CN103793597B - Model similarity measuring method based on complete backbone subsystems - Google Patents
Model similarity measuring method based on complete backbone subsystems Download PDFInfo
- Publication number
- CN103793597B CN103793597B CN201410019455.8A CN201410019455A CN103793597B CN 103793597 B CN103793597 B CN 103793597B CN 201410019455 A CN201410019455 A CN 201410019455A CN 103793597 B CN103793597 B CN 103793597B
- Authority
- CN
- China
- Prior art keywords
- transition
- model
- complete
- trunk subsystem
- similarity
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- 230000008859 change Effects 0.000 claims abstract description 8
- 230000007704 transition Effects 0.000 claims description 42
- 230000001364 causal effect Effects 0.000 claims description 10
- 238000011524 similarity measure Methods 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 2
- 238000005259 measurement Methods 0.000 abstract description 5
- 230000009286 beneficial effect Effects 0.000 abstract 1
- 230000000694 effects Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a model similarity measuring method based on complete backbone subsystems. The model similarity measuring method sequentially includes the steps of decomposing a model to obtain the complete backbone subsystems, building behavior profiles for the multiple complete backbone subsystems, carrying out similarity comparison on the complete backbone subsystems, and calculating the model similarity. In order to improve the measuring accuracy of the model similarity, the model similarity measuring method based on the complete backbone subsystems is provided, and according to the model similarity measuring method, the cause-and-effect behavior profiles are improved in three aspects, namely, the first aspect of bringing the importance degree of change pairs into the measurement range, the second aspect of depicting the cause-and-effect behavior profiles in a fine grit mode, and the third aspect of deepening the consistency of the change pairs. Accuracy of measurement on the model similarity can be improved from a behavior perspective, and support is provided for operations such as model retrieval, model combining and model reuse. The model similarity measuring method can be further used for software behavior predictability assessment and is beneficial for improving the software credibility.
Description
Technical field
The present invention relates to a kind of business model method for measuring similarity.
Background technology
Nowadays, business process model is commonly used to industry analysis operation flow or creates new model.Large scale business enterprise leads to
Often have procedural model storehouse, include hundreds of or thousands of models.These models, by different staff developments, are that enterprise is precious
Expensive intelligence wealth.Manage and operate these models for convenience, such as pattern search, model combination etc., need study model phase
Like degree measure.
There are many distortion measures at present.Method based on mark equivalence, Mutual simulation and branch's Mutual simulation
The binary whether being can only be given answer, for example, represent of equal value with 1, represent non-equivalence with 0.But in actual applications, people
It is often desirable to know the difference having much degree between different models, the application that this allows for these methods is restricted.Mark phase
It is defined as the ratio of the sum of the number of identical mark and mark in two models like degree.The method and mark equivalent method are faced equally
Problem it is simply that due to exist in model concurrent, select or ring structure, can cause that the set of mark is very big or even infinite, this gives
Calculating brings very big difficulty.And mark similarity based method is still excessively strict, two marks are slightly different all to cause whole mark not
Can coupling.For this reason, have scholar propose based on observation behavior measure, by comparing the typical behaviour in event log,
Calculate precision and recall index, with the similarity of measurement model.The method is limited by the completeness of event log.Base
It is considered as the improvement to the method based on movable dependency graph in the method changing syntopy, but this method only focuses on directly
Cause effect relation, have ignored indirect causal association, enough sensitivity lacked to the change of model.Method based on cause and effect footprint
Computational efficiency is too low.Weidlich et al. proposes the measure based on causal actant profile, and cause and effect outline definition is
Tight order relation, the set of cross reference, mutex relation and cooccurrence relation.The method can extremely efficiently calculate free choice net
Distortion, but as preceding method, the accuracy of measurement is still not enough.
Content of the invention
In order to improve the accuracy of measurement of distortion, the present invention provides a kind of model phase based on complete trunk subsystem
Like degree measure, the method is transformed to causal actant profile in terms of three.One be by transition to importance degree receive
Enter to measure category;Two is that more fine granularity portrays causal actant profile;Three be in-depth transition to consistent degree.
The technical scheme that the present invention is given is:
A kind of distortion measure based on complete trunk subsystem is it is characterised in that include successively to model
Decompose obtain complete trunk subsystem, set up behavior profile for a complete trunk subsystem, to carry out complete trunk subsystem similar
Degree compares, computation model similarity, concretely comprises the following steps:
1)For transition to importance degree, present invention employs the measure based on complete trunk subsystem.Exist first
One short circulation transition is increased on Workflow net, by seeking the extension subnet of transition invariant, model decomposition is become some complete
Trunk subsystem.One complete trunk subsystem is made up of a simple trunk subsystem and some loop structures, and it describes
The execution route of one class example.These paths at least perform all activities in simple trunk subsystem it is also possible to execute any
Activity in secondary loop structure.The present invention not directly to transition to importance degree calculate, but by each complete master
Emerge from dry subsystem Similarity Measure.The number of times of one transition appearance in each complete trunk subsystem is more, its ginseng
More with the number of times calculating, the impact for distortion is also bigger, thus showing that it is more important.
2)In order to more fine granularity portrays behavior profile, behavior profile is defined on complete trunk subsystem the present invention, often
Individual complete trunk subsystem has a behavior profile, and the set of these behavior profiles constitutes the behavior profile of model.For reducing
Information loss, no longer concurrency relation is incorporated in cross reference in behavior profile, suitably increases redundancy transition right, for carving simultaneously
Draw the common portion of simple trunk subsystem and its loop structure.The so common factor of concurrency relation and cross reference and Yan Xuguan
System may be all no longer sky with the common factor of cross reference.The transition of these redundancies can change the constraint to consistent degree as calculating.
Redundancy transition generate to by simple trunk subsystem.
3)On the basis of the similarity of causal actant profile is built upon changing to consistent degree.The present invention will change consistent degree
Be defined as with this transition to the match is successful transition to number and this transition Corresponding matching the ratio to number for the transition.So
Transition are mapped to certain value on [0,1] interval to consistent degree, and are no longer 0 or 1.This than original method computationally again before
Go a step further.
4)Each complete trunk subsystem carries out Similarity Measure with the complete trunk subsystem of alternate model successively, definition
For all transition to consistent degree and with transition to total ratio.Last distortion is defined as each complete trunk subsystem
The meansigma methodss of maximum similarity.
The angle tolerance distortion of the main subordinate act of the inventive method, does not consider that label similarity is similar with structure
Degree.The present invention by change to importance degree, transition to consistent degree all include measuring similarity category, more explication behavior
The concept of profile.The invention has the advantages that:Distortion tolerance can be improved accurately with subordinate act angle, be model
The operation such as retrieval, model combination, Model Reuse provides to be supported.The inventive method can be additionally used in the assessment of software action predictability,
Be conducive to improving software credibility.
The inventive method carrys out descriptive model using the Workflow net of unrestricted choice, and one is because that Workflow net has become current
The description most commonly used formalized model of workflow;Two are because that most of Workflow Management Systems only allow free choice net
Workflow;Three are because that free choice net has been widely studied and it can be readily appreciated that the checking of its soundness attribute and behavior wheel
Wide acquisition all can complete in polynomial time.
Brief description
Fig. 1 asks for process for complete trunk subsystem.
Fig. 2 is transition to importance degree example.
Fig. 3 is that concurrency relation and cross reference are separated necessity example.
Fig. 4 be increase redundancy transition to necessity example.
Fig. 5 is transition consistent degree refinement example.
Fig. 6 is full solution procedure example.
Fig. 7 is complete trunk subsystem.
Specific embodiment
It is described further with reference to the technical scheme that example is protected to the present invention.
What Fig. 1 illustrated complete trunk subsystem asks for process.Fig. 1 (a) is a sound unrestricted choice Workflow net,
In fact changing t* plus short a circulation between place and terminate storehouse institute, an arc pointing to t* from terminate storehouse institute, and one
Point to the arc of initial storehouse institute from t*, just constitute a short circulating net.B (), (c), (d) are three simple trunk subsystems, (e) and
F () is loop structure.By synthesis, can get (g), (h), (i) three complete trunk subsystems.
Notice that transition are rightNot only belonged to tight order relation but also belonged to cross reference in (h).Transition are right?
Belong to cross reference in (g), but be not belonging to tight order relation, and both belonged to cross reference in (h) and belonged to tight order relation.Below
The superiority of the inventive method is described by several simply examples.
The inventive method that can illustrate Fig. 2 can embody change to importance.Compared to model (a), model (b) is exchanged
The position of movable A and B, and the position of mobile C and E exchanged by model (c).According to the method for causal actant profile, (b) with
A the similarity of () is equal to the similarity of (c) and (a).But from figure 2 it can be seen that in (a) two paths all through A and B,
And only one paths through C and E it is clear that transition to (A, B) than transition more important to (C, E).Can obtain with the present invention
B () is 0.875 with the similarity of (a), and (c) and the similarity of (a) are 0.938, embody A and B very well and exchange to master mould
The fact that impact is bigger.
Fig. 3 can illustrate the benefit separating concurrency relation and cross reference.According to the method for causal actant profile, model
A () and model (b) can not be compartmentalized, their similarity is 1.And use the inventive method, can get the similar of them
Degree is 0.875.Because (B, C) belongs to tight order relation and cross reference in (a), and in (b), but belongs to concurrency relation and friendship
Fork relation.
Fig. 4 can illustrate to increase redundancy transition to being a need for.Increase after redundancy relationship it is possible to effective district sub-model
(a) and model (b).
Fig. 5 can illustrate the in-depth to transition to consistent degree for the inventive method.In Fig. 5, background and dotted line indicate model
Movable corresponding relation in (a) and model (b).According to the method for causal actant profile, transition can not successful to (B, C)
Join, because (B, C) belongs to tight order relation in (a), and in (b), (B2, C1) but belongs to concurrency relation.So the one of (B, C)
Cause degree is 0.However, other three transition are to (B1, C1), (B1, C2) and (B2, C2) is all can to mate with (B, C).At this
In inventive method technology, the consistent degree of (B, C) will correspond to 0.75.Obviously, this is more reasonable than the method for causal actant profile, meter
Calculate also finer in granularity.
Fig. 5 also illustrates another major issue:Cooccurrence relation can not solve behavior profile precision problem very well.From figure
Can be seen that sometimes cooccurrence relation in 4 example and seem too weak.And in Figure 5, cooccurrence relation but seem too strict.Example
As in Fig. 5 (a), A1, A2 and B are cooccurrence relations, because A3 and A4 can not be cooccurrence relation with B1 and B2 simultaneously in Fig. 5 (b), because
Unsuccessful during this coupling, final mask similarity only has 0.63.But over all, this cooccurrence relation is at (a) and (b)
In all be exist.It is 0.817 it is shown that more preferable reasonability that application the inventive method technology calculates acquired results.
Finally, we pass through a simply example(Fig. 6)The overall process of the inventive method technical operation is described.First,
Model (a) and (b) are decomposed, obtains complete trunk subsystem, as shown in Figure 7.
Second step, sets up behavior profile for individual complete trunk subsystem.
3rd step, carries out complete trunk subsystem similarity-rough set, and table 1 lists comparative result.
Table 1 model (a) and model (b) each complete trunk path similarity-rough set
Similarity | a1 | a2 | a3 | a4 |
b1 | 1 | 0.72 | 0.72 | 0.4 |
b2 | 0.4 | 0.62 | 0.62 | 1 |
Finally, computation model similarity is (1+0.72+0.72+1+1+1)/6=0.91.
Claims (1)
1. a kind of distortion measure based on complete trunk subsystem is it is characterised in that include successively model is divided
Solution obtain complete trunk subsystem, set up behavior profile for each complete trunk subsystem, to carry out complete trunk subsystem similar
Degree compares, computation model similarity, concretely comprises the following steps:
1) employ the measure based on complete trunk subsystem, increase short circulation transition first on Workflow net,
By seeking the extension subnet of transition invariant, model decomposition is become some complete trunk subsystems;One complete trunk subsystem
It is made up of a simple trunk subsystem and some loop structures, the described complete trunk subsystem description execution of one class example
Path;
2) behavior profile is defined on complete trunk subsystem, each complete trunk subsystem has a behavior profile, this
The set of a little behavior profiles constitutes the behavior profile of model;Concurrency relation is not incorporated in cross reference in behavior profile, simultaneously
Increase redundancy transition right, for portraying the common portion of simple trunk subsystem and its loop structure;The transition of these redundancies are made
For calculating the constraint to consistent degree for the transition;Redundancy transition generate to by simple trunk subsystem;
3), on the basis of the similarity of causal actant profile is built upon changing to consistent degree, transition is defined as to consistent degree and is somebody's turn to do
Transition to the match is successful transition to number and this transition Corresponding matching the ratio to number for the transition, transition are to consistent degree quilt
It is mapped to certain value on [0,1] interval;
4) each complete trunk subsystem carries out Similarity Measure with the complete trunk subsystem of alternate model successively, is defined as institute
Have change to consistent degree and with transition to total ratio, it is maximum that last distortion is defined as each complete trunk subsystem
The meansigma methodss of similarity.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410019455.8A CN103793597B (en) | 2014-01-16 | 2014-01-16 | Model similarity measuring method based on complete backbone subsystems |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410019455.8A CN103793597B (en) | 2014-01-16 | 2014-01-16 | Model similarity measuring method based on complete backbone subsystems |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103793597A CN103793597A (en) | 2014-05-14 |
CN103793597B true CN103793597B (en) | 2017-02-22 |
Family
ID=50669259
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410019455.8A Expired - Fee Related CN103793597B (en) | 2014-01-16 | 2014-01-16 | Model similarity measuring method based on complete backbone subsystems |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103793597B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107967545B (en) * | 2016-10-19 | 2022-05-17 | 中国辐射防护研究院 | Method for calculating Birnbaum importance and structure importance of subsystem component in probability safety evaluation |
CN108733725B (en) * | 2017-04-24 | 2022-03-25 | 西门子(中国)有限公司 | Method and system for establishing network semantic model of target system |
CN108647380B (en) * | 2018-04-02 | 2020-07-14 | 同济大学 | Concurrent system error detection method, system, medium and device based on PD-Net |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7487543B2 (en) * | 2002-07-23 | 2009-02-03 | International Business Machines Corporation | Method and apparatus for the automatic determination of potentially worm-like behavior of a program |
CN101894325A (en) * | 2010-07-13 | 2010-11-24 | 华南理工大学 | Workflow net-based semantic web service behavior matching method |
-
2014
- 2014-01-16 CN CN201410019455.8A patent/CN103793597B/en not_active Expired - Fee Related
Also Published As
Publication number | Publication date |
---|---|
CN103793597A (en) | 2014-05-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113378011B (en) | Construction method and system of complex product assembly digital twin body | |
CN103093023A (en) | Computer aided design model analysis system | |
CN104281525B (en) | A kind of defect data analysis method and the method utilizing its reduction Software Testing Project | |
CN103793597B (en) | Model similarity measuring method based on complete backbone subsystems | |
CN110543530A (en) | Method for realizing rapid coding of pumped storage power station model component by utilizing Dynamo | |
CN107491508A (en) | A kind of data base querying time forecasting methods based on Recognition with Recurrent Neural Network | |
CN103606032B (en) | A kind of method in two dimension power grid GIS data set | |
Kawakami-Arevalo et al. | Increased Productivity through a Production Model Based on Lean Manufacturing and SLP Tools in Small Furniture Manufacturing Workshops | |
Li et al. | Research and application of computer aided design system for product innovation | |
CN104794010B (en) | Information exchange optimization method in a kind of system | |
CN108984746A (en) | A kind of process model method for digging of more concurrent 2 degree of circulations of triangle | |
CN114692403A (en) | Binary decision diagram-based efficient complex system reliability evaluation method | |
CN109685453B (en) | Method for intelligently identifying effective paths of workflow | |
CN112613209A (en) | Full hexahedron unit encryption method based on finite element model | |
Fang et al. | Online incremental mining based on trusted behavior interval | |
Shikhli et al. | Data Acquisition Model for Analyzing Schedule Delays Using KDD: Knowledge Discovery and Datamining | |
CN104965923A (en) | Cloud computing application platform building method for generating cash flow statement | |
Zhang et al. | Research Review of Design Pattern Mining | |
FAHRUDIN | Sequence clustering in process mining for business process analysis using k-means | |
Singh | Software module clustering using metaheuristic search techniques: A survey | |
Guo et al. | The Application Research about Data Warehouse Based on ERP | |
CN113190929B (en) | Workpiece family construction method of delayed reconstruction manufacturing system based on machine learning | |
CN109460415A (en) | A kind of similar fixture search method based on N-dimensional vector included angle cosine | |
CN108957297B (en) | Method for constructing circuit fault diagnosis expansion decision tree by introducing working condition information | |
Bruneau et al. | A three-level signature by graph for Reverse Engineering of mechanical assemblies |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170222 |