CN104050083B - Condition/Decision-coverage-oriented test data automatic generation method - Google Patents
Condition/Decision-coverage-oriented test data automatic generation method Download PDFInfo
- Publication number
- CN104050083B CN104050083B CN201410271284.8A CN201410271284A CN104050083B CN 104050083 B CN104050083 B CN 104050083B CN 201410271284 A CN201410271284 A CN 201410271284A CN 104050083 B CN104050083 B CN 104050083B
- Authority
- CN
- China
- Prior art keywords
- branch
- branch node
- path
- function
- node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Debugging And Monitoring (AREA)
Abstract
The invention discloses a condition/decision-coverage-oriented test data automatic generation method. Test data with which branch nodes can be covered are looked for according to condition/decision coverage criteria, so that a test data set of source codes to be tested is generated. The step that the test data with which the branch nodes can be covered are looked for according to the condition/decision coverage criteria is achieved in a method that an input vector set is constructed through a random method first, operational analysis is conducted on the input vector set, then the input vector set is expanded in a linear fitting means, and the test data with which the branch nodes can be covered according to the condition/decision coverage criteria are obtained through repeated iteration. The method is highly automated, unit testing efficiency can be improved as much as possible, test cost is lowered, and test coverage criteria oriented toward condition/decision of the nodes are supported.
Description
Technical field
The present invention relates to the test conditional/judgement of a kind of automatic generation of test data, more particularly to automation cell
The generation method of the test case data of coverage test.
Background technology
Measuring technology is to ensure that software system quality is most important and one of most efficient method, all the time and industrial quarters
Ensure the topmost means of software system correctness.Limited by computing capability and time, space resources, test cannot limit it is soft
The all possible execution of part, in practice, various test sufficient degree criterions are used for driving and assess test process.Therefore, towards
How given test sufficient degree criterion, generate corresponding test case set, covers corresponding knot with the execution of driver
Structure, becomes the sport technique segment of most critical in test process.Testing example design is relied on substantially and is manually completed at present, not only
Waste time and energy, and the effectiveness of use-case is also difficult to be guaranteed.How automatically to generate for various test sufficient degree criterions has
The test case of effect, has become one of hot issue of Testing Technology Study.
Towards the Test coverage (abbreviation of condition/judgement：C/DC, i.e. Condition/Decision Coverage) criterion,
That, using a kind of wide white box testing sufficient degree criterion, it requires the enough test case of design so that in judgement each
The be possible to of condition corrects to few execution once, while all of each judgement may determine that result is at least performed once.Lift
For example, sentence S=A and (B or C) are judged, wherein, A, B, C are the condition in judging.When ABC values are TTF, S takes
It is worth for T；When ABC values are FFT, S values are F.Under test set { TTF, FFT }, each condition, i.e. condition ABC exist T and
The value situation of F, and judge that sentence S there is also the value of T and F.
Theoretical research is taken it has been proved that there is no any condition that general effective algorithm can be arbitrarily judgement in program
Value combination producing test input.Existing research work can be divided into based on static analysis and based on two class method of Dynamic Execution.Base
The Nonlinear Constraints in judging cannot be effectively processed in the method for static analysis, dynamic approach is easily trapped into local optimum
Put and feasible input cannot be found, they have significant limitation when practical problem is processed.
In science and engineering problem, generally some discrete data, root can be obtained using methods such as sampling, experiments
According to these data, we are often desirable to obtain a discrete equation and given data for approaching actual function or more crypto set
Match, this process is called fitting.Linear fit is fairly simple fit approach, and data point is attached using straight line,
As a result it is a polygon.Linear fit is easy to use, and it can obtain approximating function within the shorter time, shorten calculation
The previous work amount of method automatic Data Generation Test.
The content of the invention
Problem to be solved by this invention is to build automatically test data (test case) for source program to be tested so that
Perform under these test datas source program to be tested can cover the source program to be tested all judgements possibility value and sentence
In fixed, the possibility value of all conditions, completes condition/judging coverage.
To solve the above problems, the scheme that the present invention is adopted is as follows：
A kind of automatic generation of test data covered towards condition criterion, including analyze source program to be measured and obtain to be measured
The step of the set of the associated path collection of the set in all paths of source program and the set BS of all branch nodes and branch node
Suddenly, each branch node B during the method is also included to branch node collection BS performs following steps：
S1：Each condition of each branch node on each path before B (containing B), structure are concentrated according to the associated path of B
Build the branch function F with regard to input variable vector Xk,i,j(X)；The input variable vector X becomes for the source program input to be measured
The vector that amount is constituted；The branch function Fk,i,j(X) represent branch's letter of j-th condition of upper i-th branch node of path k
Number；The path k is through branch node B；
S2：Build initial input vector collection U；The initial input vector collection U includes at least two input vectors；
S3：Calculate the input using each input vector in input vector collection U as source program to be measured and perform source program to be measured
When each path k for associating with the branch node B on each branch function before B on each branch node of (containing B)
Fk,i,j(X) the set V and farthest public branch node m of branch function value are obtainedk；Execution road when source program to be measured is performed
Footpath calculates the value of each condition in the judgement of B and judgement when branch node B；If there is one in input vector collection U
When input vector performs source program to be measured, through branch node B and in making the judgement of B or judging, certain condition is produced execution route
New value, then record the input vector as a test data of the branch node B；If the branch node B's
Test data set cover the judgement of B and judge in the be possible to value of each condition, then return the test data set as described
The test data set of branch node B；
S4：Concentrated on each branch node for (containing B) before B on each path k according to the associated path of branch node B
Each branch function value set V and farthest common node mkFarthest common node m is located on build path kk(contain m beforek)
Each branch node each branch function fitting function LFk,i,j；LFk,i,jRepresent the jth of upper i-th branch node of path k
The linear fit function of the branch function of individual condition；
S5：According to each branch function in each branch node for (containing B) before B on each path associated with B
Linear fit function LFk,i,jFeasible interval I is calculated with source program to be measured；
S6：Randomly select in feasible interval I each input variable input be worth to test data set add to be input into
In quantity set U；
S7：Step S3 to S6 is repeated until number of times that step S3 to S6 is performed reaches the number of times of restriction；
S8：Return all records the branch node B test data as the branch node B test data
Collection.
Further, the automatic generation of test data covered towards condition criterion of the invention, step S2
Described in initial input vector collection U include Ni+1 input vector, the Ni be input variable number；Initial input
Vector set U={ u0, u1, u2..., uNi, wherein
ui={ ui,1, ui,2..., ui,Ni}；Input vector u0For random generation, uiWith u0Meet relation：
Wherein i, j ∈ 1,2 ..., Ni }.
Further, the automatic generation of test data covered towards condition criterion of the invention, step S3
Including：
S31：Unenforced input vector u is obtained from input vector collection Ut；
S32：Judge unenforced input vector utWhether there is；If there is no unenforced input vector, then perform
Step S34；Otherwise execution step S33；
S33：According to utPerform each path of associated path concentration that source code to be measured obtains the path and branch node B for performing
The public branch node m of kk,t, calculate the branch function of each branch node for (containing B) before branch node B on the k of path
Fk,i,j(X) value, each branch function F of each branch node of all associated paths of Bk,i,j(X) value constitutes set V；If
The path of execution is through branch node B, then an associated path of the path for performing for branch node B, according to the associated path
The value of each condition during the branch function of each condition is worth to the judgement of branch node B and judges in the judgement of upper B；If B
Judgement or judge that conditional generates new value, then record the input vector and test as one of the branch node B
Data；If the test data set of B covers the judgement of B and the be possible to value of each condition in judging, the test data is returned
Collect the test data set as the branch node B and terminate to generate whole branch node B the process of test data.
S34：From { the m of each associated path k record of branch Bk,1, mk,2..., mk,TUIn choose farthest public branch section
Point mk。
Further, the automatic generation of test data covered towards condition criterion of the invention, step S4
Comprise the following steps：
S41：The associated path of input vector collection U and B concentrates each branch node for (containing B) before B on each path k
Each branch function value set V and farthest common node mkBuild coordinate set Vk,y,z,j={ (u1,j, Fk,y,z(u1,j)),
(u2,j, Fk,y,z(u2,j)) ..., (uTU,j, Fk,y,z(uTU,j))}；
S42：With Vk,y,z,j={ (u1,j, Fk,y,z(u1,j)), (u2,j, Fk,y,z(u2,j)) ..., (uTU,j, Fk,y,z(uTU,j))}
Interior two adjacent coordinate points build Fk,y,z(xj) linear fit function LFk,y,z(xj)=pj*xj+qj, obtain parameter set
LFPQk,i,j={ (p1, q1), (p2, q2) ..., (pNi, qNi)}；
Wherein, Vk,y,z,jRepresent the seat of z-th branch function j-th input variable of correspondence of upper y-th branch node of path k
Mark collection, wherein y≤m；ui,jFor the value of j-th input variable of i-th input vector in input vector collection U；Fk,y,z(xj) for road
Function of k upper y-th node, z-th branch function in footpath with regard to j-th input variable；LFk,y,z(xj) for path k it is upper y-th section
Linear fit function of z-th branch function of point with regard to j-th input variable；LFPQk,i,jFor j-th of upper i-th node of path k
The parameter set of the linear fit function of branch function.
Further, the automatic generation of test data covered towards condition criterion of the invention, step S5
Including：
S51：Calculate each branch node for (containing B) before branch node B in each associated path k of branch node B
In each branch function feasible interval Dk,i,j；
S52：Merge the feasible interval D of each branch function in each branch nodek,i,jObtain the feasible of each branch node
Interval Dk,i；
S53：Merge farthest public branch node mkThe feasible interval D of front each branch nodek,iObtain branch node B to exist
Feasible interval I in associated path kk。
S54：Merge feasible interval Is of the branch node B in each associated path kk, obtain the feasible interval of branch node B
I。
Further, the automatic generation of test data covered towards condition criterion of the invention, the feasible region
Between I={ I1, I2..., Is, wherein, numbers of the s for input variable, IiFor the feasible interval of i-th input variable, Ii=
{Ii,1, Ii,2..., Ii,Mi, wherein, Ii,jInterval for i-th input variable, j-th possible segmentation, Mi is i-th input variable
Possible segmentation interval number, in step S6, when Mi is 1, from IiFor feasible interval unique segmentation of i-th input variable
Interval Ii,1Extension interval in random value build test data.
A kind of machine readable media, be stored with the computer-readable recording medium instruction set, when the instruction set is performed so that
The above-mentioned automatic generation of test data covered towards condition criterion of the executable present invention of the machine.
The technique effect of the present invention is as follows：
1st, the inventive method is without the need for the data dependence relation between each sentence on analysis program path, using Dynamic Execution
Mode, automatically determines the linear fit function of branch's predicate, efficiency high.
2nd, Nonlinear Constraints that can effectively in processing path, and by the way of dynamic step length expansion, well
Solve the problems, such as that dynamic approach is easily trapped into local best points and cannot find feasible input.
Compound condition predicate equivalence is converted into branch function point by compound condition constraint that the 3rd, can effectively in processing path
Group, quickly determines the available interval of input variable using linear fit function.
4th, constraint solver solves constrained system using the method for linear fit, makes full use of linear fit function simply easy
Advantage；If branch function is linear with regard to input variable, the linear fit function function itself for obtaining is completely
It coincide, if branch function is nonlinear with regard to input variable, the comprehensive linear fit function obtained on each interval,
The substantially situation of branch function is can be evaluated whether, as the continuous execution of program, the data for function assignment are more, estimation is obtained
Function information it is more accurate.In theory, when digital simulation function it is interval sufficiently small when, the linear fit function of acquisition is with regard to energy
Press close to the truth of function enough, just more have directive significance to automatically generating accurate test data.
5th, the inventive method is increasingly automated, can improve the efficiency of unit testing as far as possible, reduces testing cost.
6th, the inventive method can be good at supporting the Testing criteria towards condition/judgement.
Description of the drawings
Fig. 1 is source program code example to be measured of the invention.
Fig. 2 is the flow chart of step S3 of the present invention.
Specific embodiment
The present invention is described in further details with reference to Figure of description.
First, the ultimate principle of CLFF automatic generation of test data
CLFF automatic generation of test data, the i.e. automated test data generation based on linear fit function drive are calculated
Method, is application of the linear fit technology in automated test data generation.Specifically, morphology is carried out to source program to be measured first
The all paths of source program to be measured and all branch nodes are obtained after analysis, syntactic analysiss and semantic analysis, then for each point
Zhi Jiedian is individually created a set of test data (test case) and enables the test data to cover the judgement and judgement of the node
In each condition be possible to value, a set of test data of each branch node correspondence tests all branches' sections of source program
Point can obtain test data set.
When test data being generated for certain branch node, be first randomly generated an input data, then according to the input number
According to source program to be measured is performed, its implementing result is analyzed, new test data is constructed after linear fit according to implementing result, so
Source program to be measured is performed according to new test data afterwards, its implementing result is analyzed, further according to implementing result after linear fit
The new test data of construction, steps be repeated alternatively until that one group of test data of presence can cover the judgement of the branch node and sentence
All valued combinations of each condition in fixed.Additionally, also there is analyzed branch node here cannot cover judgement/bar forever
All valued combinations of part, in this case, by stopping when cycle-index reaches a certain limit value to cycle count, so
It is possible to afterwards cover the judgement of the branch node and judges that the input data of conditional a certain kind value is returned as test data.
Therefore, it is of the invention to focus on, how linear fit is carried out and how according to Linear Quasi according to implementing result
Close the new test data of construction.
2nd, build branch function
Step S1 of the present invention is to be concentrated on per paths (to contain before B according to the associated path of the branch node B
B each condition of each branch node), builds the branch function F with regard to input variable vector Xk,i,j(X).The step is referred to as
To build branch function.
The associated path collection of branch node：Through the set in the path of branch node.
Condition has two kinds, and one kind is Boolean expression：Bool(X)；For Boolean expression, branch function Fk,i,j(X)=
Bool(X).Another kind is the condition comprising conditional operator, is expressed as：L (X) rel R (X), rel ∈ ＞, >=,=, ≠,
≤, ＜ }；Here rel represents conditional operator, and L (X) represents the expression formula on the left of conditional operator, and R (X) represents conditional operation
The expression formula on symbol right side, then, branch function Fk,i,j(X)=L (X)-R (X).That is, branch function is the specification of condition
Represent, any condition can be expressed as form：Fk,i,j(X) rel0, rel ∈ ＞, >=,=, ≠ ,≤, ＜ }.It is generally every
Multiple conditions, after carrying out the canonical representation of condition, the path k upper i-th associated with the branch node B are included on individual branch node
The logical expression of individual branch node can be expressed as：cK, i=FK, i, 1(X)rel10op1Fk,i,2(X)rel20op2...ops- 1Fk,i,s(X)rels0, op hereiRepresent with (And) or or (Or) logical operator, s represents upper i-th branch node of path k
The quantity of upper condition.
In addition to above-mentioned canonical representation, the branch function of structure also has the mapping process of a parameter, that is, will divide
Prop up function Fk,i,j(X) parameter causes branch function F in mapping to the vectorial X of input variablek,i,j(X) it is with regard to input variable
The function of vectorial X.For specific path k, branch function Fk,i,j(X) functional relationship and between input variable vector X is true
Fixed.As shown in figure 1, in the example source codes write by C language, a total of 6 paths：Respectively：Path 1：P1→C1
→P2→C2→P4→P6；Path 2：P1→C1→P2→C2→C3→P5→P6；Path 3：P1→C1→P2→C2→C3→
P6；Path 4：P1→C1→P3→C2→P4→P6；Path 5：P1→C1→P3→C2→C3→P5→P6；Path 6：P1→C1
→P3→C2→C3→P6.In program, one has three branch nodes：C1, C2, C3.The path set associated with branch node C1：
Path 1,2,3,4,5,6；The path set associated with branch node C2：Path 1,2,3,4,5,6；The road associated with branch node C3
Footpath collection：Path 2,3,5,6.Example functions have 3 inputs to be respectively a, b, c, and the wherein condition of C2 branch nodes is d+e>
100, in the case of path determines, d and e relative to a, what the input of b, c was to determine, d=D (a, b, c) and e can be expressed as
=E (a, b, c), then the branch function of the condition be defined as：Fk,i,j(X)=D (X)-E (X), X={ a, b, c }.Accordingly, for
Arbitrary input x={ ax,bx,cx, branch function Fk,i,j(X) a specific functional value can be directly calculated.With path 4 it is
Example, d=(a-b) * 2, e=b, it is hereby achieved that the branch function F of the only one condition of C2 branch nodes4,2,1(a,b,c)
=d+e-100=(a-b) * 2+b-100=2*a-b-100.By taking path 2 as an example, d=(a-b) * 2, e=d=(a-
B) * 2, it is hereby achieved that the branch function F of the only one condition of C2 branch nodes2,2,1(a, b, c)=d+e-100=
(a-b) * 4-100.That is, under different paths, the branch function of the similarity condition of same branch node may be different.But
Under the path for determining, branch function has been to determine with regard to the function of input variable.
The input of this step is before source program to be measured is located at B in each path on the associated path collection of branch node B
Each condition of each branch node of (containing B), output is the branch function F of corresponding conditionsesk,i,j(X).Here X be input variable to
Amount, the vector being made up of source program input variable to be measured.Branch function Fk,i,j(X) represent the of upper i-th branch node of path k
The span of the branch function of j condition, i and j depends on branch node number and each point on the path of branch node place
Conditional number on Zhi Jiedian；Path k is through branch node B.
It should be noted that path is to carry out morphological analysis, syntactic analysiss and semantic analysis by source program to be measured here,
Final analysis obtains a certain paths that source program Program path to be tested is concentrated, and the branch node B is by Program path collection
In all paths branch node composition program branch node concentrate some branch node.Those skilled in the art manage
Solution, carries out morphological analysis, syntactic analysiss to source program to be measured and belongs to prior art, and which implements details nor institute of the present invention
The subject category being discussed.Those skilled in the art are further appreciated that is carrying out morphological analysis, syntactic analysiss to source program to be measured
And after semantic analysis, branch function can be built to each condition of all branch nodes of source program to be measured first, obtain one
Total branch function collection, then according to (containing B) before being located at B in each paths in the path set associated with branch node B
Branch node is concentrated from this total branch function and chooses corresponding branch function.
3rd, build initial input vector collection
This step is aforementioned step S2.
As the present invention builds feasible interval by linear fit, then from feasible interval interior selection test data.And line
Property fitting at least need two data samples.Therefore, when initial input vector collection U is built, initial input vector collection U is extremely
Two input vectors are included less.Here, the aforesaid input variable vector X of input vector correspondence.Each element of input vector is
Correspond to the specific value of the corresponding input variable of source program to be measured.Input vector in initial input vector collection U is by random
Generate, each element in input vector is by generating at random.Linear fit to better implement the present invention.It is preferential in the present invention
Adopt and build initial input vector collection U with the following method：
Initial input vector collection U includes Ni+1 input vector, numbers of the Ni for input variable；Initial input vector
Collection U={ u0, u1, u2..., uNi, wherein ui={ ui,1, ui,1, ui,2..., ui,Ni}.In this Ni+1 input vector, wherein the
One input vector, i.e. input vector u0For random generation.uiWith u0Meet relation：Its
Middle i, j ∈ 1,2 ..., Ni }.
The benefit of so design is, for certain branch function Fi,j(X), when carrying out linear fit, each input variable can
To carry out respectively, and multilinear fitting is avoided, while all of variable can obtain the data for carrying out effective linear fit.
By taking the path 4 of previous example as an example, C2 node branch function is F4,2,1=2*a-b-100.Assume the four of input vector collection
Individual input vector is respectively：u0={ 0, -3, -2 }, u1={ 4, -3, -2 }, u2={ 0,9, -2 }, u3={ 0, -3,7 }.
Wherein, input vector u0, u1In, input variable b is identical with the value of c, and respectively -3 and -2, at this point it is possible to build with regard to defeated
Enter the linear fit function LF of variable a4,2,1(a)=2*a-97.In the same manner, according to input vector u0, u2Can build with regard to input
The linear fit function LF of variable b4,2,1(b)=- 1*b-100.
4th, code is performed with statistics
This step is aforementioned step S3.
The input of this step is：Input vector collection U, the associated path collection of branch node B and source program to be measured.It is output as：
(contained B) before branch node B on each path k of associated path concentration of the corresponding branch node B of each input vector is each
Each branch function F of branch nodek,i,j(X) the set V that value is constituted, and the farthest common path branch section with path k
Point mk。
Specific process can pass through the flow process shown in Fig. 2 and realize, including step S31, S32, S33, S34.
S31, obtains unenforced input vector u from input vector collection Ut.Step S3～S6 is the process of iterative cycles.It is defeated
Incoming vector collection U changes with iteration cycle process, and the input of the first time iteration of iterative cycles is initial input vector collection U,
Initial input vector collection U is obtained by abovementioned steps S2.As nth iteration process has been repeated with front iterative process several times,
That is in the input vector collection U of nth iteration process, partial input vectors perform step S33 via former iterative process.
Consider treatment effeciency, the input vector that this part is repeated need not repeat execution step S33 again.Therefore it may only be necessary to from input to
Those unenforced input vector execution steps S33 are chosen in quantity set U.
S32, judges unenforced input vector utWhether there is.The step is the judgement step of the extension of S31.If not
The input vector u of executiontDo not exist, in representing input vector collection U, each input vector has been performed both by step S33, then perform step
Rapid S34.If unenforced input vector utExist, then to input vector utExecution step S33.
S33, according to utPerform source code to be measured obtain perform path and branch node B each associated path k it is public
Branch node mK, t, and calculate the branch function F of (containing B) each branch node before branch node B in associated path kk,i,j
(X) value.Here execution source code to be measured can be that simulation is performed.A kind of implementation is, by way of pitching pile, to treat
Survey.Pitching pile code can monitor the result of each step execution of source code to be measured and perform to be measured
The path of the process of source code.Second implementation is saved positioned at branch in each associated path k for directly calculate branch node B
The branch function F of (containing B) each branch node before point Bk,i,j(X) value, then according to branch function Fk,i,j(X) corresponding condition
ck,iAnalyze the path P of its executiont.Under the first implementation, including three steps：Implantation pitching pile code, performs source to be measured
Program, the associated path for calculating branch node B concentrate the branch function of each branch node for (containing B) before B on each path k
Fk,i,j(X) value.And under second implementation, the associated path for performing source program to be measured and calculating branch node B concentrates each
The branch function F of each branch node for (containing B) before B on the k of pathk,i,j(X) value can be completed with a step.Therefore it is of the invention
Second implementation is adopted preferentially.Arbitrary input vector utAfter performing source code to be measured, performed path P can be obtainedtWith
The associated path of branch node B concentrates the public branch node m of path kK, t.For input vector collection U={ u1, u2..., uKU}
For, can be with the public branch set of node M of associated path kk={ mk,1, mk,2..., mk,TU, and branch function Fk,i,j(X)
Value set V={ V1, V2..., VTU}.Here TU represents the number of element in input vector collection U.Here public point
Zhi Jiedian mk,tThe input vector u that (wherein, t ∈ { 1,2 ..., KU }) is represented for target under branch nodetUnder the k of path most
Remote public partial node.Input vector utPublic branch node under the k of path is according to input vector utAfter performing source code to be measured
Performed path PtLast branch node in the part Chong Die with path k.Correspondingly, calculate in associated path k of B
Branch function Fk,i,j(X) value is also referred to according to input vector utPerform performed path P after source code to be measuredtWith path k
Branch function F in each condition of the branch node of lapk,i,j(X) in value, namely set VWherein Fk,i(ut)={ Fk,i,1(ut),
Fk,i,2(ut) ..., Fk,i,iM(ut)}.By taking the code flow in Fig. 1 as an example.The associated path of branch node C2 concentrates one to associate
Route Routes 1 are P1 → C1 → P2 → C2 → P4 → P6.And a certain input vector utUnder={ 0, -3, -2 }, perform in Fig. 1
After code, the path P for obtainingt(i.e. path 3) is：P1 → C1 → P2 → C2 → C3 → P6, then the path for overlapping is：P1→C1→
P2 → C2, therefore public branch node m1,t=C2, or represented with the subscript of second branch node：m1,t=2.Correspondingly,
Each branch function F1,i,j(X) value is：F1,1,1=a-b=3；F1,2,1=(a-b) * 4-100=-88.The path of execution
Pt：P1 → C1 → P2 → C2 → C3 → P6 through branch node C2, then Pt(i.e. path 3) is the associated path of branch node C2, root
According to branch function value F of each condition in the judgement of C2 on the associated path path 33,2,1=(a-b) * 4-100=-88 (roads
On footpath 3, second branch node is branch node C2) obtain the judgement e+d of branch node C2>100 value is false, in judgement
Condition e+d>100 value is also false；If judgement/condition valued combinations (being judged to vacation, judge conditional as false) makes C2's
In judging or judging, certain condition produces new value, then record input vector ut={ 0, -3, -2 } are used as the branch
One test data of node C2；If the test data set of C2 covers the judgement of C2 and each condition is be possible to takes in judging
Value, returns the test data set as the test data set of the branch node C2 and terminates to generate whole branch node B and survey
The process of examination data.
S34, to every associated path k of branch node B from { mk,1, mk,2..., mk,TUIn path selection k farthest public affairs
Common branch node mk.Due to public branch node m herek,t(wherein, t ∈ { 1,2 ..., TU }) is branch node subscript, because
This, the farthest public branch node m of associated path kk=max { mk,1, mk,2..., mk,TU}。
It should be noted that farthest public branch node m herekNode B was included () before branch node B.
5th, linear fit
This step is aforementioned step S4.
The input of this step is on each branch node of (containing B) before being located at B in each associated path k of branch node B
Each branch function value set V and the farthest public branch node m of each associated pathk.The two inputs are all steps S3
Output.It is output as：The farthest public branch node m of each associated path k of branch node BkIt is each in each front branch node
The linear fit function LF of branch functionk,i,j.Here LFk,i,jRepresent the line of j-th branch function of upper i-th branch of path k
Property fitting function.It should be noted that farthest public branch node m herekIt is front including farthest public branch node mk.According to front
State input vector U={ u1, u2..., uTU, the set V={ V of branch function value1, V2..., VTU, input vector can be obtained
Coordinate set：{(u1, V1), (u2, V2) ..., (uTU, VTU)}.For z-th branch's letter of arbitrary y-th branch node on the k of path
Number Fk,y,z(X) coordinate set V can be obtainedk,y,z={ (u1, Fk,y,z(u1)), (u2, Fk,y,z(u2)) ..., (uTU, Fk,y,z
(uTU))}。
Linear fit function LFk,i,jThe mode of realization is a lot, following to have three kinds of embodiments：
The embodiment of the first linear fit is fitted using multiple linear regression.According to coordinate set Vk,y,zAnd it is polynary
Linear fit function(wherein, numbers of the Ni for input variable), you can to calculate
The multilinear fitting function LF for arrivingk,y,z(X) coefficient { a0, a1..., aNi, so as to complete the structure of linear fit function.
Second and the embodiment of the third linear fit be all that simple Linear Quasi is carried out one by one to each input variable
Close.First by Fk,y,z(X) it is considered as with regard to input variable xj(xjRepresent j-th input variable, wherein j ∈ { 1,2 ..., Ni, Ni
For the number of input variable) function Fk,y,z(xj), according to coordinate set Vk,y,zCan obtain with regard to input variable xjCoordinate set
For：Vk,y,z,j={ (u1,j, Fk,y,z(u1,j)), (u2,j, Fk,y,z(u2,j)) ..., (uTU,j, Fk,y,z(uTU,j)), wherein, ui,jTable
Show the value of j-th input variable in i-th input vector.Fk,y,z(xj) linear fit function LFk,y,z(xj)=pj*xj+qj。
The embodiment of second linear fit is with coordinate set Vk,y,z,jInterior all data pass through as measured data
Method of least square builds Fk,y,z(xj) linear fit function LFk,y,z(xj)=pj*xj+qj.Then have：
The embodiment of the third linear fit is with coordinate set Vk,y,z,j={ (u1,j, Fk,y,z(u1,j)), (u2,j, Fk,y,z
(u2,j)) ..., (uTU,j, Fk,y,z(uTU,j)) interior two adjacent coordinate points structure Fk,y,z(xj) linear fit function
LFk,y,z(xj)=pj*xj+qj.Work as ut,j≠ut+1,jWhen,Due to
Vk,y,z,jIn there is TU coordinate, therefore according to above-mentioned rule, TU-1 linear fit function can be obtained in theory.For
This can adopt rearmounted rule, always build linear fit function with most latter two adjacent coordinate points.Such as, if uTU,j≠
uTU-1,j, thenIf uTU,j=uTU-1,j=...=
uTU-t+1,j≠uTU-t,j, then
Second and the output of embodiment of the third linear fit be farthest public branch node mkEach front point
Each branch function LF in Zhi Jiediank,i,jLinear fit function LF can be obtainedk,i,jThe parameter set with regard to each input variable
LFPQk,i,j={ (p1, q1), (p2, q2) ..., (pNi, qNi), numbers of the wherein Ni for input variable.Linear fit function
LFk,i,jBy parameter set LFPQk,i,jRepresent.
The embodiment of the first linear fit has many measured datas to limit by the way of multiple linear regression.Therefore
The preferential embodiment for adopting second and the third linear fit of the invention.As linear fit is iterative process, input vector
Collection U increases in iteration, therefore, linear fit parameter p obtained using the embodiment of the third linear fitjAnd qjThan
The embodiment of two kinds of linear fits more changing property.Therefore, the present invention is with the embodiment of the third linear fit as optimum
Embodiment.
For convenience of follow-up explanation, the output { a of the embodiment of the first linear fit0, a1..., aNiCan be converted into
Two and the third linear fit embodiment output LFPQk,i,j={ (p1, q1), (p2, q2) ..., (pNi, qNi), conversion
Following method can be adopted：pi=ai；qi=Fy,z(u1)-u1,i×ai.Thus, the embodiment of three kinds of linear fits possesses phase
With form or the output result of type.
6th, feasible interval calculating
This step is aforementioned step S5.
Feasible interval refers to and produces through branch node B and sentencing of covering that branch node B completes that the covering of condition/judgements lacks
The feasible interval of the possible valued combinations of each condition in fixed and judgement.
The logical expression of i-th branch node (not contain B) before B in associated path k of branch node B
ck,iAs a example by, ck,i=Fk,i,1(X)rel10op1Fk,i,2(X)rel20op2...ops-1Fk,i,s(X)rels0, op hereiRepresent with
(And) or or (Or) logical operator, s represents the quantity of condition on upper i-th branch node of path k.When execution source to be measured
During code, cover the means suitable logical expression for (B not being contained) before branch node B in associated path k of branch node B
ck,iIt is false.Wherein, certain branch function Fk,i,j(X) when more than 0, logical expression ck,iIt is false to be possible to, it means that only
Work as Fk,i,j(X) portion of before branch node B (without B) in associated path k of branch node B could when more than 0, be covered
Point.When X is located at interval Dk,i,jWhen, meet Fk,i,j(X) more than 0, then interval Dk,i,jFor feasible interval.Due to interval Dk,i,jWith point
Prop up function Fk,i,j(X) it is corresponding, interval Dk,i,jIt is also called branch function Fk,i,j(X) feasible interval.Merge logical expression
ck,iThe feasible interval D of each interior branch functionk,i,jLogical expression c is obtainedk,iFeasible interval Dk,i.Logical expression
Formula ck,iFeasible interval Dk,iIt is also called the feasible interval of upper i-th branch node of path k.
Again with the associated path k top set node B's (assuming B as h-th branch node on the k of path) of branch node B
Logical expression ck,hAs a example by, ck,h=Fk,h,1(X)rel10op1Fk,h,2(X)rel20op2...ops-1Fk,h,s(X)rels0, here
opiRepresent with (And) or or (Or) logical operator, s represents the quantity of condition on upper h-th branch node of path k.When holding
During row source code to be measured, cover B and complete the judgement/condition valued combinations requirement logical expression that condition/judgement covering lacks
Formula ck,hIt is false.Wherein, certain branch function Fk,h,j(X) when (j ≠ w) is more than 0, logical expression ck,hBe possible to be it is false and
Fk,h,j(X)relj0 corresponding condition obtains unlapped value.This means only to work as Fk,h,j(X), when more than 0, B could be covered
Complete condition/judgement and cover the judgement/condition valued combinations for lacking.When X is located at interval Dk,h,jWhen, meet Fk,h,j(X) it is big
In 0, then interval Dk,h,jFor feasible interval.Due to interval Dk,h,jWith branch function Fk,h,j(X) it is corresponding, interval Dk,h,jIt is also called
Branch function Fk,h,j(X) feasible interval.Merge logical expression ck,hThe feasible interval D of each interior branch functionk,h,jI.e.
Logical expression c is obtainedk,hFeasible interval Dk,h.Logical expression ck,hFeasible interval Dk,hIt is also called path k upper h-th
The feasible interval of branch node.
Merge farthest public branch node mkThe feasible interval D of each front branch nodek,iBranch node B can be obtained
Feasible interval I in associated path kk.Therefore, step S5 is divided into following four step：
S51：Calculate each branch node for (containing B) before branch node B in each associated path k of branch node B
In each branch function feasible interval Dk,i,j；
S52：Merge the feasible interval D of each branch function in each branch nodek,i,jObtain the feasible of each branch node
Interval Dk,i；
S53：Merge farthest public branch node mkThe feasible interval D of front each branch nodek,iObtain branch node B to exist
Feasible interval I in associated path kk。
S54：Merge feasible interval Is of the branch node B in each associated path kk, obtain the feasible interval of branch node B
I。
In step S51, branch function Fk,i,j(X) feasible interval Dk,,jBy branch function Fk,i,j(X) corresponding Linear Quasi
Close function LFk,i,j(X) calculate and obtain.The embodiment of aforementioned three kinds of linear fits is output as LFPQk,i,j={ (p1, q1),
(p2, q2) ..., (pNi, qNi)}.Use LFPQk,i,jThe linear fit function LF of expressionk,i,j(X) it is to distinguish by each input variable
The linear fit collection of functions of definition, is calculating feasible interval Dk,i,jWhen, can in the same way, i.e., by each input variable point
Ji Suan not inequation LFk,i,j(xk)reli0=pk*xk+qk reli0 with regard to xwInterval solutions Dk,i,j,w, Dk,i,j,wFor path k
The feasible interval of w-th input variable of j-th branch function of upper i-th branch node.It will be appreciated by those skilled in the art that so
The x for obtainingwInterval solutions Dk,i,j,wIt is very wide, generally (a, ∞) or (- ∞, form a).Notice linear fit function LFk,i,j
(xw)=pw*xw+qwBy Vk,y,z,j={ (u1,j, Fk,y,z(u1,j)), (u2,j, Fk,y,z(u2,j)) ..., (uTU,j, Fk,u,z
(uTU,j)) calculate acquisition.Therefore can be by feasible interval Dk,i,j,wIt is limited in { u1,j, u2,j..., uTU,jIn related interval RD,
I.e. feasible interval Dk,i,j,wIt is the common factor of related interval RD and the interval solutions of inequation.Related interval RD can be：(2×
umin,j- umax,j, umax,j) or (2 × umin,j- umax,j, 2 × umax,j- umin,j) or (umin,j, 2 × umax,j- umin,j) or
(umin,j, umax,j) or (ura,j, urb,j) or (2 × ura,j- urb,j, urb,j) or (2 × ura,j- urb,j, 2 × urb,j- ura,j) or
(ura,j, 2 × urb,j- ura,j) etc..Wherein, umin,j=min { u1,j, u2,j..., uTU,j, umax,j=max { u1,j,
u2,j..., uTU,j}；ura,j, urb,j∈{u1,j, u2,j..., uTU,j, it is aforementioned calculatingIn, ura,j=min { uTU-t, j, uTU,j, urb,j=max { uTU-t, j,
uTU,j}.Each input variable can obtain a feasible interval Dk,i,j,w, it is hereby achieved that branch branch function Fk,i,j(X)
Feasible interval Dk,i,j={ Dk,i,j,1, Dk,i,j,2..., Dk,i,j,s, numbers of the wherein s for input variable.
The feasible interval D of branch node in step S52k,iUsing with logical expression ck,i=Fk,i,1(X)
rel10op1Fk,i,2(X)rel20op2...ops-1Fk,i,s(X)rels0 consistent friendship union operation rule is merged：
Dk,i=Dk,i,1 op1 Dk,i,2 op2 ...ops-1 Dk,i,s。
Feasible interval I in step S53kMerged using intersection operation rule：
Wherein mkFor farthest public branch node.
In step S54, feasible interval I is merged using union operation rule：
Wherein NBAssociated path for branch node B concentrates path number；
The feasible interval I={ I for finally giving1, I2..., Is}.Numbers of the wherein s for input variable, IiIt is defeated for i-th
Enter the feasible interval of variable.The feasible interval I of input variableiIt is the possible segmentation Interval Set being made up of possible segmentation interval, Ii=
{Ii,1, Ii,2..., Ii,Mi, wherein, Ii,jInterval for i-th input variable, j-th possible segmentation, Mi is i-th input variable
Possible segmentation interval number.
7th, the extension of input vector collection
This step is aforementioned step S6, and the input that each input variable is randomly selected in feasible interval I is worth to survey
Examination data set is added into input vector collection U.
Due to I={ I1, I2..., Is, and Ii={ Ii,1, Ii,2..., Ii,Mi}.Only need to from each be input in this step
The possible segmentation Interval Set I of variableiEach possible segmentation interval in randomly choose a data can the new input vector of structure
Collection T.As each input variable has Mi possible segmentation interval, therefore each input variable can therefrom select Mi randoms number.
M will can be produced after the random number combination of each possible segmentation of all of input variable interval gained1×M2×...×MsIt is individual defeated
Incoming vector.Build after new input vector collection T is merged with original input vector collection U and obtain new input vector collection U.
The input variable more special for part, its correspondence possible segmentation interval number only have 1.Under this situation, can
With the random value in extension interval.Interval point of left extension of extension is interval, and right extension is interval.That assumes i-th input variable can
The interval unique segment identifier of row is shown in Ii,1For (a, b), a<B, then left extension is interval for (2b-a, a), right extension interval is：(b,2a-
b)。
8th, the end of iteration and iteration
This step is aforementioned step S7 and step S8.
Step S7, repeats step S3 to S6 until number of times that step S3 to S6 is performed reaches the number of times of restriction.Step
S3 to S6 is the process of iterative cycles, and the iterative cycles of wheel per Jing mono-, the input vector in input vector collection U also change therewith.
If through the branch node B when in step s3, there is an input vector execution source program to be measured in input vector collection U
And the valued combinations that judgement and judgement conditional do not occur in one B of covering, then record the input vector and save as the branch
One test data of point B, if there is no unlapped judgement and judge the valued combinations of conditional in the branch node B,
Test data set of the test data set for then recording as the branch node B, whole iterative process terminate.If iterative cycles
Number of times when reaching restriction number of times set in advance, show that branch node B can not complete the covering of all of condition/judgements.If
The unreachable test data set for needing also exist for returning branch node B of branch node B, i.e. step S8.
Step S8, the test data that covers of condition/judgement of the test data set of the B of return recording as branch node B
Collection.
Claims (4)
1. a kind of automatic generation of test data covered towards condition criterion, including analysis source program to be measured obtains source to be measured
The step of set of the associated path collection of the set in all paths of program and the set BS of all branch nodes and branch node,
Characterized in that, each branch node B during the method is also included to branch node collection BS performs following steps：
S1：Each condition of each branch node on each path before B is concentrated according to the associated path of B, is built with regard to defeated
Enter the branch function F of variable vector Xk,i,j(X)；The input variable vector X is that the source program input variable to be measured is constituted
Vector；The branch function Fk,i,j(X) represent the branch function of j-th condition of upper i-th branch node of path k；The road
Footpath k is through branch node B；Sequence numbers of the k for the associated path concentration path of branch node B；
S2：Build initial input vector collection U；The initial input vector collection U includes at least two input vectors；
S3：Calculate when input using each input vector in input vector collection U as source program to be measured performs source program to be measured with
The associated path of the branch node B concentrates each branch function F on each path on each branch node before Bk,i,j
(X) the set V and farthest public branch node m of branch function value are obtainedk；Execution route Jing when source program to be measured is performed
The value of each condition in the judgement of B being calculated when crossing branch node B and being judged；If there is an input in input vector collection U
When vector performs source program to be measured execution route through branch node B and in making the judgement of B or judging certain condition produce it is new
Value, then record the input vector as a test data of the branch node B；If the test of the branch node B
Data set cover the judgement of B and judge in the be possible to value of each condition, then return the test data set as the branch
The test data set of node B；The farthest public branch node mkFor the farthest public branch node of path k；
S4：Each branch's letter on each path on each branch node before B is concentrated according to the associated path of branch node B
The set V of numerical value and farthest common node mkFarthest common node m is located on build path kkEach branch node before
The fitting function LF of each branch functionk,i,j；LFk,i,jRepresent the branch function of j-th condition of upper i-th branch node of path k
Linear fit function；
S5：Each branch's letter on each path in each branch node before B is concentrated according to the associated path of branch node B
Several linear fit function LFk,i,jFeasible interval I is calculated with source program to be measured；The feasible interval I={ I1, I2..., Is,
Wherein, numbers of the s for input variable, IiFor the feasible interval of i-th input variable, Ii={ Ii,1, Ii,2..., Ii,Mi, its
In, Ii,jInterval for i-th input variable, j-th possible segmentation, Mi is i-th input variable possible segmentation interval number；
S6：The input that each input variable is randomly selected in feasible interval I is worth to test data set and adds to input vector collection U
In；When Mi is 1, from IiFor feasible interval unique piecewise interval I of i-th input variablei,1Extension interval in take at random
Value builds test data；
S7：Step S3 to S6 is repeated until number of times that step S3 to S6 is performed reaches the number of times of restriction；
S8：Return all records the branch node B test data as the branch node B test data set；
Step S5 includes：
S51：The associated path for calculating branch node B concentrates each branch node Zhong Ge branches letter on each path before B
Several feasible interval Dk,i,j；The feasible interval Dk,i,jFor upper i-th branch node of path k j-th branch function it is feasible
It is interval；
S52：Merge the feasible interval D of each branch function in each branch nodek,i,jObtain the feasible interval of each branch node
Dk,i；The feasible interval Dk,iFor the feasible interval of upper i-th branch node of path k；
S53：Merge farthest public branch node mkThe feasible interval D of front each branch nodek,iBranch node B is obtained in path k
On feasible interval Ik；The feasible interval IkFor the feasible interval of path k；
S54：Merge branch node B feasible interval I on each pathk, obtain the feasible interval I of branch node B.
2. the automatic generation of test data for covering towards condition criterion as claimed in claim 1, it is characterised in that
Initial input vector collection U described in step S2 includes Ni+1 input vector, and the Ni is input variable
Number；Initial input vector collection U={ u0, u1, u2..., uNi, wherein ui={ ui,1, ui,2..., ui,Ni}；Input vector u0
For random generation, uiWith u0Meet relation：Wherein i, j ∈ 1,2 ..., Ni }；I represent input to
Sequence number in quantity set, j represent the sequence number of input variable.
3. the automatic generation of test data for covering towards condition criterion as claimed in claim 1, it is characterised in that described
Step S3 includes：
S31：Unenforced input vector u is obtained from input vector collection Ut；
S32：Judge unenforced input vector utWhether there is；If there is no unenforced input vector, then execution step
S34；Otherwise execution step S33；
S33：According to utPerform the public affairs that source code to be measured obtains each path of associated path concentration of the path and branch node B for performing
Common branch node mk,t, calculate the branch function F of each branch node on the k of path before branch node Bk,i,j(X) value,
The associated path of branch node B concentrates each branch function F of each branch node in all pathsk,i,j(X) value constitutes set
V；If the path for performing is through branch node B, an associated path of the path for performing for branch node B, according to the association
The value of each condition during the branch function of each condition is worth to the judgement of branch node B and judges in the judgement of B on path；It is false
Such as the judgement of B or judgement conditional generate new value, then record the input vector as the branch node B one surveys
Examination data；If the test data set of B covers the judgement of B and the be possible to value of each condition in judging, the test number is returned
According to collection is as the test data set of the branch node B and terminates to generate whole branch node B the process of test data；
S34：{ the m recorded by the path k concentrated from the associated path of branch node Bk,1, mk,2..., mk,TUIn choose farthest public
Common branch node mk；
Wherein t is input vector utSequence number in input vector collection U, TU are the number of element in input vector collection U.
4. the automatic generation of test data for covering towards condition criterion as claimed in claim 1, it is characterised in that described
Step S4 is comprised the following steps：
S41：The associated path of input vector collection U and branch node B concentrates each branch node on each path before B
The set V of each branch function value and farthest common node mkBuild coordinate set Vk,y,z,j={ (u1,j, Fk,y,z(u1,j)), (u2,j,
Fk,y,z(u2,j)) ..., (uTU,j, Fk,y,z(uTU,j))}；
S42：With Vk,y,z,j={ (u1,j, Fk,y,z(u1,j)), (u2,j, Fk,y,z(u2,j)) ..., (uTU,j, Fk,y,z(uTU,j)) in two
Individual adjacent coordinate points build Fk,y,z(xj) linear fit function LFk,y,z(xj)=pj*xj+qj, obtain parameter set LFPQk,i,j
={ (p1, q1), (p2, q2) ..., (pNi, qNi)}；
Wherein, Vk,y,z,jThe coordinate set of z-th branch function j-th input variable of correspondence of upper y-th branch node of path k is represented,
Wherein y≤m；ui,jFor the value of j-th input variable of i-th input vector in input vector collection U；Fk,y,z(xj) on the k of path
Function of y-th node, z-th branch function with regard to j-th input variable；LFk,y,z(xj) for the upper y-th node z of path k
Linear fit function of the individual branch function with regard to j-th input variable；LFPQk,i,jFor upper i-th node, j-th branch of path k
The parameter set of the linear fit function of function；TU is the number of input vector in input vector collection U；Ni is the individual of input variable
Number.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410271284.8A CN104050083B (en) | 2014-06-17 | 2014-06-17 | Condition/Decision-coverage-oriented test data automatic generation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410271284.8A CN104050083B (en) | 2014-06-17 | 2014-06-17 | Condition/Decision-coverage-oriented test data automatic generation method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104050083A CN104050083A (en) | 2014-09-17 |
| CN104050083B true CN104050083B (en) | 2017-05-03 |
Family
ID=51502969
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410271284.8A Active CN104050083B (en) | 2014-06-17 | 2014-06-17 | Condition/Decision-coverage-oriented test data automatic generation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104050083B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104881359B (en) * | 2015-06-10 | 2017-11-24 | 南京大学 | A kind of automatic generation of test data covered with linear fit realizing route |
| CN108647146B (en) * | 2018-05-11 | 2021-06-08 | 北京信息科技大学 | Test case generation method for judging combination coverage based on correction condition |
| CN112463607B (en) * | 2020-11-26 | 2024-03-15 | 苏州洞察云信息技术有限公司 | Determination statement value confirmation method, device and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101145130A (en) * | 2007-08-02 | 2008-03-19 | 中兴通讯股份有限公司 | Simple network management protocol software test device and test method |
| CN101162426A (en) * | 2007-11-15 | 2008-04-16 | 中兴通讯股份有限公司 | Example design test method for testing based on data driving and automatically testing platform |
| CN101814055A (en) * | 2010-04-06 | 2010-08-25 | 南京大学 | Sampling method for test cases in clusters |
| CN102368226A (en) * | 2011-10-10 | 2012-03-07 | 南京大学 | Method for automatically generating test cases based on analysis on feasible paths of EFSM (extended finite state machine) |
-
2014
- 2014-06-17 CN CN201410271284.8A patent/CN104050083B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101145130A (en) * | 2007-08-02 | 2008-03-19 | 中兴通讯股份有限公司 | Simple network management protocol software test device and test method |
| CN101162426A (en) * | 2007-11-15 | 2008-04-16 | 中兴通讯股份有限公司 | Example design test method for testing based on data driving and automatically testing platform |
| CN101814055A (en) * | 2010-04-06 | 2010-08-25 | 南京大学 | Sampling method for test cases in clusters |
| CN102368226A (en) * | 2011-10-10 | 2012-03-07 | 南京大学 | Method for automatically generating test cases based on analysis on feasible paths of EFSM (extended finite state machine) |
Non-Patent Citations (1)
| Title |
|---|
| 分量线性拟合函数驱动的测试数据自动生成技术研究;丁文旭;《中国优秀硕士学位论文全文数据库 信息科技辑》;20121015(第10期);第5-28页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104050083A (en) | 2014-09-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103942147B (en) | A kind of automatic generation of test data of linear fit function drive | |
| Hamraz et al. | A matrix-calculation-based algorithm for numerical change propagation analysis | |
| Huang et al. | Forecasting container throughput of Qingdao port with a hybrid model | |
| Taleb et al. | Big data pre-processing: closing the data quality enforcement loop | |
| Sharma et al. | Design and evaluation of layout alternatives to enhance the performance of industry | |
| CN104050083B (en) | Condition/Decision-coverage-oriented test data automatic generation method | |
| Plehn et al. | Assessing the impact of changes and their knock-on effects in manufacturing systems | |
| JPWO2018079225A1 (en) | Automatic prediction system, automatic prediction method, and automatic prediction program | |
| CN104050082B (en) | Test data automatic generation method oriented toward modified condition/decision coverage | |
| Janssens et al. | Modeling and enacting enterprise decisions | |
| Guttmann | Relation-algebraic verification of Prim’s minimum spanning tree algorithm | |
| CN113159441A (en) | Prediction method and device for implementation condition of banking business project | |
| JP6909596B2 (en) | Knowledge model construction system and knowledge model construction method | |
| Seref et al. | Software code maintainability: a literature review | |
| Lantow et al. | Towards a classification framework for approaches to enterprise architecture analysis | |
| Ljubić et al. | Stochastic survivable network design problems | |
| Maravilha et al. | A new algorithm based on differential evolution for combinatorial optimization | |
| CN104881359B (en) | A kind of automatic generation of test data covered with linear fit realizing route | |
| Anand et al. | Information flow in parameterized quantum circuits | |
| Peñaloza | Inconsistency-tolerant instance checking in tractable description logics | |
| Barbier et al. | Model-driven engineering applied to crop modeling | |
| JP6618454B2 (en) | COMPUTER DEVICE, METHOD, AND PROGRAM | |
| CN114254764B (en) | Feedback-based machine learning model searching method, system, equipment and medium | |
| Tiwari et al. | To find the effectiveness of barriers in reverse logistics by using ISM | |
| Jamkhaneh et al. | Fuzzy system reliability analysis based on confidence interval |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |