process to the test process to the maintenance process. The development function block 610 has each function of the software development support apparatus 100 described above, and inherits the equivalence relation of the software parts set as an invariant between the abstract hierarchies, while satisfying the specifications to be satisfied by the software parts set. By incrementally refining for each abstract hierarchy, software that matches a given required specification is automatically generated from a software component set. The development function block 610 executes software requirement analysis, system design, program design, and program implementation, and the developed software is stored in the software storage unit 640. The software storage unit 640 holds all programs in required specifications, development, testing, and maintenance processes.

  The test function block 620 tests whether the software generated by the development function block 610 conforms to the detailed specification in each abstract layer, and if a problem is found, Modifying the requirement conditions, the development function block 610 incrementally refines the specifications to be satisfied by the software component set for each abstract hierarchy, starting from the abstract hierarchy in which the requirement conditions are modified.

  Testing of the developed software must be done at any stage of requirements analysis, system design, program design, and program implementation. For example, when the test function block 620 finds and corrects a defect in the required specifications, the development function block 610 returns to the upper hierarchy such as the phase space level and the adhesion space level of the abstract hierarchy model, and starts from that hierarchy. Redo the design details to eliminate the problem. In addition, when the test function block 620 finds a problem in the number of function arguments or the data type of the variable in the program implementation, the development function block 610 determines the subordinate level such as the cell space level and the expression level of the abstract hierarchical model. Redo the design from the hierarchy and resolve the problem. The software development support apparatus 100 based on the incremental and modular abstract hierarchy can reconfigure software by linear operation regardless of which hierarchy is redone.

  The maintenance function block 630 changes the requirement condition in the abstract hierarchy that should reflect the modification or function extension when the software being operated needs to be modified or expanded, and the development function block 610 is a requirement condition. Starting from the abstract hierarchy in which is changed, the specifications to be satisfied by the software component set are incrementally detailed again for each abstract hierarchy.

  There are two types of software maintenance: corrective maintenance and preventive maintenance, which can be corrected to deal with or prevent system failures. There are perfective maintenance and adaptive maintenance.

  In the correction maintenance, when a failure occurs in the current system, the cause of the failure is discovered and necessary software is corrected. The correction maintenance is an operation for correcting a residual defect of software, and generally does not change the specification. For example, correction maintenance using the abstract hierarchy model is performed by adjusting a requirement condition in a relatively lower hierarchy such as a cell space level or an expression level.

  Preventive maintenance is the correction and change of an aspect of the system in order to discover a potential failure and prevent a system failure before it has actually failed. Similarly, an abstract hierarchical model can be used.

  Improving maintenance is not changing software to avoid failures, but changing software to improve certain functions and performance of the system. Improvement maintenance involves changes in software specifications. For improved maintenance using the abstract hierarchy model, the software is redesigned while following the abstract hierarchy toward the lower hierarchy after adjusting the requirements at a relatively higher hierarchy such as the phase space level and the adhesive space level. Is done by doing.

  In adaptive maintenance, software is changed according to changes in hardware and software technology. Adaptive maintenance does not deal with failures, but changes to adapt to an evolving system. Adaptive maintenance can be performed using an abstract hierarchical model as well as other maintenance.

  The software maintenance process includes the software development process and the software testing process in that it analyzes new requirements, evaluates system and program designs, reviews code, and makes modifications as necessary. Contains common activities. Just as the development process and the test process are performed using the abstract hierarchical model, the maintenance process can be efficiently performed using the abstract hierarchical model effectively, and the maintenance burden can be reduced.

  In general, defect correction has a ripple effect that causes another defect correction. This is because maintenance rework can affect other parts of the system and create new obstacles. In general, software development depends on the developer, so it is difficult to accurately predict the impact of maintenance even if the specifications created by the developer are carefully reviewed. However, according to the software life cycle management apparatus 600 of the present embodiment, software development is performed using an abstract hierarchical model, and developer-dependent design elements are eliminated, so that the same abstraction can be achieved in maintenance. By referring to the hierarchical model, after grasping what is equivalent to what, the correct correction is efficiently performed by accurately predicting how the correction of one part will affect the other part. be able to.

  Even if it is necessary to define a new equivalence relationship for a certain part of software during correction maintenance, or an equivalence relationship that should have been defined from the beginning is discovered at the maintenance stage, the abstract hierarchy model is used. By referring to and redefining the equivalence relation at the adhesion space level and dropping the equivalence relation to the lower cell space level and the expression level, it is possible to ensure consistency. In addition, even when adding new functions to software or setting new parameters for extended maintenance and adaptive maintenance, the software is redesigned incrementally by setting new equivalence relationships as appropriate in the abstract hierarchical model. It can be performed. Also, in preventive maintenance, even if there is a difference in data type or data length between corresponding variables and it is predicted that inconvenience will occur, the abstract hierarchy model is used to define equivalence relations for these variables. Inconsistency can be adjusted for that part.

  As described above, according to the software life cycle management apparatus 600 of the present embodiment, by using the abstract hierarchical model, even if a change in the requirement occurs in the maintenance stage, the equivalence relationship is determined based on the requirement specification, The equivalence relation can be inherited as an invariant up to the design and implementation level, so that the specification can be refined and embodied, and the change can be flexibly and efficiently dealt with. By using the abstract hierarchical model, software operations in the maintenance process are guaranteed linear interoperability, as in the development process. Can be deleted.

  Various models such as a waterfall model, a rapid prototyping model, and a spiral model have been proposed as software life cycle models. The software life cycle management apparatus 600 according to the present embodiment is one of the life cycle models. It does not depend on any of the above and is not applied on the assumption of any life cycle model. However, it does not prevent the software life cycle management apparatus 600 of the present embodiment from being used in a form adapted to any one of the life cycle models. The main advantage of the software life cycle management apparatus 600 according to the present embodiment is that the equivalence relation between software component sets is inherited as an invariant between the abstract hierarchies by using the incremental modular abstract hierarchy model. In addition to guaranteeing performance, it guarantees linear integration and linear operation of a set of software components, and has consistent lifecycle management from software development to testing and maintenance.

  The embodiment has been described above. The embodiments are exemplifications, and various modifications are possible, and those skilled in the art will understand that such modifications are also included in the present invention.

It is a figure explaining a mode that adhesion space is generated from two mutually disjoint phase spaces. It is a figure explaining a mode that the phase space of a customer and a trading company is pasted up. It is a figure explaining a mode that the phase space of a seat and a support part is pasted up. It is a block diagram of the software development assistance apparatus which concerns on embodiment. It is a block diagram of the adhesion space level processing block of FIG. It is a block diagram of the cell space level processing block of FIG. It is a block diagram of the expression level processing block of FIG. It is a figure explaining a mode that two phase spaces are adhere | attached and the adhesion | attachment space is produced | generated by the adhesion space level process block of FIG. It is a figure explaining the cell structure of an object. It is a figure explaining the cell structure of an object. It is a figure explaining a mode that inconsistency in an expression level is adjusted by the expression level process block of FIG. It is a figure explaining a mode that inconsistency in an expression level is adjusted by the expression level process block of FIG. It is a figure explaining a mode that inconsistency in an expression level is adjusted by the expression level process block of FIG. It is a figure explaining a mode that two programs are integrated. It is a figure explaining the input / output interface of two modules matched by the adhesive mapping. It is a figure explaining a mode that a program module is updated to a new version. It is a figure explaining a mode that two program modules are synthesize | combined. It is a figure explaining a mode that data are shared between two program modules. It is a figure explaining a mode that two program modules are combined. It is a block diagram of the software life cycle management apparatus which concerns on embodiment.

Explanation of symbols

  20 Corresponding element extraction unit, 22 Adhesive mapping acquisition unit, 24 Equalization mapping acquisition unit, 26 Procedure acquisition unit, 28 Procedure recording unit, 44 Equivalence relationship setting unit, 52 Equivalence relationship holding unit, 62 Cell junction unit, 64 Cell dimension adjustment , 66 correspondence table holding unit, 72 expression generation unit, 74 expression format adjustment unit, 76 conversion table holding unit, 100 software development support device, 110 homotopy level processing block, 120 set level processing block, 130 phase space level processing block , 140 adhesive space level processing block, 150 cell space level processing block, 160 expression level processing block, 170 view level processing block, 190 object set storage unit, 600 software life cycle management device, 610 development function block 620, test function block, 630 maintenance function block, 640 software storage unit.

Claims (14)

  Using an incremental and modular abstract hierarchical model that guarantees linear interoperability between software component sets, consistently manages the life cycle from software development to testing and maintenance. Software life cycle management method characterized by the above.   The development process is given by incrementally refining the specifications to be satisfied by the software component set for each abstract layer while inheriting the equivalence relation of the software component set between the abstract layers as an invariant. 2. The software life cycle management method according to claim 1, wherein software that matches the required specifications is automatically generated from the software component set.   The test step tests whether the software generated by the development step conforms to detailed specifications in each abstract layer, and according to the problem found, the requirement condition in any abstract layer 3. The software life cycle management method according to claim 2, wherein the problem is eliminated by repeating the development process from the abstract hierarchy after correcting the problem.   In the maintenance process, when it is necessary to modify or extend the function of the software in operation, the development condition is changed from the abstract layer after changing the requirements in the abstract layer to reflect the modification or function extension. 4. The software life cycle management method according to claim 2, wherein the software is maintained by repeating the steps.   Software development function blocks, test function blocks, and maintenance function blocks using an incremental modular abstract hierarchy model that guarantees linear interoperability between software component sets, and the software life cycle is abstracted Software lifecycle management device characterized by consistent management based on hierarchical model.   The development function block is provided by incrementally refining the specifications to be satisfied by the software component set for each abstract layer while inheriting the equivalence relation of the software component set between the abstract layers as an invariant. 6. The software life cycle management apparatus according to claim 5, wherein software that conforms to the required specification is automatically generated from the software component set.   The test function block tests whether the software generated by the development function block conforms to a detailed specification in each abstract layer, and when a problem is found, The software life cycle management apparatus according to claim 6, wherein the requirement condition is modified in step, and the development function block repeats the refinement from an abstract hierarchy in which the requirement condition is modified.   The maintenance function block changes a requirement condition in an abstract hierarchy that should reflect the modification or function extension when the software in operation needs to be modified or expanded, and the development function block includes the development function block 8. The software life cycle management apparatus according to claim 6, wherein the refinement is performed again from an abstract hierarchy whose request condition has been changed. The abstract hierarchical model is configured to be modularized so that operations on software component sets can be performed in each hierarchy,
A set design unit for performing a set operation on the software component set to design a desired plurality of software component sets;
A phase space design unit that designs the software component set as a phase space by defining a phase in a set having a subset of the software component set as an element;
An adhesion space design unit that defines an equivalence relation on the plurality of software component sets each designed as a phase space, and designs an adhesion space in which the partial spaces associated with the equivalence relation are adhered,
The cell space design unit for expressing the attribute related to the specification of the software component belonging to the plurality of software component sets in the adhesion space by a cell, and designing the dimension of the cell,
The software life cycle management apparatus according to claim 5, further comprising an expression design unit that designs an expression format of the cell.   The modular abstract hierarchy model designs a view for the software developer or maintainer with respect to the plurality of software component sets, and for each developer or maintainer the plurality of software component sets. The software life cycle management apparatus according to claim 9, further comprising a view design unit that varies an operable range.   In the modular abstract hierarchical model, a change in the software component set caused by an operation on the software component set by each design unit in any step of development, testing, and maintenance of the software is homotopy equivalent. The software life of claim 9 or 10, further comprising a homotopy design unit that records the operation procedure so that the software component set can be returned to the state before the change. Cycle management device. The bonding space design part is:
For the first phase space X and a second phase space Y are disjoint, the continuous function from the subspace Y ₀ of Y of the second phase space into the first phase space X, the equivalence An adhesive map acquisition unit that acquires the adhesive map f representing the relationship;
Each point y in said subspace Y ₀ of the second phase space Y, by identified with the image f (y) in the first phase space X by the adhesive mapping f, the second phase space The software life cycle management according to claim 9, further comprising: an equalization map acquisition unit that adheres Y to the first phase space and acquires the adhesion space as an exclusive OR of equivalence classes. apparatus. The cell space design unit
A cell joining portion for joining the software components of the first phase space X and the software components of the second phase space Y, which are identified by the adhesion map f, in association with each other in the cell space;
13. The software life cycle management apparatus according to claim 12, further comprising a cell dimension adjusting unit that adjusts an inconsistency related to a dimension of an attribute related to a specification of the software component associated in the cell space. 14. The expression design unit includes an expression format adjusting unit that adjusts inconsistencies related to attribute expression formats related to specifications of the software components associated with each other in the cell space according to an accepted expression. Software life cycle management device described in 1.

Images