CN104835015B - Workflow mining method based on predecessor task - Google Patents
Workflow mining method based on predecessor task Download PDFInfo
- Publication number
- CN104835015B CN104835015B CN201510272608.4A CN201510272608A CN104835015B CN 104835015 B CN104835015 B CN 104835015B CN 201510272608 A CN201510272608 A CN 201510272608A CN 104835015 B CN104835015 B CN 104835015B
- Authority
- CN
- China
- Prior art keywords
- task
- predecessor
- workflow
- log
- event log
- 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
Abstract
The invention discloses the Workflow mining methods based on predecessor task to be analyzed by task in analysis event log including predecessor task in the event log to workflow；It is input with event log, the Work flow model with Petri network description is output result；This method uses the event log based on predecessor task, and it includes the information of predecessor task that is, in event log that it is the input of current task that predecessor task, which refers to that current task executes the set for the task that the preceding needs relied on are completed,；The formal definitions of the event log of predecessor task are as follows: T is task-set, and T* is the task sequence comprising n task, and E=[θ] T is event set on the basis of task-set T；Predecessor task sequence is expressed as σ ∈ E*, and the event log of predecessor task is expressed as WE*.The present invention theoretically proposes novel method for digging, and all realizes actual tool on Activiti platform and ProM platform.
Description
Technical field
It is to use the invention belongs to the workflow mining technology in workflow technology field, especially workflow technology field
In the technology of the excacation stream process model from workflow logs.
Background technique
Workflow process is defined as file, information or activity according to a series of program or rule from a participant
It is transferred to the business procedure all or in part of another participant.Workflow system is one for managing workflow concentratedly
Automated system.Now, most of information systems describe task nexus using defined Work flow model and safeguard entire industry
Business process.But as business procedure is more and more, single business procedure becomes increasingly complex, Work flow model unavoidably has low
Effect, the problem of even error.It is necessary for being monitored and being improved to business procedure for this, and these demands are all
Need to obtain the real behavior of Work flow model.
Workflow mining technology aims to solve the problem that the above problem.Workflow mining technology passes through to tired during workflow execution
Long-pending mass data is effectively analyzed, and the operating condition of real scene personnel and workflow process are obtained, and is the work in later period
The monitoring and analysis for making flow model provide support, as shown in Fig. 2, the Practical significance of workflow mining technology is as shown in figure 2
Dotted box portion.Workflow mining technology exports corresponding workflow process model by analysis event log, reverse push.
The present invention only considers the case where event log information is complete and there is no noises, does not consider that event log information is possible incomplete
And the case where information errors.
Workflow mining is the technology of the anti-push model from event log (executing sequence), then with certain side
Relationship expression between task is come out (current, generally to describe entire Work flow model using Petri network) by formula.Therefore workflow is dug
Pick is a technical problem, and how the key of problem is from event log from anti-push model.Event log is event rail
The set of mark, every track are made of multiple events.Workflow mining technology can analyze event log and therefrom calculate and go out on missions
Between relationship, mainly causality, choice relation and concurrency relation, then further according to the anti-push model of these relationships,
Petri network is to realize a means of the technology here.
Currently, Workflow net is a kind of popular modeling method in Workflow Process Modeling field, Workflow net is one
The special Petri network of class.Petri network can describe sequence in process model, selection, circulation and concurrently and together with being apparent
Step structure has the advantages that in terms of describing process model: the semanteme of formalization, intuitive graphical representation, it can be readily appreciated that
Solid mathematical theory basis and mature analytical technology etc., therefore Petri network is comparative maturity and popular process model building work
Tool.For Petri network from structure, Petri network is a triple PN=(P, T, F), and wherein P is library institute (place) set,
T be transition (transition) set, and P ∩ T=φ, F=(P × T) ∪ (T × P) be library institute change between camber line collection
It closing, x={ y ∈ P ∪ T | (y, x) ∈ F } indicates preceding collection library institute or changed, x=y ∈ P ∪ T | (x, y) ∈ F }
Indicate a library or transition rear collection.
Workflow net is compared with ordinary Petri net, and there are two specific conditions: first is that there are two special in Workflow net
Library institute, be referred to as initial storehouse institute i and terminate library institute o, initial storehouse institute do not input, and terminates library and does not export；Second
Condition is that auxiliary transition a t*, extended model PN=(P, the T ∪ { t* }, F ∪ of composition are added between library institute o and library institute i
{ (o, t*), (t*, i) }) it is strong continune.Here, transition indicate the activity of workflow, library institute and the distribution of token expression work
Make the execution state flowed, the ignition condition of Petri network indicates movable execution condition, and generally, Workflow net can pass through
Petri network structure understands the logic of the business procedure of expression workflow.In Workflow net, transition (transition) represent work
Task in flowing, dependence between task by with library connection indicate that Tokken (token) is in set of library
The state of distribution situation expression process model.
Theoretical according to Petri network, the executable condition of the task (transition indicate) in Workflow net is the task institute
All respectively have a Tokken (token) in the front damming institute of corresponding transition, referred to as can ignition condition, be sometimes referred to as enabled condition
(enabled).The firing rule of one task (transition indicate) is: respectively moving from all input magazines institute for the transition lighted a fire
Except a Tokken, a Tokken is respectively added to all output libraries for the transition lighted a fire.It corresponds in Workflow system, one
The execution step of a task is: judging precondition, executes task, postcondition is arranged.Precondition refers to a task energy
The precondition enough executed, the i.e. executable condition of task, a task is only in the feelings for obtaining all executable conditions
Under condition, which could be executed.After the completion of postcondition refers to a task, before the task terminates, what which was done
Some rehabilitations, it may inform the end of whole process, it is also possible to precondition be arranged for its subsequent tasks.
Workflow engine is to execute work process flow by parsing " work process flow definition ".Work process flow uses
The mode of XML defines the relationship between activity and activity.As shown in figure 3, WorkflowProcess is used to define the process of workflow
Information, Activities are used for the active set of definitim flow.Each Activity element in Activities is for defining list
A activity, each activity are uniquely identified using Id.Transitions is used to define the transition process between activity,
Transition process between interior each Transition two activities of expression, the incipient activity of From attribute expression transition, and To table
Show the ending activity of transition.Therefore, by From the and To information in Transition, workflow engine is defined in analytic process
When with the execution of the task of decision, can recorde current task rely on all predecessor tasks for being over.
Workflow mining is the technology of the anti-push model from event log (executing sequence), if be reversely pushed out
Process model Workflow net description, then the essence of workflow mining is exactly event log (executing sequence) directional structure vectorical structure workflow
The technology of net, in Workflow net in triple structure PN=(P, T, F), wherein transition set directly (is held by workflow logs
Row sequence) in set of tasks composition, therefore excacation just becomes excavating set of library and set of library and transition collection therein
Between connection camber line, this anti-push technology needs analysis by task nexus (relation), existing α method, α+Side
Method, α++Method and β method are all based on this thinking design.
Currently, the Workflow mining method based on event log mainly has α method, α+Method, α++Method, β method and the side χ
Method.Wherein α method, α+Method and α++Event in method log is simple task names, and the event in β method log contains
There is the beginning and end information of task.α method can only handle the process model of SWF web frame constraint, cannot handle short circulation knot
Structure, implicit cause and effect dependency structure and implicit library institute structure；α+Method extends the mining ability of α method, it can excavate short follow
Ring structure；α++Method further expands the mining ability of α method, it can excavate most implicit cause and effect dependency structure；β
Method introduces new event type, it can excavate the process model for meeting the constraint of SWF web frame, short loop structure, but cannot
Handle implicit cause and effect dependency structure and implicit library institute structure.χ method is based on subsequent tasks method, and this method is from subsequent tasks
Relation information is extracted, has similar thinking direction with this patent method；But has that constitutionally is different with this patent method, rely on
Complete log is unable to the infull log of mined information.
Although Workflow mining method known to major part can all consider some events type, such as timestamp, operator
Member etc., but work on hand stream method for digging is all by the cause and effect between the task close proximity mining task in analysis event log
Relationship and concurrency relation, and then the choice relation between mining task.Although these methods can excavate a part of workflow mould
Type, but for being relied on and implicit library is difficult to excavate as implicit cause and effect, or even it cannot excavate.Above-mentioned based on event log
Workflow mining method in, α++The mining ability of method is strongest, although this method can excavate SWF structure, short circulation
Structure, most of implicit cause and effect dependency structure, but implicit library institute structure cannot be handled, and α++Method excavate implicit cause and effect according to
It needs when relying structure using complicated logic task relationship analysis, which greatly enhances the complexities of this method.χ method can dug
It digs in the range of structure and the performance of method for digging and has a clear superiority, but in the log range and log recording performance that can be excavated
The upper method proposed not as good as this patent.
Summary of the invention
The technical problem to be solved by the present invention is providing a kind of to comprising predecessor task, (current task relies on when executing
Task/the resource completed) information the workflow process method for digging that is excavated of predecessor task.This method can not only expand
Exhibition Workflow mining method excavates range, and can simplify cause and effect dependence in excacation flow model and it is potential simultaneously
Hair relationship, while the incomplete non-fully log of log information can be excavated (this patent is known as the complete log of forerunner).
The technical solution of the present invention is as follows: the Workflow mining method based on predecessor task, it is characterized in that passing through analysis event
Task in log is analyzed including predecessor task in the event log to workflow；It is input with event log, with Petri
The Work flow model of net description is output result；This method uses the event log based on predecessor task, and predecessor task, which refers to, to be worked as
The set for the task that the needs relied on before preceding task execution are completed is the input of current task, i.e., before including in event log
The information of drive task；The formal definitions of the event log of predecessor task are as follows: T is task-set, and T* is the task comprising n task
Sequence, E=[θ] T are event sets on the basis of task-set T；Predecessor task sequence is expressed as σ ∈ E*, the event of predecessor task
Log is expressed as WE*.
The method for digging overall flow is as shown in Figure 1, comprise the steps of (as shown in Figure 4):
(1) Work flow model based on Petri network description, initializes the return value N of the process, the knot according to Petri network
Structure definition, N is by set of library PW, task-set TWWith camber line collection FWIt constitutes；
(2) event log W is analyzed, task-set T is calculatedW, initial task TIWith the T that ends taskO；The work to be excavated is set
Make stream process model initial value N=(PW,TW,FW), wherein PW=TW=FW=φ；Event log W is analyzed, task-set T is calculatedW、
Initial task TIWith the T that ends taskO；
(3) single step circulation is extracted, Hash table HT is obtained；Preprocessing tasks collection TW；
(4) according to the relationship between task, task nexus collection X is calculatedW；
(5) task nexus collection X is removedWIn redundant elements, calculate final task set of relations YW；
(6) according to YW, calculate set of library PW；
(7) according to YWAnd PW, calculate camber line collection FW；
(8) by storage place single step circular insertion F in HTW；
(9) the workflow process model N of Petri network description is returned；
(10) according to process model N, show flow chart using tool.
Workflow mining method based on predecessor task, is the Workflow mining method of complete set, and method for digging includes
Above-mentioned 10 listed (in Fig. 1) step.This patent statement protection emphasis be step exclusive in the overall flow (3) and its
Sub-step, step (4) and its sub-step, step (8), comprising the following steps:
(3) single step circulation is extracted, is stored into HT, preprocessing tasks collection TW；
(4) according to the relationship between task, task nexus collection X is calculatedW；
(8) by the single step circular insertion in Hash table HT to FWIn；
Wherein step (3) specifically refines, comprising the following steps:
(3-1) constructs preprocessing tasks collection TWIn being formed in all tasks two-by-two of the task to St；
(3-2) defines Hash table HT, for storing single step circulation；
(3-3) traverses all tasks pair in St, finds out respectively by the task pair of some task a beginning and end, with HT
The mode of { (task to) }={ a } is inserted into HT, meanwhile, in TWMiddle exclusion task a.
Wherein step (4) specifically refines, comprising the following steps:
(4-1) is from task-set TWConstruct all task nexus collection XA；
(4-2) is using cause and effect dependence to XAIt is filtered, obtains XB；
(4-3) uses non-causal dependence, potential concurrency relation and loose potential choice relation, to XBIt is filtered,
Obtain XC；
(4-4) uses stringent choice relation, to XCIt is filtered, obtains final set of relations XW；
It is different about between task to define a series of and other methods for Workflow mining method based on predecessor task
Relationship.These relationships are used in the step in claim 1.These relationships include cause and effect dependence, non-causal dependence
Relationship, potential concurrency relation, non-causal dependence, loose potential choice relation and stringent potential choice relation.These relationships
Be specifically defined and be described as follows:
(A) cause and effect dependence (labeled as a →wB): under the conditions of [θ] b ∈ σ, σ ∈ W, a ∈ θ if it exists, then task b
Dependent on task a；
(B) non-causal dependence (is labeled as(a→wB)): the inverse relation of cause and effect dependence, i.e. task a and task
B be unsatisfactory for a →wb；
(C) potential concurrency relation (labeled as a//wB): task a and task b arbitrarily meet one of following two condition:
(C-1) under the conditions of [θ] t ∈ σ, there are a ∈ θ and b ∈ θ；Alternatively,
(C-2) in certain logged sequence σ=t1t2t3…tnIn, there are two log [θ1] a ∈ σ and [θ2] b ∈ σ, it is full
Sufficient θ1∩θ2=φ and b is adjacent to after a.
(D) concurrency relation non-potential (is labeled as(a//wB)): task a and task b be unsatisfactory for a//wb；
(E) loose potential choice relation (is labeled as a#LB): task a and task b is not present in a certain log sequence simultaneously
In column；
(F) stringent potential choice relation (is labeled as a#SB): task a and b meets following two condition simultaneously:
(F-1) task a and b meets loose alternative condition, i.e. a#LB, and
(F-2) task a and b is respectively present in two logged sequence σiAnd σjIn, and the intersection of the forerunner of two tasks is
Sky, i.e. θa∩θb=φ, that is to say, that two tasks do not have common predecessor task.
Workflow mining method based on predecessor task, can excavate non-fully log, thus with the other methods in the field
Compared to there is significant capability improving.The log range that this method can be excavated is named as the complete log of forerunner, and provides forerunner
The formal definitions of complete log.Its formal definitions is as follows:
(1) N=(P, T, F) is a reasonable SWF structure, that is,W is the workflow logs of structure N,
NamelyAnd each logged sequence σ ∈ W is that structure is terminated from initial state [i] start and ending state [o]
Igniting sequence；When W meets next (2), (3), (4) condition simultaneously；Claiming W is the complete log of forerunner；
(2) to any one task t ∈ N, all there is a logged sequence σ ∈ W and make in logged sequence that there are events
[θ]t∈σ；That is, each task occurs once at least in log；
(3) to any two task a and b, if in actual Work flow model N, task a and b be some library institute p it
Choice relation afterwards, then a and b centainly meet choice relation, that is, a# in workflow logs WS b。
(4) to any two task a and b, if task a and b are immediately some in actual Work flow model N
Task tiConcurrency relation later, then certainly existing logged sequence σ=t in workflow logs W1t2t3…tn∈W；So that
There are event [θ in logged sequence1]tk∈ σ and [θ2]tk+1∈ σ meets θ1∩θ2≠ φ and tk=a, tk+1=b；It is, appointing
Business a and b mono-, which is scheduled in some logged sequence, is adjacent to appearance once.
In the method, the effect of step (3) (including sub-step) is the log rule generated using single step circulation, will
The task of single step circulation proposes out, so that ground to be excavated log no longer contains single step circulation.Excavate entire workflow it
Afterwards, then by the task of single step circulation it is inserted into Petri network, to correctly excavate single step cycle task.
It this method the step of in (4), needs to pre-process the task nexus in log, calculates in log and own
Task nexus between task.Between task in log relationship preprocess method (as shown in Figure 5) include: cause and effect dependence, it is non-
Cause and effect dependence, potential concurrency relation, non-causal dependence, loose potential choice relation and stringent potential choice relation.
Content according to the present invention, we developed " the predecessor task equipments of recording based on Activiti platform " and
" the predecessor task digging tool based on ProM platform ".The former can use enterprise work of the Activiti as workflow engine
Make in Workflow Management System, in the form of predecessor task, records the execution journal of task in the system.The latter can be from the log of input
In information, process model is excavated, and by the process model in the form of Petri network graphic element, visually shows stream
Journey figure comes.(note: Activiti is the workflow engine of a open source, and official website ishttp://www.activiti.org/；
ProM is the open source software project in workflow mining field, and official website ishttp://www.processmining.org/)
The beneficial effects of the present invention are: the mining ability that this method not only improves Workflow mining method (can excavate
Implicit library institute structure out can excavate non-fully the namely complete log of forerunner), and simplify excavate cause and effect dependence and
The process of potential concurrency relation.Because implicit library does not influence the behavior of Work flow model, current all process model minings
Method is all not concerned with this special construction.But the redundancy relationship between going out on missions shown by implicit library, this has to a certain extent
Performance and security risk.This method has paid close attention to implicit library institute structure, it can excavate the implicit library institute structure of part, this can be with
Better support is provided for the analysis, verifying and monitoring of Work flow model；Simultaneously for it is complicated under production environment can not be complete
The non-fully log that the Workflow system of concurrent branch generates is covered, unique branch for other methods is also provided
It holds.
Detailed description of the invention
Fig. 1 is the flow chart of the Workflow mining method based on predecessor task.
Effect of Fig. 2 workflow mining technology in Workflow Management System.
Fig. 3 work process flow definitions example figure.
Fig. 4 is the main flow of the Workflow mining method based on predecessor task.
Fig. 5 relationship preprocess method between task.
Fig. 6 is academic dissertation management process example.
Fig. 7 is the Work flow model that present example can excavate, which includes implicit library institute.
Fig. 8 is the log information according to example, and development and usage tool shows flow chart.
Fig. 9 is the system assumption diagram of the predecessor task equipments of recording based on Activiti platform.
Figure 10 is the System Design figure of the predecessor task digging tool based on ProM platform.
Specific embodiment
In order to make the objectives, technical solutions and advantages of the present invention clearer, With reference to embodiment and join
According to attached drawing, the present invention is described in further details.These descriptions are merely illustrative, and is not intended to limit of the invention be applicable in
Range.In addition, in the following description, the description to already present structure and technology is omitted, to avoid this is unnecessarily obscured
The concept of invention.
The present invention is mainly using new event type and by all between task in relationship pretreatment acquisition log
Relationship between business, and it is added on the basis of α method the amendment step to task nexus collection.The method for digging overall flow is such as
Shown in Fig. 1.Its specific implementation is as follows:
1, the main flow of this method is as shown in Fig. 4 top half.
(1) step 1 initializes the return value N (Work flow model of Petri network description) of the process, according to Petri network
Structure definition, N is by set of library PW, task-set TWWith camber line collection FWIt constitutes；
(2) step 2, analysis event log calculate task-set TW(all titles included in log are different to appoint
Business), the initial task collection T of each execution track σICollect T with ending taskO；
(3) step 3 extracts single step circulation, is stored into Hash table HT；Preprocessing tasks collection TW；
(4) step 4 calculates task nexus collection X according to the relationship between taskW；
(5) step 5, by deleting X'WThe element of middle redundancy calculates final task nexus collection YW。
(6) step 6 calculates the set of library P of Work flow modelW, element YWIn element, initial storehouse institute and end
Library set；
(7) step 7, according to PWAnd YWObtain the transition arc collection F of Work flow modelW；
(8) step 8, by storage place single step circular insertion F in HTW；
(9) step 9 returns to Work flow model N；
(10) final step shows flow chart using tool according to process model N.
Wherein, it is as follows to extract the step of single step circulation for the step 3 of this method:
(3-1) step 1 constructs task-set TWIn all tasks the forming two-by-two of the task to St；
(3-2) step 2 defines Hash table HT, for storing single step circulation.
(3-3) step 3 traverses all tasks pair in St, finds out respectively by the task of some task a beginning and end
It is right, it is inserted into HT in a manner of HT { (task to) }={ a }, meanwhile, in TWMiddle exclusion task a.
Wherein, the step 4 of this method, calculating task set of relations XWProcess.It is closed between task described in the process application drawing 5
It is preprocess method, the specific steps are as follows:
(4-1) step 1, from task-set TWConstruct all task nexus collection XA。
(4-2) step 2, using cause and effect dependence to XAIt is filtered, obtains XB。
(4-3) step 3, using non-causal dependence, potential concurrency relation and loose potential choice relation, to XBIt carries out
Filtering, obtains XC。
(4-4) step 4, using stringent choice relation, to XCIt is filtered, obtains final set of relations XW。
Illustrate implementation of the invention below by specific example.
Example of the invention will excavate the Work flow model of Fig. 6 from event log, and the model is by 11 library institutes, 12
Transition are constituted.The Work flow model describes the management process to graduate's academic dissertation, be mainly concerned with the inside and outside evaluation of paper,
Thesis defense, paper examination.For convenience of analysis, it would be desirable to the Chinese of example are mapped to mathematic sign, such as
Shown in table 1.Table 2 is the event log of academic dissertation management process example, and the event log is by the input as present example
Data.
1 mission number of table and its Chinese table of comparisons
The event log of 2 academic dissertation management process example of table
For the example, we will implement this method using following steps:
1. initializing return value N, (Work flow model of Petri network description, the structure according to Petri network define, and N is by library institute
Collect PW, task-set TWWith camber line collection FWConstitute) so that PW=TW=FW=φ.
2. obtaining event task-set T from event logW={ t1,t2,t3,t4,t5,t6,t7,t8,t9,t10,t11,t12,
Obtain initial task collection TI={ t1And end task and collect TO={ t12}。
3. extracting single step circulation, it is stored into Hash table HT, preprocessing tasks collection TW, the specific steps are as follows:
(1) task that all tasks form two-by-two in construction task-set is to St={ (t1,t2),(t1,t3),(t1,t4),…,
(t11,t12)}。
(2) sky Hash table HT={ } is defined.
(3) all tasks pair in St are traversed, are found out respectively by the task pair of some task beginning and end.It here can be with
Find t4And t7, so after the end of the step, HT={ { t3,t5}=> { t4},{t3,t6}=> { t7 } }, TW={ t1,t2,t3,t5,
t6,t8,t9,t10,t11,t12}
4. calculating task nexus collection X according to the relationship between taskW, the specific steps are as follows:
(1) from task-set TWConstruct all task nexus collection:
XA={ ({ t1},{t2}),({t1},{t3}),…,({t1,t2},{t3}),…,({t11},{t12})}。
(2) step 2, using cause and effect dependence to XAIt is filtered, obtains:
XB={ ({ t1},{t2}),({t3},{t5}),…,({t1,t2},{t3,t5}),…,({t11},{t12})}。
(3) step 3, using non-causal dependence, potential concurrency relation and loose potential choice relation, to XBIt carried out
Filter, obtains:
XC={ ({ t1},{t2}),({t11},{t2}),({t3,t8},{t5})…,({t1,t2},{t3}),…,({t11},
{t12})}.
(4) step 4, using stringent choice relation, to XCIt is filtered, obtains final set of relations XW={ ({ t1},
{t2}),({t11},{t2}),({t1,t11},{t2}),({t2},{t3}),({t3},{t5}),({t3},({t5},{t8}),
({t6},{t8}),({t8},{t9}),({t9},{t11}),({t9},{t10}),({t9},{t10,t11}),({t10},{t12})}
5. deleting amendment task nexus collection X' according to the step 6 of main flowWThe element of middle redundancy and obtain final task pass
Assembly YW, the final task set of relations are as follows: { ({ t1,t11},{t2}),({t2},{t3}),({t3},{t5}),({t3},{t6}),
({t5},{t8}),({t6},{t8}),({t8},{t9}),({t9},{t10,t11}),({t10},{t12})}。
6. according to the step 7 of method and final task collection YW, this method can obtain set of library PW, the set of library are as follows: { iw,
ow,p({t1,t11},{t2}),p({t2},{t3}),p({t3},{t5}),p({t3},{t6}),p({t5},{t8}),p({t6},{t8}),p({t8},{t9}),
p({ t9 }, { t10, t11 }),p({t10},{t12})}.Wherein iwAnd owRespectively initial storehouse institute and end library institute.
7. according to the step 8 of main flow and applying set of library PWWith task-set YW, this method acquisition camber line collection FW, the camber line
Collection are as follows: { (iw,t1),(t1,p({t1,t11},{t2})),(p({t1,t11},{t2}),t2),…,(t12,ow)}。
8. by single step circular insertion to FWIn, obtain complete camber line collection are as follows: { { (iw,t1),(t1,p({t1,t11},{t2})),
(p({t1,t11},{t2}),t2),(t4,p({t3},{t5})),(p({t3},{t5}),t4),(t7,p({t3},{t6})),(p({t3},{t6}),t7)…,
(t12,ow)}}。
9. so far, this method just completely obtains the Work flow model N=(P described by Petri networkW,TW,FW)。
10. showing flow chart using tool, as shown in Figure 8 according to process model N.
Above step obtains Work flow model N, can get as represented by Fig. 6 by Petri network graphical representation tool
Work flow model.Although the model contains SWF structure, short loop structure, even implicit cause and effect dependency structure, this method
It can correctly excavate, i.e., this method can correctly excavate SWF structure, short loop structure and implicit cause and effect dependency structure.Certainly, should
Method can also excavate Work flow model as shown in Figure 7, which contains implicit library institute structure P1。
" Workflow mining method based on predecessor task " is embodied as the forerunner based on Activiti platform by us
Task record tool and predecessor task digging tool based on ProM platform.Wherein Activiti platform is one and is widely used
Workflow engine into Enterprise Workflow Management System, by the log recording for realizing this patent the method on the platform
This patent method can be applied in actual enterprise work flow management by tool, be provided with production valence for the method for this patent
The input of value.And ProM platform is the frame that a workflow process being widely used by industry and academia is excavated and studied
Frame has many tools for being used to carry out workflow process analysis and research thereon.By realizing this patent on the platform
The digging tool of method is stated, can specifically implement this patent the method, and visually show this patent the method
Output.
Predecessor task equipments of recording based on Activiti platform are applied in academic paper management system (its core by us
Heart Work flow model is as shown in Figure 6), the predecessor task (as shown in table 2) of the available workflow, the log can be with text text
The form of shelves is stored on disk.Then, the predecessor task digging tool based on ProM platform is used to the predecessor task, it can be with
It excavates corresponding to the log as a result, to run screenshot as shown in Figure 8, it can be seen that the result and Fig. 6 excavated are complete
It is identical.
The part of specific implementation as the Workflow mining method based on predecessor task, before Activiti platform
It drives task record tool and the predecessor task digging tool based on ProM is the enhancing and supplement to this patent.Therefore, pass through Fig. 9
And Figure 10, the architecture Design of two tools is supplemented, respectively to reinforce applying for this patent the explanation of the right of protection.
It should be intended that, above-mentioned specific embodiment of the invention is used only for exemplary illustration or explains of the invention
Principle, but not to limit the present invention.Therefore, that is done without departing from the spirit and scope of the present invention is any
Modification, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.In addition, appended claims purport of the present invention
Covering the whole variations fallen into attached claim scope and boundary or this range and the equivalent form on boundary and is repairing
Change example.
Claims (4)
1. the Workflow mining method based on predecessor task, it is characterized in that by task in analysis event log, including to work
Predecessor task is analyzed in the event log of stream；It is input with event log, the Work flow model with Petri network description is defeated
Result out；This method uses the event log based on predecessor task, and predecessor task refers to that current task executes the preceding needs relied on
The set of the task of completion, is the input of current task, i.e., includes the information of predecessor task in event log；Predecessor task
The formal definitions of event log are as follows: T is task-set, and T* is the task sequence for including n task；E=[θ] T is in task-set T
On the basis of event set；Predecessor task sequence is expressed as σ ∈ E*, and the event log of predecessor task is expressed as WE*；
Overall flow comprises the steps of:
(1) Work flow model based on Petri network description, initializes the return value N of the process, and the structure according to Petri network is fixed
Justice, N is by set of library PW, task-set TWWith camber line collection FWIt constitutes；
(2) event log W is analyzed, task-set T is calculatedW, initial task TIWith the T that ends taskO；The workflow to be excavated is set
Process model initial value N=(PW,TW,FW), wherein PW=TW=FW=φ；
(3) single step circulation is extracted, Hash table HT is obtained；Preprocessing tasks collection TW；
(4) according to the relationship between task, task nexus collection X is calculatedW；
(5) task nexus collection X is removedWIn redundant elements, calculate final task set of relations YW；
(6) according to YW, calculate set of library PW；
(7) according to YWAnd PW, calculate camber line collection FW；
(8) by storage place single step circular insertion F in HTW；
(9) the workflow process model N of Petri network description is returned；
(10) according to process model N, show flow chart using tool.
2. the Workflow mining method according to claim 1 based on predecessor task,
It is characterized in that step (3) specifically includes the following steps:
(3-1) constructs preprocessing tasks collection TWIn being formed in all tasks two-by-two of the task to St；
(3-2) defines Hash table HT, for storing single step circulation；
(3-3) traverses all tasks pair in St, finds out respectively by the task pair of some task a beginning and end, { (is appointed with HT
Be engaged in)=the mode of { a } is inserted into HT, meanwhile, in TWMiddle exclusion task a；
Wherein step (4) specifically refines, comprising the following steps:
(4-1) is from task-set TWConstruct all task nexus collection XA；
(4-2) is using cause and effect dependence to XAIt is filtered, obtains XB；
(4-3) uses non-causal dependence, potential concurrency relation and loose potential choice relation, to XBIt is filtered, obtains
XC；
(4-4) uses stringent choice relation, to XCIt is filtered, obtains final set of relations XW。
3. a series of Workflow mining method according to claim 1 based on predecessor task, it is characterized in that defining passes
Relationship between task；These relationships include cause and effect dependence, non-causal dependence, potential concurrency relation, non-causal
Dependence, loose potential choice relation and stringent potential choice relation；Being specifically defined for these relationships is described as follows:
(A) cause and effect dependence, labeled as a →wB: under the conditions of [θ] b ∈ σ, σ ∈ W if it exists, a ∈ θ, then task b is depended on
Task a；
(B) non-causal dependence, be labeled asThe inverse relation of cause and effect dependence, i.e. task a and task b are discontented
Sufficient a →wb；
(C) potential concurrency relation, labeled as a//wB: task a and task b arbitrarily meets one of following two condition:
(C-1) under the conditions of [θ] t ∈ σ, there are a ∈ θ and b ∈ θ；Alternatively,
(C-2) in certain logged sequence σ=t1t2t3…tnIn, there are two log [θ1] a ∈ σ and [θ2] b ∈ σ, meet θ1∩
θ2=φ and b is adjacent to after a；
(D) concurrency relation non-potential, be labeled asTask a and task b be unsatisfactory for a//wb；
(E) loose potential choice relation, be labeled as a#LB: task a and task b is not present in simultaneously in a certain logged sequence；
(F) stringent potential choice relation, be labeled as a#SB: task a and b meets following two condition simultaneously:
(F-1) task a and b meets loose alternative condition, i.e. a#LB, and
(F-2) task a and b is respectively present in two logged sequence σiAnd σjIn, and the intersection of the forerunner of two tasks is sky,
That is θa∩θb=φ, that is to say, that two tasks do not have common predecessor task.
4. the Workflow mining method according to claim 1 based on predecessor task, it is characterized in that the log range excavated
It is named as the complete log of forerunner, and provides the formal definitions of the complete log of forerunner；Its formal definitions is as follows:
(1) N=(P, T, F) is a reasonable SWF structure, that is,W is the workflow logs of structure N, also
It isAnd each logged sequence σ ∈ W is the point that structure is terminated from initial state [i] start and ending state [o]
Fiery sequence；When W meets next (2), (3), (4) condition simultaneously；Claiming W is the complete log of forerunner；
(2) to any one task t ∈ N, all there is a logged sequence σ ∈ W and make in logged sequence that there are event [θ] t
∈σ；That is, each task occurs once at least in log；
(3) to any two task a and b, if in actual Work flow model N, after task a and b are some library institute p
Choice relation, then a and b centainly meet stringent potential choice relation, that is, a# in workflow logs WSb；
(4) to any two task a and b, if task a and b are immediately some task t in actual Work flow model Ni
Concurrency relation later, then certainly existing logged sequence σ=t in workflow logs W1t2...tn∈W；So that in log
There are event [θ in sequence1]tk∈ σ and [θ2]tk+1∈ σ meets θ1∩θ2≠ φ and tk=a, tk+1=b；It is, task a and b
One is scheduled in some logged sequence that be adjacent to appearance primary.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510272608.4A CN104835015B (en) | 2015-05-25 | 2015-05-25 | Workflow mining method based on predecessor task |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510272608.4A CN104835015B (en) | 2015-05-25 | 2015-05-25 | Workflow mining method based on predecessor task |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104835015A CN104835015A (en) | 2015-08-12 |
| CN104835015B true CN104835015B (en) | 2019-01-22 |
Family
ID=53812889
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510272608.4A Active CN104835015B (en) | 2015-05-25 | 2015-05-25 | Workflow mining method based on predecessor task |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104835015B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105260188A (en) * | 2015-10-24 | 2016-01-20 | 北京航空航天大学 | Time characteristic model and modeling method thereof |
| CN106779594A (en) * | 2016-12-01 | 2017-05-31 | 江苏鸿信系统集成有限公司 | A kind of Workflow management method based on Activiti |
| CN108647253B (en) * | 2018-04-23 | 2022-09-06 | 南京理工大学 | Mining algorithm containing time constraint workflow |
| CN108710645B (en) * | 2018-04-23 | 2021-09-10 | 南京理工大学 | Process mining method based on mixed event log |
| CN108717625B (en) * | 2018-05-28 | 2022-05-20 | 北京交通大学 | Generation method of railway electric service workflow |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102332125A (en) * | 2011-11-08 | 2012-01-25 | 南京大学 | Workflow mining method based on subsequent tasks |
-
2015
- 2015-05-25 CN CN201510272608.4A patent/CN104835015B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102332125A (en) * | 2011-11-08 | 2012-01-25 | 南京大学 | Workflow mining method based on subsequent tasks |
Non-Patent Citations (3)
| Title |
|---|
| 一种基于后继任务的过程挖掘算法;王栋毅 等;《计算机应用与软件》;20121015;第29卷(第10期);17-21 |
| 基于事件日志的工作流挖掘算法研究;梁艳;《中国优秀硕士学位论文全文数据库信息科技辑》;20111215(第S1期);I138-479 |
| 基于工作流网的过程挖掘算法研究;闻立杰;《中国博士学位论文全文数据库信息科技辑》;20080815(第08期);I138-1 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104835015A (en) | 2015-08-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104835015B (en) | Workflow mining method based on predecessor task | |
| CN102332125B (en) | Workflow mining method based on subsequent tasks | |
| Giaglis | A taxonomy of business process modeling and information systems modeling techniques | |
| Schimm | Mining exact models of concurrent workflows | |
| Baier et al. | Bridging abstraction layers in process mining by automated matching of events and activities | |
| Hegedüs et al. | Back-annotation of simulation traces with change-driven model transformations | |
| Jander et al. | Goal-oriented processes with GPMN | |
| Kožíšek et al. | Business process modelling languages | |
| Popova et al. | From petri nets to guard-stage-milestone models | |
| Metzker et al. | An interdisciplinary approach for successfully integrating human-centered design methods into development processes practiced by industrial software development organizations | |
| Kim et al. | A comparison of software product line traceability approaches from end-to-end traceability perspectives | |
| Jugel et al. | Providing EA decision support for stakeholders by automated analyses | |
| Zemni et al. | An automated approach for merging business process fragments | |
| Herzog et al. | A 4‐Box Development Model for Complex Systems Engineering | |
| Fernández-Ropero et al. | Graph-Based Business Process Model Refactoring. | |
| Fraternali et al. | Multi-level tests for model driven web applications | |
| Lautenbacher et al. | Planning support for enterprise changes | |
| Berriche et al. | Combining case-based reasoning and process mining to improve collaborative decision-making in products design | |
| Shershakov | DPMine: modeling and process mining tool | |
| Genga et al. | Subgraph mining for anomalous pattern discovery in event logs | |
| Sahlabadi et al. | Evaluation and extracting factual software architecture of distributed system by process mining techniques | |
| Miao et al. | A formal specification-based integration testing approach | |
| Kirchner et al. | Using CUTA4BPM to support participative development of expert-driven processes | |
| El Hichami et al. | An algebraic method for analysing control flow of BPMN models | |
| Ungureanu et al. | Civil Systems Engineering: Definition and Basic Concepts |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |