CN110109921A - Event log and process model calibration method based on event similarity - Google Patents
Event log and process model calibration method based on event similarity Download PDFInfo
- Publication number
- CN110109921A CN110109921A CN201910250146.4A CN201910250146A CN110109921A CN 110109921 A CN110109921 A CN 110109921A CN 201910250146 A CN201910250146 A CN 201910250146A CN 110109921 A CN110109921 A CN 110109921A
- Authority
- CN
- China
- Prior art keywords
- event
- similarity
- concept
- attribute
- mark
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/22—Indexing; Data structures therefor; Storage structures
- G06F16/2228—Indexing structures
- G06F16/2246—Trees, e.g. B+trees
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention belongs to information technology field more particularly to a kind of event log based on event similarity and model calibration methods.This method is first defined Event Concepts;Secondly, calculating the similarity between each attribute of two events based on ontology tree;Finally, the similarity based on event calibrates mark and model.The present invention is from the event with multiple attributes, the calibration problem of research process model and event log.For the consistency for ensuring event description in log and model, event is indicated with its functional semantics, and is marked on domain body.Similarity by calculating the multiple attributes of event calculates the similarity between the event with same campaign.Similarity based on event provides the calibration algorithm between process model and event log.Compared with for the simple alignment of single attribute event, calibration of the present invention is the finer description to calibration, can provide more detailed foundation for the problems in improved model or discovery log.
Description
Technical field
The invention belongs to information technology field more particularly to a kind of event logs and process model based on event similarity
Calibration method.
Background technique
The method detected currently used for the consistency to event log and process model has very much.For determine log with
The degree of model bias, Adriansyah A etc. propose the concept of optimal calibration, the Solve problems of optimal calibration are converted into asking
Solve optimum path problems.Wherein, the path with minimum cost corresponds to log and the optimal calibration of model.Meanwhile according to being
The no importance for distinguishing each event, optimal calibration are further divided into the optimal calibration based on cost value (cost) and are based on moving
Two kinds of situations of optimal calibration of dynamic step number (move).For the calibration problem of Enterprise Data and model, Song etc. is according to model
Structure and behavior feature, reduce search space on a large scale using effective heuristic, to improve the efficiency of calibration.
The calibration research for adapting to distributed computing and big data environment also continuously emerges.Jorge M G etc. is by SESE structure
Process model be divided into multiple submodels and calibrate respectively, the time required to it can substantially reduce calibration, while being also easy to carry out
The diagnosis of deviation.The decomposition method of the application Petri network such as Verbeek, by the process model indicated with Petri network and its log it
Between calibration problem, be converted into the subnet of Petri network compared between log mapping, propose feasibility that calibration decomposes,
A series of theorems such as the cost lower limit of puppet calibration.
In the calibration currently about model and event log, each event (event) pertains only to one category of activity
Property.But the event log information in the actual environment, recorded often relates to executor, operation object and the event hair of event
The information in raw time, place etc., i.e. event may have multiple attributes, be not limited solely to a movable attribute.Meanwhile if
When in view of other attributes other than the activity of event, it is currently used in the optimal calibration of reflection model and log deviation, it may not be still
It is so optimal calibration, it is therefore desirable to which new method carrys out the deviation of computation model and log.
Summary of the invention
To solve the above problems, multiple attribute synthesis that event is related to by the present invention take in, a kind of base is provided
In the event log and process model calibration method of event similarity, it can preferably reflect the symbol between event log and process model
Conjunction property.
The technical solution adopted by the present invention are as follows: a kind of event log based on event similarity and process model calibration side
Method, comprising the following steps:
Firstly, being defined to Event Concepts, reflected including process model, the mark of event, event log and activity
It penetrates;
Secondly, calculating the similarity between each attribute of two events based on ontology tree;Specific calculating process are as follows:
Step 1. inputs the event e in markσWith the event e in process modelN, the activity attributes of two events need identical;
Step 2. is by two events its functional semantics FS (eσ) and FS (eN) form indicate, then will be in its functional semantics
Each attribute be labeled on ontology tree respectively and calculate its similarity;
Step 3. further calculates the similarity of two events according to the similarity of each attribute of two events;
Finally, based on the similarity of event in event log mark and model calibrate;Specific calculating process is as follows:
Step 1. reference standard likelihood function defines similarity likelihood function and its cost value;
Step 2. is based on similarity likelihood function, defines the cost value cost (γ) of calibration;
The mark σ of event is converted to event net by step 3., has same campaign attribute in calculating process model and event net
Event synchronizing moving similarity, and the product net of tectonic event net and process model;
Step 4. finds out the different paths in product net from beginning state to final state, and the transition in product net are mapped as
Movement in calibration;
For step 5. according to the definition for calibrating cost value under similarity likelihood function, the smallest path of cost value is optimal school
Quasi- γ.
It is defined as follows in the first step:
It defines 1. process models: process model being expressed as a binary group (PN, Attr), wherein PN=(P, T;F,
It M), is a Petri network, wherein P is that a set of library is closed, and T is a transition set,It is that stream closes
The set of system, M:P → Num.M ∈ NumPIt is a mark function, wherein Num indicates natural number set, M ∈ NumPIndicate M:P
→ Num is a function being defined on set P;Attr is a label function, and has Attr:T → ξ ∪ τ, wherein ξ is institute
The event sets being likely to occur, τ are null event, and Attr function is that the transition of each of Petri network are mapped as event set ξ
Or the element in null event τ, event e are an elements in set ξ, be expressed as (e, #res (e) [: value]), whereineFor
Event classifier indicates with the activity name of event, #res (e) [: value] it indicateseRespective attributes and attribute value, wherein
[: value] expression attribute value is option;
Define the mark of 2. events: when an event has multiple attributes, trace description is the finite sequence of an event, i.e.,
<(e 1,#res(e1)[:value]),…,(e n,#res(en) [: value]) >, event log is the multiset of mark;
Define 3. active maps: the method that the event with multiple attributes is indicated with its activity attributes, it is referred to as living
Dynamic mapping, is denoted as↓act。
Ontology tree is constructed referring to WordNet and HowNet in the case where domain expert participates in, is used when to construction ontology tree general
Thought is defined:
Define the functional semantics of 4. event e
The functional semantics of event e are described as multi-component system: FS (e)=< activity, resource, [option],
Constraint >, in which:
Activity=C | C ∈ CaIndicate activity corresponding to event e, wherein CaIt is movable general to indicate that certain field describes
Set is read, #act (e) is denoted as;
Resource=C | C ∈ CrIndicate resource involved by event e, wherein CrIndicate that certain field describes resource
Concept set, is denoted as #res (e);
Option=C | C ∈ CoIndicate other attributes involved by event e, wherein CoIndicate that the description of certain field is corresponding
The concept set of other attributes;
Constraint=Q1∧Q2∧…∧Qn, indicate each attribute constraint condition for occurring to meet when some event,
Middle QiThe specific constraint condition of (i=1,2 ..., n) expression a certain attribute of event;Constraint condition Q1, Q2..., QnBetween be defined as
" and " relationship;
Define 5. ontology trees
Enabling O is the ontology tree on the D of field, then has:
O=< ({ C }, { R }) | Ci∈ D, i=1,2 ..., m;Rj∈ D, j=1,2 ..., n >;
Wherein, C is concept set, indicates the related notion on the D of field, and can be divided into class concepts and Feature concept, reality
Example concept and Factors ' Concept;Wherein class concepts characterization has the set of same nature object;Feature concept reflects concept C institute
The feature having;Instance concepts illustrate the respective instance of concept;Factors ' Concept characterizes institute's element of concept;R is to close
The relationship of the set of system, ontology definition has: is-a, instance-of, element-of and trait-of, wherein is-a is indicated
Inheritance between concept, instance-of indicate the relationship between concept and concept specific example, part-of refer to concept with
Relationship between the element of concept, trait-of indicate the relationship between concept and the feature of concept;
Define 6. Ontologies mark
It is the process of the concept in ontology tree, the referred to as Ontology of concept by given concept k replacement or partial replacement
Mark, is denoted as remark (k);
Define the Ontology mark of 7. events
The functional semantics FS (e) of event e is subjected to ontology mark substantially by each attribute of event in ontology tree
Carry out the process of semantic tagger;Firstly, the concept #attr (e) for being related to event e carries out Ontology mark in ontology tree
Note, if some attribute of event e has specific attribute value, provides its occurrence according to the trait-of marked.
Based on event attribute the step of the mark on ontology tree seeks similarity are as follows:
If eNFor an event in model, eσIt is an event in mark, and hase N=e σ, two events are expressed as it
Functional semantics FS (eN) and FS (eσ) form, and by FS (eN) and FS (eσ) in each attribute be labeled on ontology tree;
If eNAnd eσMark of a certain attribute in ontology tree is expressed as x and y, then concept between (x, y) similarity sim (x,
Y) following situation can be divided to discuss:
(1) y ∈ (C if it does not existr(eσ)∪Co(eσ)), so that kind (x)=kind (y), then sim (x, y)=0;
(2) y ∈ (Cr (e if it existsσ)∪Co(eσ)), so that kind (x)=kind (y), and
If y ∈ descendant (x) or y ∈ instance-of (x), then sim (x, y)=1;
IfAndBetween then sim (x, y)=Dis (x, y), i.e. x, y
Similarity sim (x, y) is its distance in ontology tree;
(3) y ∈ (C if it existsr(eσ)∪Co(eσ)), so that kind (x)=kind (y), and
Each constraint Q of x and y will be limitedi, it is mapped as its feature trait-ofi(x) and trait-ofi(y):
If (a)Then sim (Qi(xi,yi))=Dis (x, y);
If (b)And trait-ofi(y) it is unsatisfactory for trait-ofi(x), then sim (Qi(xi,yi))=
0;
If (c)Then trai-tofi(y) meet trait-ofi(x)
If y ∈ descendant (x) or y ∈ instance-of (x), sim (Qi(xi,yi))=1;
IfAndThen sim (Qi(xi,yi))=Dis (x, y);
Then, sim (x, y)=average (sim (Qi(xi,yi));
In the above calculating process, function kind indicates the affiliated type of event attribute;If there is kind (x)=kind (y), andsim(Qi(xi,yi)) it is in QiRestriction under xi,yiSimilarity, and sim (x, y) be equal in various different QsiUnder
sim(Qi(xi,yi)) average value, wherein trait-of is one of relationship present in ontology tree.
Reference standard likelihood function enables the cost value of log movement, model mobile, weak synchronizing moving and strong synchronizing moving
Cost is denoted as (1,1,1,1-Sim-event (e respectivelyσ,eN)), it is called similarity likelihood function, wherein Sim-event (eσ,
eN) indicate event to (eσ,eN) between similarity, be the average value of above-mentioned given each attributes similarity of event;Based on phase
Like degree likelihood function, calibrating cost value cost (γ) is to calibrate the mobile number of the log for including, the mobile number of model, weak synchronous shifting in γ
Dynamic several and strong synchronizing moving number * (1-Sim-event (eσ,eNThe sum of)).
The present invention is from the event with multiple attributes, the calibration problem of research process model and event log.It is true
The consistency for protecting event description in log and model, event and its attribute is indicated with its functional semantics, and on domain body
It is marked.Have same campaign attribute event to it by calculating the similarity of multiple attributes between event pair, and then calculating
Between similarity.Similarity based on event provides the calibration algorithm between process model and event log.
Through the invention it is described calibration and simple alignment comparison it is found that simple alignment reflection be deviation matter, and this
The invention calibration reflects the amount of deviation on the basis of matter.Therefore, calibration of the present invention to the description of calibration more
Refinement causes, and can provide more detailed foundation for the problems in improved model or discovery log.
Detailed description of the invention
Fig. 1 is the process model N of the training course described with Petri network1;
Fig. 2 is the partial function ontology schematic diagram of teaching field;
Fig. 3 is process model schematic diagram described in embodiment;
Fig. 4 is event net schematic diagram described in embodiment;
Fig. 5 is the product net schematic diagram of process model described in embodiment Yu event net;
Fig. 6 is that simple alignment and optimal calibration of the present invention are compared, wherein (a) is simple optimal calibration, (b) is this
The invention optimal calibration;
Fig. 7 is that the structure of simple alignment and calibration of the present invention compares, wherein (a) is the structure chart of simple alignment, (b)
It is the structure chart of calibration of the present invention.
Specific embodiment
The present invention is described in detail with reference to the accompanying drawings and examples.
Event log and process model calibration method based on event similarity, comprising the following steps:
Firstly, concept involved in event calibration process is defined and is explained.
Define 1. process models
Process model is expressed as a binary group (PN, Attr), wherein
(1) PN=(P, T;F, M), it is a Petri network, wherein and P is that a set of library is closed, and T is a transition set,It is the set of flow relation, M:P → Num.M ∈ NumPIt is a mark function, wherein Num is indicated
Natural number set, M ∈ NumPExpression M:P → Num is a function being defined on set P.
(2) Attr is a label function, and has Attr:T → ξ ∪ τ, wherein ξ is be likely to occur event set
It closes, τ is null event.Each of Petri network is changed the member being mapped as in event set ξ or null event τ by label function Attr
Element.Event e is an element in ξ, be expressed as (e, #res (e) [: value]), whereineFor event classifier, with event
Activity name indicates, #res (e) [: value] it indicateseRespective attributes and attribute value, wherein [: value] indicates attribute value
For option.
Define the mark of 2. events
When an event has multiple attributes, trace description is the finite sequence of an event, i.e., < (e 1,#res(e1)[:
value]),…,(e n,#res(en)[:value])>.Event log is the multiset of mark.
Define 3. active maps
The method that event with multiple attributes is indicated with its activity attributes, referred to as active map, are denoted as↓act。
Similarly, the mark of event can be mapped as its corresponding simple mark by active map, and it is opposite that event log mapping is also known as it
Process model is also mapped as a simple procedure model by the simple event log answered.
Unless otherwise specified, the event in the present invention refers both to the event of multiple attributes.Process model refers to more as a result,
Attribute event model, mark refer to that the mark of more attribute events, event log refer to more attribute event logs.And by only activity belongs in event
Model, event log, mark and the calibration of property are referred to as naive model, simple event log, simple mark and simple alignment.
The concepts such as the mark for illustrating process model and event below by an example.
Fig. 1 describes the process model N an of training course with Petri network1.Its main activities is listed below: 1. student
Correlation circumstance is seeked advice from Xiang assistant (assistant) and submits application (apply request);2. registering personal information
(register);3. assistant verifies student's personal information (check information);4. student pays training cost (pay
financial staff);5. assistant (assistants) prepares correlated curriculum data (prepare);6. Sam (teacher) exists
Classroom B201 carries out training;7. student previews next class journey (preview);8. being examined (examine) to training result
Or examination (assess).Attribute and attribute value of the event in addition to activity are indicated in figure with dashed box, are the limitations to event.Scheming
The attributes such as each event in 1, including activity, resource resource and place place.
If there is mark σ1And σ2, wherein σ1=< (apply request, assistant:Cindy), (register,
assistant:Jack),(pay,financial staff),(prepare,assistant),(teach course,
teacher:Sam,classroom:B201),(preview,student),(teach course,teacher:Sam,
classroom:B201),(examine,teacher:Sam)>.And mark σ2Except teach course occur in addition to B202, other
With mark σ1It is identical.
To mark σ1And σ2Active map is carried out, σ is obtained1↓act=σ2↓act=< apply request, register, pay,
prepare,teach course,preview,teach course,examine>.That is σ1And σ2In N1Naive model on etc.
Valence, and be 1 with the fitness degree (fitness) of model.But when the multiple attributes for the event of considering, by mark σ2In N1Upper progress
It recurs, then its fitness degree is less than 1.
Secondly, calculating the similarity between each attribute of two events based on ontology tree.
Similarity between attribute can be calculated using ontology tree.And current ontology tree is still incomplete, therefore utilize
WordNet and HowNet etc. is using concept representated by morphology as description object, to disclose between different concepts and concept category
The lexical semantic net of relationship, the similarity calculation of Lai Jinhang attribute between sexual intercourse.In semantic net, the tool calibrated as needed
The relationship that body Model and specific event log select it to indicate.
Building method and the semantic application in Services Composition referring to WordNet, it is believed that a process model
It is a series of program that functions are realized towards specific area.Each single item function can be completed by an event, and an event
It can be indicated again by multiple attributes.The functional semantics of defined herein event describe, and embody the common attribute of event in description, such as
Activity attributes and Resource Properties, other attributes are as option.
Define the functional semantics of 4. event e
The functional semantics of event e can be described as multi-component system: FS (e)=< activity, resource,
[option], constraint >, in which:
Activity=C | C ∈ CaIndicate activity corresponding to event e, wherein CaIt is movable general to indicate that certain field describes
Set, such as teach, exam, assess are read, #act (e) is denoted as.
Resource=C | C ∈ CrIndicating resource involved by event e, main body, the event initiated comprising activity are made
Object etc., wherein CrIndicate that certain field describes the concept set of resource, such as teacher, student, assistant,
It is denoted as #res (e).
Option=C | C ∈ CoIndicate other attributes involved by event e, and such as place, time, wherein CoIt indicates
Certain field describes the concept set of corresponding other attributes, is related to concept such as lab, the classroom in place;It is related to the concept of time
Such as year, time.[option] indicates that this is option in the functional semantics of event e.
For all properties of event, defined function kind indicates the affiliated type of event attribute.Such as kind (#res (e))
=res.
Constraint=Q1∧Q2∧…∧Qn, indicate each attribute constraint condition for occurring to meet when some event,
Middle QiThe specific constraint condition of a certain attribute occurs for (i=1,2 ..., n) expression event.It will be between the constraint condition of different attribute
Be defined as " and " relationship.If (capacity > 50) ∧ (multi-media=true) is two conditions that place selects,
Capacity > 50 indicates that its galleryful should be greater than 50, multi-media=true expression and should have multimedia equipment herein.
From the foregoing, it will be observed that Event Function semantic description is a multi-component system, the core of multi-component system is movable activity, resource
Resource is the main body or object of active action, and option indicates other attributes of event, and constraint is to event category
Property carry out restriction and specification.
Identical describing mode is used when to guarantee mark and model calibration, establishing domain body tree, the function of event is retouched
It states and is mapped in ontology tree.On the one hand the foundation of ontology tree can avoid causing semantic conflict, be on the other hand used as event similarity
Calculating.
Define 5. ontology trees
Enabling O is the ontology tree on the D of field, then has:
O=< ({ C }, { R }) | Ci∈ D, i=1,2 ..., m;Rj∈ D, j=1,2 ..., n >,
Wherein, C is concept set, indicates the related notion on the D of field, and can be divided into class concepts and Feature concept, reality
Example concept and Factors ' Concept.Wherein class concepts characterization have same nature object set, as teacher, place,
Classroom etc.;Feature concept reflects feature possessed by concept C, as the feature of classroom has capacity (to accommodate
Number);Instance concepts illustrate the respective instance of concept;Factors ' Concept characterizes institute's element of concept.
R is the set of relationship.Herein, the relationship of ontology definition has: is-a, instance-of, element-of and
trait-of.Wherein, is-a indicates the inheritance between concept, and instance-of is indicated between concept and concept specific example
Relationship, part-of refers to the relationship between concept and the element of concept, trait-of indicate concept and concept feature it
Between relationship.The partial function ontology tree that Fig. 2 illustrates as teaching field.
Define 6 Ontologies mark
It is the process of concept in field function ontology tree, the referred to as sheet of concept by given concept k replacement or partial replacement
Body semantic tagger, is denoted as remark (k).
Define the Ontology mark of 7 events
The functional semantics FS (e) of event e is subjected to ontology mark substantially by each attribute of event in ontology tree
Carry out the process of semantic tagger.Firstly, the concept #attr (e) for being related to event e carries out this in the ontology tree in certain field
Body semantic tagger remark (#attr (e)).If some attribute of event e has attribute value value (#attr (e)), basis
The trait-of attribute of remark (#attr (e)) provides the occurrence of value (#attr (e)).Such as in Fig. 1, process model
N1Transition t6Represented event e, #act (e) is teach course, #res1It (e) is classroom, value is limited to
B201.The functional semantics of story part e can be described as FS (e)=< teach course, classroom, classroom=B201
>.Ontology tree according to Fig.2, having remark (classroom)=classroom, B201 is the limit value of classroom,
And it is capacity and multi-media that classroom, which has trait-of, then can acquire capacity (B201) and multi-
media(B201)。
Similarly, the attribute of event each in mark can be marked on ontology tree.As there are event e, #act in mark
It (e) is teach course, #res1It (e) is classroom, value B202.Therefore it is found that remark (classroom)=
Then classroom acquires capacity (B202) and multi-media (B202).
Semantic tagger of the event attribute in ontology tree when especially relating to attribute occurrence, needs manually to mark.Cause
This, the semantic tagger of each attribute of event is semi-automatic completion.Meanwhile the ontology tree in each field can be in the case where domain expert participates in
It is constructed referring to WordNet and HowNet.
Process based on ontology tree calculating event similarity is as follows:
It can be exchanged into the calculating of similarity between the multiple attributes of event to the similarity calculation of event in process model and mark.
The similarity calculation of attribute can be carried out based on ontology tree.Specific step is as follows:
Event e in step 1, input markσWith the event e in modelN, the activity attributes of two events need identical.
Step 2, respectively by two event eσAnd eNIt is expressed as the form D S (e of its functional semanticsσ) and FS (eN), by function language
Adopted FS (eσ) and FS (eN) each attribute be labeled on ontology tree.
Step 3 calculates similarity of every a pair of of attribute on ontology tree in two events, records the type and respectively of each attribute
The corresponding similarity of attribute, then the similarity Sim-event (e using its average value as two eventsσ,eN)。
Below with event e in modelN, event e in markσ(two events or activities attributes are identical, i.e.,e N=e σ) for, classification is said
The computation rule of bright similarity:
Each attribute (and attribute in functional semantics) of two events is labeled on ontology tree and distinguishes table
It is shown as (x, y), the characteristics of x and y therein are according to its ontology tree, has allowed the features such as traitof.To concept to (x, y),
Similarity sim (x, y) can divide following situation to discuss:
(1) y ∈ (C if it does not existr(eσ)∪Co(eσ)), so that kind (x)=kind (y), then sim (x, y)=0;
(2) y ∈ (C if it existsr(eσ)∪Co(eσ)), so that kind (x)=kind (y), and
If y ∈ descendant (x) or y ∈ instance-of (x), then sim (x, y)=1;
IfAndBetween then sim (x, y)=Dis (x, y), i.e. x, y
Similarity sim (x, y) is its distance in ontology tree;
(3) y ∈ (C if it existsr(eσ)∪Co(eσ)), so that kind (x)=kind (y), and
Each constraint Q of x and y will be limitedi, it is mapped as its feature trait-ofi(x) and trait-ofi(y):
If (a)sim(Qi(xi,yi))=Dis (x, y);
If (b)And trait-ofi(y) it is unsatisfactory for trait-ofi(x), then sim (Qi(xi,yi))=
0;
If (c)Then trai-tofi(y) meet trait-ofi(x)
If y ∈ descendant (x) or y ∈ instance-of (x), sim (Qi(xi,yi))=1;
IfAndThen sim (Qi(xi,yi))=Dis (x, y);
Then, sim (x, y)=average (sim (Qi(xi,yi));
In the above calculating process, function kind indicates the affiliated type of event attribute;If there is kind (x)=kind (y), andsim(Qi(xi,yi)) it is in QiRestriction under xi,yiSimilarity, and sim (x, y) be equal in various different QsiUnder
sim(Qi(xi,yi)) average value, wherein trait-of is one of relationship present in ontology tree.WhereinIt indicates to the condition
There is no limit, such asExpression is limited to sky to x.Trait-of indicates the feature of attribute, is the collection in ontology tree
One of conjunction relationship.
In similarity calculation of the attributive concept to (x, y) of the above event, the event e of (1) finger markσIn be not present and mould
The event e of typeNIn the corresponding concept y of attribute x, as there is resource in x, and do not have in y, at this time sim (x, y) be 0;
(2) refer to that working as (x, y) is identical concept pair, and not to the restriction of x in constraint condition, if y is the descendants's node or x of x
Example, then it is assumed that concept y meets x, similarity 1;(such as two attributes are all concept resource, and x is corresponding general on semantic tree
It reads place and is constrained without specific, and y belongs to the specific example of classroom or place);If y be not x descendants's node or
Example, then with x, distance of the y on semantic tree indicates its similarity;(such as x corresponds to concept Lab, and y corresponds to classroom);
(3) if there is the concept similar with x in y, and there is the restriction constraint to x, then each of which is constrained into QiIt is mapped as its feature
trait-ofi(x) and trait-ofi(y).If constraining Q herein to the sky that is constrained to of yiOn, sim (Qi(xi,yi)) it is Dis
(x,y);If the constraint to y is not sky, compare trait-ofi(y) and trait-ofi(x).If trait-ofi(y) it is unsatisfactory for
trait-ofi(x), then sim (Qi(xi,yi))=0;If trait-ofi(y) meet trait-ofi(x), if y is descendants's knot of x
When point or example, then (x, y) is in constraint QiOn similarity sim (Qi(xi,yi)) it is 1, if y is not the descendants's node or reality of x
Example, then (x, y) is in constraint QiOn similarity sim (Qi(xi,yi)) it is Dis (x, y).For example, it is assumed that the constraint condition to x is
Classroom=B201, then being mapped trait-of (classroom) is capacity (B201);If being constrained to y
Sky, at this time under this constraint, the similarity of (x, y) is the distance Dis (x, y) of the two in the body;Assuming that being constrained to y
Lab=C305 is then mapped as capacity (C305), then compares capacity (B201) and capacity (C305).
If capacity (C305) is unsatisfactory for capacity (B201), under this constraint, the similarity of (x, y) is 0;If capacity
(C305) meet capacity (B201), because y is the example or descendants's node that Lab is not classroom, then under this constraint,
The similarity of (x, y) is Dis (x, y);If y is the example or descendants's node of classroom, under this constraint, the phase of (x, y)
It is 1 like degree.When to x, there are when multiple constraints, the similarity of (x, y) is the average value of each properties similarity of constraint.In sim
In the calculating of (x, y), Dis (x, y) indicates the distance of (x, y) in semantic tree.Meet defining for condition, it can be according to concrete concept
Trait-of depending on, be generally understood as " being better than ".It is false if multi-media is true better than multi-media.
Model N as shown in Figure 11In, to transition t6Corresponding event e, #act (e)=teacher course, #
Resource (e)=teacher:sam, #place (e)=classroom:B201.Event e ', # in corresponding mark
Act (e ')=teachercourse, #resource (e ')=teacher:Sam, #place (e ')=classroom:B202.
By above-mentioned steps it is found that resource attribute, sim (x, y)=1, to place attribute, #place (e)=#place (e ')=
Classroom is identical concept, but to event e, is constrained to Q (x)=B201, and to event e ', it is constrained to B202, therefore
This judges whether each trait-of of B202 meets the requirement of B201.It is assumed that capacity (B202) < capacity (B201),
But multi-media (B201)=true, and multi-media (B202)=false, constraint #place (e)=
Under classroom:B201, the similar value of attribute place is 0.5.Therefore two event similarity calculation results are as follows: Sim-event
(e, e ')=0.75, com_attr [0]={ resource, 1 }, com_attr [1]={ place, 0.5 }.
Below based on the similarity between event, the related definition of calibration is provided.
Define 8 calibrations based on event similarity
For the mark σ and model N of event, the calibration γ based on event similarity is a series of set of movements.Calibration makes
The series that constitutes of first element is mark σ (remove > >) in log, the sequence that second element is constituted is to change in model N
The generation sequence (remove > >) of corresponding event.Wherein, each effectively movement is event to (m, n):
(1) (m, n) claims log mobile, ifm∈σ↓actAndn=> >;
(2) (m, n) is mobile for model, ifm=> > andn∈N↓act;
(3) (m, n) is weak synchronizing moving, ifm∈σ↓act,n∈N↓act,m=nAnd there is Sim (m, n) < λ.
(4) (m, n) is strong synchronizing moving, ifm∈σ↓act,n∈N↓act,m=nAnd there is Sim (m, n) >=λ.
Wherein, λ is an adjustable threshold value, and 0≤λ≤1.
Define the cost value of 9 calibrations
Reference standard likelihood function enables the cost value of log movement, model mobile, weak synchronizing moving and strong synchronizing moving
Cost is denoted as (1,1,1,1-Sim-event (e respectivelyσ,eN)), it is called similarity likelihood function.Wherein, Sim-event (eσ,
eN) indicate event to (eσ,eN) between similarity.
Based on similarity likelihood function, the cost value cost (γ) of a calibration γ is the log movement for including in calibrating γ
The mobile number of number, model, weak synchronizing moving number and strong synchronizing moving number * (1-Sim-event (eσ,eNThe sum of)).
Define 10 optimal calibrations
Ψ (σ, N) is enabled to indicate mark σ and model N based on the calibration set under similarity likelihood function.γ if it exists1∈
Ψ (σ, N) is rightThere is cost (γ)≤cost (γ1), then claim γ1Similarity likelihood function is based on for σ and N
Optimal calibration.Similarly, the optimal calibration set between σ and N based on similarity likelihood function is expressed as Ψ0(σ,N)。
Calibration process are as follows: firstly, mark σ is converted to its corresponding event net, it is calculated to the identical event of activity attributes
The similarity of event, and the product net of tectonic event net and model net, the movement transition in product net being mapped as in calibration.Its
It is secondary, the cost value of each moved further is provided according to the definition for calibrating cost value under similarity likelihood function, from the initial mark of product net
Knowing end of identification has several ways diameter, and in these paths, the smallest optimal path of cost value corresponds to mark and model most
Excellent calibration.
Fig. 3 show the model N indicated with Labelled Petri Net2.Fig. 4 show mark σ=< (a, Resource1:x1),(b,
Resource2:m1,Place1:y1),(c,Place2:z1),(e,Place3:k1) > represented by event net N2′.Model N2And thing
Part net N2' product net N " it is as shown in Figure 5.In figs. 3 and 4, the label of each transition t is an event e, the work of event e
Dynamic attributee, such as a, b, c etc..In product net N ", each synchronizing moving has the similarity and thing of two events in the movement
Similarity between each attribute and attribute of part.Transition in product net are mapped as the movement in calibration, and according to similarity
Likelihood function provides the cost value of the corresponding calibration in different paths, and wherein the smallest path of cost value is optimal calibration.
Just calibration is compared with simple alignment below.
It, will be in calibration and simple log based on event log L and model N on the basis of discussing event similitude
Mark is compared with the calibration of naive model, can obtain following inference.To any bar mark σ and model N in event log L,
Corresponding simple mark and naive model are respectively σ↓actAnd N↓act。
Inference 1 is to any bar mark σ and model N, cost (Ψ in event log L0(σ,N))≥cost(Ψ0(σ↓act,
N↓act))。
It proves: enabling γ ∈ Ψ0(σ, N), then cost (γ) be the mobile number of log, the mobile number of model, weak synchronizing moving number and
Strong synchronizing moving * (1-Sim-event (eσ,eNThe sum of)).And to γ ' ∈ Γ0(σ↓act,N↓act), cost (γ ') is log in γ
The sum of mobile number and the mobile number of model.Because it is identical for moving about log with the definition of model movement in two kinds of calibrations,
Be (it is movable in mark, > >) and (> >, it is movable in model).Because of 0≤Sim-event (eσ,eN)≤1, so to model N and day
Mark σ in will L has cost (γ) >=cost (γ ').
Inference 1 illustrates, if considering attribute of the event in addition to activity, the cost of mark and model calibration in event log
Value will increase, i.e., the degree that mark and model are consistent can decline.This is because when considering multiple attributes of event, school at this time
There is weak synchronizing movings and strong synchronizing moving in standard.Even if activity on the move is identical, it is still necessary to from the similar of other attributes
Degree further calculates its cost value.This is from other side explanation, " matter " of log and model bias that simple alignment judges, and
Calibration then on the basis of matter, provides " amount " of log and model bias.
Inference 2 is to γ ∈ Ψ0(σ, N), not necessarily there is γ1∈Γ0(σ↓act,N↓act).Wherein, γ1In activity on the move
For the active map of event in γ.Vice versa.
It proves: according to inference 1, if γ ∈ Ψ0(σ, N), then it is rightThere is cost (γ)≤cost
(γ′).Under the definition of similarity likelihood function, the cost value of γ and γ ' are that number is moved in log therein, model moves number, weak
Synchronizing moving number and strong synchronizing moving * (1-Sim-event (eσ,eN)).When γ and γ ' is mapped as simple alignment γ1And γ1′
When, γ1And γ1' cost value be γ1And γ1The sum of the mobile number of ' middle log and the mobile number of model.But cost (γ)≤cost
(γ ') can not necessarily release cost (γ1)≤cost(γ1′).Therefore, γ1∈Γ0(σ↓act,N↓act) not necessarily set up.
It can similarly be issued a certificate to its inverse proposition.
2 explanation of inference, it is contemplated that with the similarity between more attribute events, calibration and simple alignment are not necessarily present pair
It should be related to.
Inference 3 is to the mark σ in log L1IfAnd γ2∈Ψ(σ1, N), and cost (γ1)=cost
(γ2).If then γ1And γ2When middle log is mobile, model is mobile, the number of weak synchronizing moving and strong synchronizing moving is all the same, then
There is cost (γ1')=cost (γ2'), wherein γ1′∈Γ(σ1↓act,N↓act), γ2′∈Γ(σ↓act,N↓act).Otherwise not at
It is vertical.
It proves: can similarly be proved with conclusion 2.
Inference 3 illustrates, if there are two the identical optimal calibrations of cost value for mark, four kinds of shiftings only in the two optimal calibrations
When dynamic number difference is identical, their optimal simple alignment also has identical cost value.But when there are two cost value is identical for mark
Optimal simple alignment when, the cost value of optimal calibration is not necessarily the same.
As described in being pushed forward by 2, the optimal simple alignment of mark and model and optimal calibration result may be inconsistent, in conjunction with inference
1, optimal calibration can more accurately reflect the degree of consistency of the mark and model in log.
If having the event log L of process model N and it, wherein event has activity attributes and two kinds of resources attribute res1With
res2, such as the res of teaching process model1For teacher, res2For course.Each event in N and L is mapped as its work
It is dynamic, obtain its corresponding simple procedure model N↓actWith event log L↓act。
(1) the optimal mark in event log and optimal example
Fig. 6 (a) show simple procedure model N↓actWith event log L↓actIn every mark optimal simple alignment and its generation
Value.As shown in Fig. 6 (a), by comparing the optimal calibration cost value of each example (case) mark, it can be obtained in event log
The smallest mark σ of cost valuei↓act, so that cost (γi)=min (cost (γj)), 1≤j≤n.Claim mark σi↓actFor L↓actOn most
Excellent mark, σi↓actCorresponding example caseiFor optimal example.
Fig. 6 (b) show the optimal calibration of more attribute events of every mark and its generation in process model N and event log L
Value.It is similar with optimal simple alignment, according to the cost value of optimal calibration, optimal cost value can be obtained in event log L most
Small optimal mark optimal example corresponding with its.Meanwhile it can be from res1And res2Angle evaluated, to being related to res1's
The corresponding event log of example is denoted as L (res1), L (res can be obtained1) in optimal mark and optimal example.Similarly, to event
Log L (res2), L (res also can be obtained2) in optimal mark and optimal example.
Thus, it can be known that calibration can be evaluated in event log from more angles compared with optimal simple alignment
Mark and example.
(2) simple alignment is compared with calibration structure
Fig. 7 gives the structural schematic diagram of simple alignment and calibration.In Fig. 7 (a), each movement in simple alignment is
Event in mark and model is to (remove > >) and cost value.In the calibration shown in Fig. 7 (b), its weak synchronizing moving and strong same is enabled
The cost value of moved further is zero, and from the point of view of " diminution ", its corresponding simple alignment can be obtained.When " amplification " more attribute events
Calibration when, then can be further discovered that the reason of its synchronizing moving similarity may not be for 1.Therefore, the calibration of more attribute events
On the one hand reflect the position and quantity that log and model bias occur from " matter ", it is on the other hand also more anti-from " amount "
Position and the reason for reflecting deviation appearance, as Place is inconsistent.Therefore, calibration of the present invention is the more careful of simple alignment
It indicates, also provides more detailed foundation for the problems in improved model or discovery log.
Claims (5)
1. a kind of event log based on event similarity and process model calibration method, it is characterised in that the following steps are included:
Firstly, being defined to Event Concepts, including process model, the mark of event, event log and active map;
Secondly, calculating the similarity between each attribute of two events based on ontology tree;Specific calculating process are as follows:
Step 1. inputs the event e in markσWith the event e in process modelN, the activity attributes of two events need identical;
Step 2. is by two events its functional semantics FS (eσ) and FS (eN) form indicate, then will be each in its functional semantics
A attribute is labeled on ontology tree respectively and calculates its similarity;
Step 3. further calculates the similarity of two events according to the similarity of each attribute of two events;
Finally, based on the similarity of event in event log mark and model calibrate;Specific calculating process is as follows:
Step 1. reference standard likelihood function defines similarity likelihood function and its cost value;
Step 2. is based on similarity likelihood function, defines the cost value cost (γ) of calibration;
The mark σ of event is converted to event net, the thing with same campaign attribute in calculating process model and event net by step 3.
The synchronizing moving similarity of part, and the product net of tectonic event net and process model;
Step 4. finds out the different paths in product net from beginning state to final state, and the transition in product net are mapped as calibrating
In movement;
For step 5. according to the definition for calibrating cost value under similarity likelihood function, the smallest path of cost value is optimal calibration
γ。
2. the event log according to claim 1 based on event similarity and process model calibration method, feature exist
In: it is defined as follows in the first step:
It defines 1. process models: process model being expressed as a binary group (PN, Attr), wherein PN=(P, T;F, M), be
One Petri network, wherein P is that a set of library is closed, and T is a transition set,It is flow relation
Set,
M:P → Num.M ∈ NumPIt is a mark function, wherein Num indicates natural number set, M ∈ NumPIndicate M:P → Num
The function being defined on set P for one;Attr is a label function, and has Attr:T → ξ ∪ τ, wherein ξ be it is all can
The event sets that can occur, τ are null event, and Attr function is that the transition of each of Petri network are mapped as event set ξ or sky
Element in event τ, event e are an elements in set ξ, be expressed as (e, #res (e) [: value]), whereineFor event
Classifier indicates with the activity name of event, #res (e) [: value] it indicateseRespective attributes and attribute value, wherein [:
Value] expression attribute value be option;
Define the mark of 2. events: when an event has multiple attributes, trace description is the finite sequence of an event, i.e., <
(e 1,#res(e1)[:value]),…,(e n,#res(en) [: value]) >, event log is the multiset of mark;
Define 3. active maps: the method that the event with multiple attributes is indicated with its activity attributes, referred to as activity are reflected
It penetrates, is denoted as↓act。
3. the event log according to claim 1 based on event similarity and process model calibration method, feature exist
In: in the case where domain expert participates in referring to WordNet and HowNet construction ontology tree, the concept used when to construction ontology tree is carried out
Definition:
Define the functional semantics of 4. event e
The functional semantics of event e are described as multi-component system: FS (e)=< activity, resource, [option],
Constraint >, in which:
Activity=C | C ∈ CaIndicate activity corresponding to event e, wherein CaIndicate that certain field describes movable concept set
It closes, is denoted as #act (e);
Resource=C | C ∈ CrIndicate resource involved by event e, wherein CrIndicate that certain field describes the concept of resource
Set, is denoted as #res (e);
Option=C | C ∈ CoIndicate other attributes involved by event e, wherein CoIndicate that the description of certain field is corresponding other
The concept set of attribute;
Constraint=Q1∧Q2∧…∧Qn, indicate each attribute constraint condition for occurring to meet when some event, wherein Qi
The specific constraint condition of (i=1,2 ..., n) expression a certain attribute of event;Constraint condition Q1, Q2..., QnBetween be defined as " simultaneously
And " relationship;
Define 5. ontology trees
Enabling O is the ontology tree on the D of field, then has:
O=< ({ C }, { R }) | Ci∈ D, i=1,2 ..., m;Rj∈ D, j=1,2 ..., n >;
Wherein, C is concept set, indicates the related notion on the D of field, and can be divided into class concepts and Feature concept, example are general
Thought and Factors ' Concept;Wherein class concepts characterization has the set of same nature object;Feature concept, which reflects concept C, to be had
Feature;Instance concepts illustrate the respective instance of concept;Factors ' Concept characterizes institute's element of concept;R is relationship
Set, the relationship of ontology definition have: is-a, instance-of, element-of and trait-of, wherein is-a indicates concept
Between inheritance, instance-of indicates the relationship between concept and concept specific example, and part-of refers to concept and concept
Element between relationship, trait-of indicates the relationship between concept and the feature of concept;
Define 6. Ontologies mark
It is the process of concept in ontology tree by given concept k replacement or partial replacement, the referred to as Ontology of concept marks, note
Make remark (k);
Define the Ontology mark of 7. events
It is substantially to carry out each attribute of event in ontology tree that the functional semantics FS (e) of event e, which is carried out ontology mark,
The process of semantic tagger;Firstly, the concept #attr (e) for being related to event e carries out Ontology mark in ontology tree, if
Some attribute of event e has specific attribute value, then provides its occurrence according to the trait-of marked.
4. the event log according to claim 1 based on event similarity and process model calibration method, feature exist
In: based on event attribute the step of the mark on ontology tree seeks similarity are as follows:
If eNFor an event in process model, eσIt is an event in mark, and hase N=e σ, two events are expressed as its function
It can semanteme FS (eN) and FS (eσ) form, and by FS (eN) and FS (eσ) in each attribute be labeled on ontology tree;If
eNAnd eσMark of a certain attribute in ontology tree is expressed as x and y, then concept is to the similarity sim (x, y) between (x, y)
Following situation can be divided to discuss:
(1) y ∈ (C if it does not existr(eσ)∪Co(eσ)), so that kind (x)=kind (y), then sim (x, y)=0;
(2) y ∈ (C if it existsr(eσ)∪Co(eσ)), so that kind (x)=kind (y), and
If y ∈ descendant (x) or y ∈ instance-of (x), then sim (x, y)=1;
IfAndIt is similar between then sim (x, y)=Dis (x, y), i.e. x, y
Spending sim (x, y) is its distance in ontology tree;
(3) y ∈ (C if it existsr(eσ)∪Co(eσ)), so that kind (x)=kind (y), and
Each constraint Q of x and y will be limitedi, it is mapped as its feature trait-ofi(x) and trait-ofi(y):
If (a)Then sim (Qi(xi,yi))=Dis (x, y);
If (b)And trait-ofi(y) it is unsatisfactory for trait-ofi(x), then sim (Qi(xi,yi))=0;
If (c)Then trai-tofi(y) meet trait-ofi(x)
If y ∈ descendant (x) or y ∈ instance-of (x), sim (Qi(xi,yi))=1;
IfAndThen sim (Qi(xi,yi))=Dis (x, y);
Then, sim (x, y)=average (sim (Qi(xi,yi));
In the above calculating process, function kind indicates the affiliated type of event attribute;If there is kind (x)=kind (y), andsim(Qi(xi,yi)) it is in QiRestriction under xi,yiSimilarity, and sim (x, y) be equal in various different QsiUnder
sim(Qi(xi,yi)) average value, wherein trait-of is one of relationship present in ontology tree.
5. the event log according to claim 1 based on event similarity and process model calibration method, feature exist
In: reference standard likelihood function enables cost value cost points of mobile log, model movement, weak synchronizing moving and strong synchronizing moving
It is not denoted as (1,1,1,1-Sim-event (eσ,eN)), it is called similarity likelihood function, wherein Sim-event (eσ,eN) table
Show event to (eσ,eN) between similarity, be the average value of above-mentioned given each attributes similarity of event;Based on similarity
Likelihood function, calibration cost value cost (γ) are to calibrate the mobile number of the log for including, model mobile number, weak synchronizing moving number in γ
With strong synchronizing moving number * (1-Sim-event (eσ,eNThe sum of)).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910250146.4A CN110109921B (en) | 2019-03-29 | 2019-03-29 | Event log and process model calibration method based on event similarity |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910250146.4A CN110109921B (en) | 2019-03-29 | 2019-03-29 | Event log and process model calibration method based on event similarity |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110109921A true CN110109921A (en) | 2019-08-09 |
CN110109921B CN110109921B (en) | 2021-08-06 |
Family
ID=67484749
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910250146.4A Expired - Fee Related CN110109921B (en) | 2019-03-29 | 2019-03-29 | Event log and process model calibration method based on event similarity |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110109921B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102012918A (en) * | 2010-11-26 | 2011-04-13 | 中金金融认证中心有限公司 | System and method for excavating and executing rule |
US8745057B1 (en) * | 2011-11-28 | 2014-06-03 | Google Inc. | Creating and organizing events in an activity stream |
CN105069306A (en) * | 2015-08-18 | 2015-11-18 | 山东科技大学 | Event log audit method based on workflow net model |
CN106204345A (en) * | 2016-06-29 | 2016-12-07 | 安徽理工大学 | The optimization method of service interaction process model based on configuration information |
US20180365654A1 (en) * | 2017-06-20 | 2018-12-20 | Microsoft Technology Licensing, Llc | Automatic association and sharing of photos with calendar events |
-
2019
- 2019-03-29 CN CN201910250146.4A patent/CN110109921B/en not_active Expired - Fee Related
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102012918A (en) * | 2010-11-26 | 2011-04-13 | 中金金融认证中心有限公司 | System and method for excavating and executing rule |
US8745057B1 (en) * | 2011-11-28 | 2014-06-03 | Google Inc. | Creating and organizing events in an activity stream |
US9223803B2 (en) * | 2011-11-28 | 2015-12-29 | Google Inc. | Creating and organizing events in an activity stream |
CN105069306A (en) * | 2015-08-18 | 2015-11-18 | 山东科技大学 | Event log audit method based on workflow net model |
CN106204345A (en) * | 2016-06-29 | 2016-12-07 | 安徽理工大学 | The optimization method of service interaction process model based on configuration information |
US20180365654A1 (en) * | 2017-06-20 | 2018-12-20 | Microsoft Technology Licensing, Llc | Automatic association and sharing of photos with calendar events |
Also Published As
Publication number | Publication date |
---|---|
CN110109921B (en) | 2021-08-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wang et al. | Attribute‐level and pattern‐level classification consistency and accuracy indices for cognitive diagnostic assessment | |
Berryman | Review of software platforms for agent based models | |
US20080227064A1 (en) | System for evaluating information competency and method thereof | |
Mao et al. | Analysis of entrepreneurship education in colleges and based on improved decision tree algorithm and fuzzy mathematics | |
Fauzan et al. | Automated Class Diagram Assessment using Semantic and Structural Similarities. | |
CN113220908A (en) | Knowledge graph matching method and device | |
Zhang | A method of recommending physical education network course resources based on collaborative filtering technology | |
Jahanshahloo et al. | Using the minimum distance of DMUs from the frontier of the PPS for evaluating group performance of DMUs in DEA | |
Song et al. | Handling multi-granular hesitant information: A group decision-making method based on cross-efficiency with regret theory | |
Huang et al. | Goal programming models for incomplete interval additive reciprocal preference relations with permutations | |
Hua et al. | Application of support vector machine model based on machine learning in art teaching | |
CN112257959A (en) | User risk prediction method and device, electronic equipment and storage medium | |
CN110109921A (en) | Event log and process model calibration method based on event similarity | |
Jordans et al. | Leadership in Africa: Past, present, and future perspectives | |
CN116186297A (en) | Graph manifold learning-based literature relationship discovery method and system | |
Pratiwi | Implementation of function point analysis in measuring the volume estimation of software system in object oriented and structural model of academic system | |
CN114639044A (en) | Label determining method and device, electronic equipment and storage medium | |
Xian et al. | DHHFL-MABAC approach based on distance measure and comprehensive weight for sewage treatment company selection | |
Accoto et al. | A probabilistic method for reconstructing the Foreign Direct Investments network in search of ultimate host economies | |
Wilkerson | Customising structure of graphical models | |
Sekiya et al. | Development of a curriculum analysis tool | |
Wróblewska et al. | SYNAT system ontology: design patterns applied to modeling of scientific community, preliminary model evaluation | |
CN116610964B (en) | Text similarity matching method and device and computer equipment | |
Deepthi et al. | Barriers to Industry 4.0 adoption: mapping the global research trends using bibliometric analysis | |
Wu et al. | Evaluation of teaching quality based on binary tree support vector machine |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210806 |
|
CF01 | Termination of patent right due to non-payment of annual fee |