CN105389192A - Method for measuring importance of software class based on weighted q2 index - Google Patents

Abstract

The invention discloses a method for measuring the importance of a software class based on a weighted q2 index. The method comprises the following steps: abstracting a software source code written by Java language into a feature dependence network with the feature granularity; constructing a class dependence network based on the feature dependence network; calculating the weighted h index of a node based on the class dependence network; calculating the weighted m index of the node based on the class dependence network; and calculating the weighted q2 index of the node based on the weighted h index and the weighted m index of the node, and taking the weighted q2 index of the node as the measurement index of the class importance. According to the invention, the disadvantage that measurement of the class importance is little related in the prior art can be made up; and, to understand software, increasing of the code maintenance efficiency has an important significance.

Description

A kind of software class importance measures method based on weighting q2 index

Technical field

The present invention relates to a kind of software class importance measures method, especially relate to a kind of software class importance measures method based on weighting q2 index.

Background technology

Computer software has penetrated into our work and the various aspects of daily life, is changing and will continue our life of change.Along with the development of software engineering and the universal of internet, the dependence of people to software grows with each passing day, more and more higher to the requirement of software quality.Which results in the day by day complicated of the surge of system scale and software application environment, such that the risk of software development increases, software quality is difficult to be controlled effectively.

Simultaneously evolutive is one of essential attribute of software.The same biology of software systems, in its life cycle, also constantly must develop, otherwise just likely be eliminated in advance.One of important content of Software Evolution is the amendment to software source code.But to code be revised, we must have certain understanding to software systems.When maintenance resources is limited, for the developer that newly adds, accelerate it, to the understanding of system, there is important meaning.Can by recommending important software element (as class, bag etc.) for person newly developed thus accelerating its understanding to system.Although there has been the research work of many software metrics aspects at present, as LOC (LinesofCode) code line, McCabe cyclomatic complexity (CyclomaticComplexity), Halstead measure, CK set of measurements, MOOD set of metrics etc., certain help can be provided for the complicacy being familiar with software, but still have following deficiency:

(1) existing work mainly concentrates on the complicacy of tolerance code element itself, lacks the tolerance to code element importance.

(2) work on hand is mainly for the tolerance of Element-Level, the local feature of the software often of tolerance, as measured a method, a class, lacking the work carrying out software metrics from overall angle, more lacking the work from overall angle tolerance software element importance.

Class is one of main composition element of the object-oriented software of current main flow.Therefore, providing a kind of effective class importance measures method, from the importance of overall angle tolerance class, for understanding software, improving code maintenance efficiency significant.

Summary of the invention

The object of the invention is to for the deficiencies in the prior art, provide a kind of software class importance measures method based on weighting q2 index.

Above-mentioned technical matters of the present invention is mainly solved by following technical proposals: a kind of software class importance measures method based on weighting q2 index, and the method comprises the following steps:

(1) software source code write by Java language is abstract at characteristic particle size is characteristic dependence net FDN=(N _f, D _f).Wherein, N _ffor the set of characteristic node in source code; D _f={ (f _i, f _j) (f _i∈ N _f, f _j∈ N _f) be the set of nonoriented edge, the dependence between representation feature.Feature includes attribute in Java source code and method.Dependence between feature includes call relation between method and method to the use relation of attribute.

(2) FDN completed based on step (1) builds class and relies on net CDN=(N _c, D _c, P).Wherein, N _cfor the set of category node in source code; D _c(c _i∈ D _c, c _j∈ D _c) be the set of a nonoriented edge, the dependence between representation class; P is one | N _c| × | N _c| (| N _c| return N _cin nodes) matrix, represent the intensity matrix of dependence between class.Class contains class, inner classes, abstract class and interface in Java.Dependence between class is that the dependence between the feature that comprises according to class obtains, and has dependence between the feature that namely class comprises, then also Existence dependency relationship between corresponding class.

(3) weighting h index h (i) of the CDN computing node i completed based on step (2).

(4) weighting m exponent m (i) of the CDN computing node i completed based on step (2).

(5) based on the weighting g2 index of step (3) and step (4) computing node i

(6) based on the weighting g2 index of all nodes in step (3), (4) and (5) calculating CDN, as the importance values of node respective class.

Further, in above-mentioned steps (2), the structure of CDN specifically comprises following sub-step:

(2.1) extract all classes in the source code write of Java language, build the CDN only having node not have limit, i.e. a CDN=(N _c, Φ, P).Φ representative edge collection is empty, and P is a null matrix simultaneously.

(2.2) step (1) D is got _fin a limit (f _i, f _j) ∈ D _f, obtain f according to source code _iand f _jthe class defined, if f _idefine in class k, f _jdefine in class p, if k ≠ p, then (k, p) is added D _c, in step (2.1), the P (k, p) of P relevant position adds 1 certainly simultaneously; If k and p is equal, be then left intact.

(2.3) step (2.2) is repeated, until traveled through all limits in FDN.

Further, the calculating of weighting h index h (i) of above-mentioned steps (3) interior joint i specifically comprises following sub-step:

(3.1) node weight of all nodes in step (2) gained CDN is asked.The node weight s of node j _jbe defined as all limits be connected with this node in CDN weight and, that is:

$s_j={\mathrm{\Σ}}_{k\∈v_j}P(j,k),$

Wherein, v _jit is the neighbor node set of node j.

(3.2) the neighbor node set v of step (2) gained CDN interior joint i is asked _i.

(3.3) by v _iin node by its node weight descending sort (if there is the equal situation of node weight, then equal a kind of possible sequence of several value Stochastic choice), obtain the array list after sorting, that node that what list [1] position was deposited is node weight is maximum.

(3.4) from list [1], travel through each node list [q] in list list successively, find the node list [n+1] that first meets node weight and be less than (n+1), then weighting h index h (i) of node i is n.

Further, the calculating of weighting m exponent m (i) of above-mentioned steps (4) interior joint i specifically comprises following sub-step:

(4.1) node weight of all nodes in step (2) gained CDN is asked.The node weight s of node j _jbe defined as all limits be connected with this node in CDN weight and, that is:

$s_j={\mathrm{\Σ}}_{k\∈v_j}P(j,k),$

Wherein, v _jit is the neighbor node set of node j.

(4.2) the neighbor node set v of step (2) gained CDN interior joint i is asked _i.

(4.3) by v _iin node by its node weight descending sort (if there is the equal situation of node weight, then equal a kind of possible sequence of several value Stochastic choice), obtain the array list after sorting, that node that what list [1] position was deposited is node weight is maximum, list [| v _i|] position deposit be minimum that node of node weight (| v _i| be v _iinterior joint number).

(4.4) basis weighting m exponent m (i) of computing node i.

Compared with prior art, the present invention has the following advantages and good effect:

(1) in the present invention, the structure of weighting q2 index all considers software impact integrally, because the weighting h exponential sum weighting m index that structure q2 index uses all considers the node weight of other node in network, be a kind of overall viewing angle, overcome existing method to a certain extent and only pay close attention to the problem that local feature ignores global feature.

(2) the present invention proposes the importance by class in weighting q2 exponential metric software, overcome the tolerance that existing method only pays close attention to software element complicacy to a certain extent, ignore the problem of software element importance measures, can for understanding software, improve code maintenance efficiency and provide support.

Accompanying drawing explanation

The source code fragment that Fig. 1 Java language of the present invention is write;

The FDN that Fig. 2 embodiments of the invention build;

The boundless CDN that Fig. 3 embodiments of the invention build;

The corresponding P of boundless CDN that Fig. 4 embodiments of the invention build;

The boundless CDN that Fig. 5 embodiments of the invention build adds the CDN behind a limit;

The boundless CDN that Fig. 6 embodiments of the invention build adds the corresponding P of CDN behind a limit;

The final CDN that Fig. 7 embodiments of the invention build;

The corresponding P of final CDN that Fig. 8 embodiments of the invention build.

Embodiment

Also by reference to the accompanying drawings technical scheme of the present invention is further described below by embodiment:

A kind of software class importance measures method based on weighting q2 index that the present invention proposes, concrete steps are as follows:

(1) software source code write by Java language is abstract at characteristic particle size is characteristic dependence net FDN=(N _f, D _f).Shown in Fig. 1 is a Java source code fragment.The Java source code fragment of giving according to Fig. 1, can build corresponding FDN (as shown in Figure 2), and the word on node limit is the name of node individual features (name by wrapping name, class name, feature name is connected with ". " and is formed).Wherein, N _f={ p1.classX.a (), p1.classX.v, p1.classX.c (), p1.classX.b (), p2.classY.d (), p2.classZ.e (), p2.classZ.f () } be the set of characteristic node, D _f={ (p1.classX.b (), p1.classX.a ()), (p1.classX.a (), p1.classX.b ()), (p1.classX.a (), p1.classX.v), (p1.classX.v, p1.classX.a ()), (p1.classX.v, p1.classX.c ()), (p1.classX.c (), p1.classX.v), (p1.classX.a (), p2.classY.d ()), (p2.classY.d (), p1.classX.a ()), (p2.classY.d (), p2.classZ.e ()), (p2.classZ.e (), p2.classY.d ()), (p1.classX.a (), p2.classZ.f ()), (p2.classZ.f (), p1.classX.a ()), (p1.classX.c (), p2.classZ.f ()), (p2.classZ.f (), p1.classX.c ()) } be the set of nonoriented edge, dependence between representation feature.

(2) FDN completed based on step (1) builds class and relies on net CDN=(N _c, D _c, P).The structure of CDN specifically comprises following sub-step:

(2.1) extract all classes in the source code write of Java language, build the CDN only having node not have limit, i.e. a CDN=(N _c, Φ, P).Φ representative edge collection is empty, and P is a null matrix simultaneously.The Java source code fragment of giving according to Fig. 1, can build corresponding CDN=(N _c, Φ, P) and (as shown in Figure 3), wherein, N _c=p1.classX, p2.classY, p2.classZ} are the set of category node in source code, and the word on node limit be the name of node respective class (name by wrap name and class name to be connected with ". " formed).P is a null matrix, as shown in Figure 4.

(2.2) step (1) D is got _fin a limit (f _i, f _j) ∈ D _f, obtain f according to source code _iand f _jthe class defined, if f _idefine in class k, f _jdefine in class p, if k ≠ p, then (k, p) is added D _c, in step (2.1), the P (k, p) of P relevant position adds 1 certainly simultaneously; If k and p is equal, be then left intact.As shown in Figure 2, if get limit (p1.classX.a (), p2.classZ.f ()), because p1.classX.a () defines in p1.classX class, p2.classZ.f () defines in class p2.classZ, and p1.classX and p2.classZ is not same class, therefore (p1.classX, p2.classZ) is added D _c, P shown in CDN and Fig. 6 shown in Fig. 5 can be obtained.

(2.3) repeat step (2.2), until traveled through all limits in FDN, the CDN shown in Fig. 7 can be obtained, wherein, N _c={ p1.classX, p2.classY, p2.classZ} are the set of category node in source code; D _c={ (p1.classX, p2.classY), (p2.classY, p1.classX), (p1.classX, p2.classZ), (p2.classZ, p1.classX), (p2.classY, p2.classZ), (p2.classZ, p2.classY) } be the set of a nonoriented edge, the dependence between representation class.Corresponding P as shown in Figure 8.

(3) weighting h index h (i) of the CDN computing node i completed based on step (2).Specifically comprise following sub-step:

(3.1) node weight of all nodes in step (2) gained CDN is asked.The node weight s of node j _jbe defined as all limits be connected with this node in CDN weight and, that is:

$s_j={\mathrm{\Σ}}_{k\∈v_j}P(j,k),$

Wherein, v _jit is the neighbor node set of node j.Therefore, the node weight s of Fig. 7 interior joint p1.classX _p1.classXthe node weight s of=2+1=3, p2.classY _p2.classYthe node weight s of=1+1=2, p2.classZ _p2.classZ=2+1=3.

(3.2) the neighbor node set v of step (2) gained CDN interior joint i is asked _i.Therefore, if make node i be node p1.classX in Fig. 7, then its neighbor node set v _p1.classX={ p2.classY, p2.classZ}.

(3.3) by v _iin node by its node weight descending sort (if there is the equal situation of node weight, then equal a kind of possible sequence of several value Stochastic choice), obtain the array list after sorting, that node that what list [1] position was deposited is node weight is maximum.Therefore, if make node i be node p1.classX in Fig. 7, v _p1.classX={ the node weight s of the node p2.classY in p2.classY, p2.classZ} _p2.classY=1+1=2, the node weight s of p2.classZ _p2.classZ=2+1=3, therefore can obtain list array for { p2.classZ, p2.classY}, i.e. list [1]=p2.classZ, list [2]=p2.classY.

(3.4) from list [1], travel through each node list [q] in list list successively, find the node list [n+1] that first meets node weight and be less than (n+1), then weighting h index h (i) of node i is n.Therefore, for p2.classZ node: as q=1, s _p2.classZ>1; As q=2, s _p2.classY=2; Again without other neighbor nodes, therefore the value of h (p1.classX) is 2.

(4) weighting m exponent m (i) of the CDN computing node i completed based on step (2), specifically comprises following sub-step:

(4.1) node weight of all nodes in step (2) gained CDN is asked.The node weight s of node j _jbe defined as all limits be connected with this node in CDN weight and, that is:

$s_j={\mathrm{\Σ}}_{k\∈v_j}P(j,k),$

Wherein, v _jit is the neighbor node set of node j.Therefore, the node weight s of Fig. 7 interior joint p1.classX _p1.classXthe node weight s of=2+1=3, p2.classY _p2.classYthe node weight s of=1+1=2, p2.classZ _p2.classZ=2+1=3.

(4.2) the neighbor node set v of step (2) gained CDN interior joint i is asked _i.Therefore, if make node i be node p1.classX in Fig. 7, then its neighbor node set v _p1.classX={ p2.classY, p2.classZ}.

(4.3) by v _iin node by its node weight descending sort (if there is the equal situation of node weight, then equal a kind of possible sequence of several value Stochastic choice), obtain the array list after sorting, that node that what list [1] position was deposited is node weight is maximum, list [| v _i|] position deposit be minimum that node of node weight (| v _i| be v _iinterior joint number).Therefore, if make node i be node p1.classX in Fig. 7, v _p1.classX={ the node weight s of the node p2.classY in p2.classY, p2.classZ} _p2.classY=1+1=2, the node weight s of p2.classZ _p2.classZ=2+1=3, therefore can obtain list array for { p2.classZ, p2.classY}, i.e. list [1]=p2.classZ, list [2]=p2.classY.|v _i|＝|{p2.classY,p2.classZ}|＝2。

(4.4) basis weighting m exponent m (i) of computing node i.Therefore,

(5) based on the weighting g2 index of step (3) and step (4) computing node i therefore, for p1.classX node, its q2 index

$q2\left(p1.classX\right)=\sqrt{h\left(p1.classX\right)m\left(p1.classX\right)}=\sqrt{2*1}\≈\mathrm{1.41.}$

(6) based on the weighting g2 index of all nodes in step (3), (4) and (5) calculating CDN, as the importance values of node respective class.Therefore, can obtain $q2\left(p2.classY\right)=\sqrt{h\left(p2.classY\right)m\left(p2.classY\right)}=\sqrt{2*1}\≈1.41;$ $q2\left(p2.classZ\right)=\sqrt{h\left(p2.classZ\right)m\left(p2.classZ\right)}=\sqrt{2*1}\≈\mathrm{1.41.}$

Specific embodiment described herein is only to the explanation for example of the present invention's spirit, in embodiment, the value of q2 (p1.classX), q2 (p2.classY) and q2 (p2.classZ) is equal, this is a kind of situation possible in reality, but does not represent all situations and be not always the case.Those skilled in the art can make various amendment or supplement or adopt similar mode to substitute to described specific embodiment, but can't depart from spirit of the present invention or surmount the scope that appended claims defines.

Claims (8)

1., based on a software class importance measures method for weighting q2 index, it is characterized in that, comprise the following steps:

(1) software source code write by Java language is abstract at characteristic particle size is characteristic dependence net FDN=(N _f, D _f).Wherein, N _ffor the set of characteristic node in source code; D _f={ (f _i, f _j) (f _i∈ N _f, f _j∈ N _f) be the set of nonoriented edge, the dependence between representation feature.

(2) FDN completed based on step (1) builds class and relies on net CDN=(N _c, D _c, P).Wherein, N _cfor the set of category node in source code; D _c(c _i∈ D _c, c _j∈ D _c) be the set of a nonoriented edge, the dependence between representation class; P is one | N _c| × | N _c| (| N _c| return N _cin nodes) matrix, represent the intensity matrix of dependence between class.

(3) weighting h index h (i) of the CDN computing node i completed based on step (2).

(4) weighting m exponent m (i) of the CDN computing node i completed based on step (2).

(5) based on the weighting g2 index of step (3) and step (4) computing node i

(6) based on the weighting g2 index of all nodes in step (3), (4) and (5) calculating CDN, as the importance values of node respective class.

2. a kind of software class importance measures method based on weighting q2 index according to claim 1, is characterized in that, the feature in described step (1) includes attribute in Java source code and method.

3. a kind of software class importance measures method based on weighting q2 index according to claim 1, is characterized in that, the dependence in described step (1) between feature includes call relation between method and method to the use relation of attribute.

4. a kind of software class importance measures method based on weighting q2 index according to claim 1, it is characterized in that, the class in described step (2) contains class, inner classes, abstract class and interface in Java.

5. a kind of software class importance measures method based on weighting q2 index according to claim 1, it is characterized in that, dependence in described step (2) between class is that the dependence between the feature that comprises according to class obtains, between the feature that namely class comprises, there is dependence, then also Existence dependency relationship between corresponding class.

6. a kind of software class importance measures method based on weighting q2 index according to claim 1, is characterized in that, in described step (2), the structure of CDN specifically comprises following sub-step:

(2.1) extract all classes in the source code write of Java language, build the CDN only having node not have limit, i.e. a CDN=(N _c, Φ, P).Φ representative edge collection is empty, and P is a null matrix simultaneously.

(2.2) step (1) D is got _fin a limit (f _i, f _j) ∈ D _f, obtain f according to source code _iand f _jthe class defined, if f _idefine in class k, f _jdefine in class p, if k ≠ p, then (k, p) is added D _c, in step (2.1), the P (k, p) of P relevant position adds 1 certainly simultaneously; If k and p is equal, be then left intact.

(2.3) step (2.2) is repeated, until traveled through all limits in FDN.

7. a kind of software class importance measures method based on weighting q2 index according to claim 1, it is characterized in that, the calculating of weighting h index h (i) of described step (3) interior joint i specifically comprises following sub-step:

(3.1) node weight of all nodes in step (2) gained CDN is asked.The node weight s of node j _jbe defined as all limits be connected with this node in CDN weight and, that is:

$s_j={\mathrm{\Σ}}_{k\∈v_j}P(j,k),$

Wherein, v _jit is the neighbor node set of node j.

(3.2) the neighbor node set v of step (2) gained CDN interior joint i is asked _i.

(3.3) by v _iin node by its node weight descending sort (if there is the equal situation of node weight, then equal a kind of possible sequence of several value Stochastic choice), obtain the array list after sorting, that node that what list [1] position was deposited is node weight is maximum.

(3.4) from list [1], travel through each node list [q] in list list successively, find the node list [n+1] that first meets node weight and be less than (n+1), then weighting h index h (i) of node i is n.

8. a kind of software class importance measures method based on weighting q2 index according to claim 1, it is characterized in that, the calculating of weighting m exponent m (i) of described step (4) interior joint i specifically comprises following sub-step:

(4.1) node weight of all nodes in step (2) gained CDN is asked.The node weight s of node j _jbe defined as all limits be connected with this node in CDN weight and, that is:

$s_j={\mathrm{\Σ}}_{k\∈v_j}P(j,k),$

Wherein, v _jit is the neighbor node set of node j.

(4.2) the neighbor node set v of step (2) gained CDN interior joint i is asked _i.

(4.3) by v _iin node by its node weight descending sort (if there is the equal situation of node weight, then equal a kind of possible sequence of several value Stochastic choice), obtain the array list after sorting, that node that what list [1] position was deposited is node weight is maximum, list [| v _i|] position deposit be minimum that node of node weight (| v _i| be v _iinterior joint number).

(4.4) basis weighting m exponent m (i) of computing node i.