CN105787072B - A kind of domain knowledge of Process-Oriented extracts and method for pushing - Google Patents
A kind of domain knowledge of Process-Oriented extracts and method for pushing Download PDFInfo
- Publication number
- CN105787072B CN105787072B CN201610117829.9A CN201610117829A CN105787072B CN 105787072 B CN105787072 B CN 105787072B CN 201610117829 A CN201610117829 A CN 201610117829A CN 105787072 B CN105787072 B CN 105787072B
- Authority
- CN
- China
- Prior art keywords
- knowledge
- vector
- factor
- user
- extracts
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/338—Presentation of query results
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
- G06F18/2193—Validation; Performance evaluation; Active pattern learning techniques based on specific statistical tests
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Abstract
A kind of domain knowledge of Process-Oriented extracts and method for pushing, method are: extracting knowledge from knowledge resource and is indicated with knowledge vector, process vector is extracted from flow tasks, extracts user property vector from user's description;By carrying out matching primitives between process vector, user property vector sum knowledge vector, the degree of correlation of knowledge requirement and knowledge is obtained;Knowledge score value is calculated by user behavior variable, is finally corrected with result of the knowledge score value to knowledge relatedness computation, obtains the knowledge sort closely related with flow tasks as a result, realizing more acurrate push of the knowledge to flow tasks.It is considered herein that the essence of knowledge push is that in the form of correct, correct knowledge is transmitted to correct people in the correct time, the domain knowledge of Process-Oriented provided by the invention extracts that work efficiency is high, accuracy is good with supplying system and method.
Description
Technical field
The present invention relates to a kind of computer literacy application systems, especially relevant to customer service activity know for extracting
Know, and intelligently pushes knowledge according to actual needs to the movable method and system of related service.
Background technique
Under environment of knowledge-based economy, knowledge is just becoming the most important resource of more and more enterprises, and its essence is be considered as knowledge
Most important corporate resources gives appropriate Knowledge delivery to appropriate user, thus effectively by the acquisition of knowledge and shared
The knowledge innovation that ground improves Company Knowledge utilization rate, promotes employee makes enterprise obtain competitive advantage.
Information management and it is integrated be that improve Company Knowledge reuse horizontal, prevent Organizational Knowledge Loss Caused, realize operation flow automation,
Intelligentized effective means.Enterprise has accumulated a large amount of experience and knowledge in process of production, but these knowledge are mostly with number
Exist according to the form of implicit knowledge in library, knowledge base, paper document and human brain, the mode that people obtain knowledge mainly passes through craft
Retrieval not can guarantee the timeliness of knowledge acquisition, accuracy and comprehensive.So studying existing knowledge and product design activity
It is associated, it is integrated with design process, so that the participant of R & D design process obtains knowing for needs in the appropriate time
Knowledge seems necessary.
R & D design process is knowledge-intensive active set, and a large amount of rule knowledge, reality are applied in one side design activity
Example knowledge, empirical data carry out decision and design;Another aspect design activity is also the source that knowledge is generated and obtained.Know
Knowing integrated target is exactly by Design Knowledge Management in conjunction with design process, by design activity role, object, target, need
The description for seeking equal attributes, is matched with relevant knowledge, to reduce manual knowledge search work, realization will rapidly and accurately be set
Meter knowledge is pushed to designer.
Some scholars and tissue are studied and have been attempted to Company Knowledge push both at home and abroad at present.It is being based on process
Aspect is studied with event driven knowledge supplying system, main book has: in document: Wang Shengfa, Gu Xinjian, Guo Jianfeng etc. are write
Paper: the knowledge active push research towards product design is loaded in " computer integrated manufacturing system ", 2007,13 (2): 234-
239, the shortcomings that knowledge cannot actively be passed to designer appropriate in due course for existing system, propose one
Kind of knowledge based management, by the product-design knowledge active push architecture of workflow logic nets and in a manner of;But its knowledge
It is relatively more with rule involved in system and object, to affect the efficiency and accuracy of knowledge matching and push;In document:
Liu Xinyu writes paper: the theoretical frame of the knowledge supplying system based on process and application study.Shenyang: Northeastern University, 2005.It is right
Knowledge supplying system based on process has mainly carried out system research to its theoretical, frame and application;It is main to knowledge push technology
Describe the method for pushing of the operation knowledge towards post and the project knowledge towards employee;But still ken is not accounted for
Influence, method has retrieving that data volume is big, push precision is not high;In document: Chen Hao, Liu Nian, Hu Yanjun write opinion
Text: the knowledge Modeling research based on workflow, be loaded in " automated manufacturing ", 2005,27 (5): 8-12 is proposed based on work
The knowledge Modeling and its system framework of stream discuss the integrated skill for supporting knowledge and knowledge, knowledge and people and knowledge and process
Art, and method knowledge being applied in the operation flow of enterprise;But this article only considered the mutual of workflow and knowledge base
Effect, does not account for knowledge agent --- and people is in influence wherein, and it is related to simply describing only to establish conceptual model
Technology.The common drawback of the studies above be propose knowledge matching with method for pushing there are low efficiency, accuracy is poor the problems such as.
Summary of the invention
The purpose of the present invention is: above-mentioned background technique there are aiming at the problem that and the neck of the new Process-Oriented of one kind that proposes
Domain knowledge extracts and supplying system and method.It is considered herein that the essence of knowledge push is in the correct time, with correct shape
Correct knowledge is transmitted to correct people by formula.Only knowledge, business events flow path and people's triplicity are got up, are just able to achieve
Effective knowledge push.
The technical scheme is that a kind of domain knowledge of Process-Oriented extracts and supplying system, including following module:
Line module, user are the targets of knowledge push, and the knowledge of user's frame of reference push executes flow tasks, goes forward side by side
Other information management activities of row;
Knowledge resource module provides a memory space for Company Knowledge, and carries out Classification Management to it, and responding to knowledge pushes away
Request is sent, the employee of line module is assisted to complete flow tasks;
Knowledge pushing module calculates reconstruction of layer by representation of knowledge abstraction, layer and knowledge and constitutes, for reducing knowledge matching model
It encloses, matching symbol closes the domain knowledge of flow characteristic, and is pushed to user, improves matching efficiency and precision;
Process module is the source of knowledge push, and process is generated by the specific requirements of enterprise, can be divided into several opposite
The attribute of knowledge can be arranged in independent sub-process task, each sub-process according to the relevant knowledge that it is related to.
Other information management activities that the line module carries out include being evaluated the knowledge of push and being fed back, issued
The implicit knowledge of oneself customizes interested knowledge, recommends knowledge etc. to other users.
The knowledge resource module is disclosed by Company Knowledge library, enterprise staff implicit knowledge, customer knowledge, enterprise external
One knowledge space of the knowledge composition of network etc..
The knowledge pushing module calculates reconstruction of layer by representation of knowledge abstraction, layer and knowledge and constitutes.
The representation of knowledge abstraction, layer includes Knowledge Extraction layer and representation of knowledge layer.
The Knowledge Extraction layer refers to be extracted from knowledge resource numerous in enterprise using text information extractive technique
Required knowledge content.
The text information extractive technique refers to extracts certain defined a kind of or a few class specific information from one section of text,
And carried out the process of structuring processing;Its treatment process includes, first the rule set that extracts of tectonic information, recycles these
Rule extracts specified information from text.
The representation of knowledge layer is that the result for extracting text information carries out information rule in the form of tables of data of database etc.
Generalized, representation of knowledge layer are corresponding with knowledge vector.
The knowledge calculates reconstruction of layer and refers on the basis of the representation of knowledge, and the demand according to flow tasks to knowledge is led to
The matching operation to knowledge is crossed, obtains qualified knowledge, and obtained knowledge is pushed to the use for executing flow tasks
Family.
The matching operation of the knowledge includes the following steps:
First, correlation calculations refer to demand (i.e. process vector) and the knowledge itself according to operation flow to knowledge
Feature (i.e. knowledge vector) and user profession, preference and experience (i.e. user property vector), know knowledge requirement is met
Know and carries out correlation calculations;
Second, knowledge score value, which calculates, refers to that the scoring to knowledge generated based on user behavior vector is calculated;
Third, knowledge sort refer to that on the basis of correlation calculations, knowledge score value calculated result intervenes correlation calculations
Sequence to knowledge, the knowledge score after being corrected are ranked up knowledge according to the sequence of score from high to low, by result
It is pushed to the user for completing a certain particular task, to realize that of the present invention and flow tasks have the knowledge being closely connected
Push.
A kind of domain knowledge of Process-Oriented extracts and method for pushing, which comprises the steps of:
First, knowledge is extracted from knowledge resource and is indicated with knowledge vector, and process vector is extracted from flow tasks, from
User property vector is extracted in user's description;
Second, by carrying out matching primitives between process vector, user property vector sum knowledge vector, show that knowledge needs
The degree of correlation for knowledge of summing;
Knowledge score value is calculated by user behavior variable in third, finally with knowledge score value to knowledge relatedness computation
Result be corrected, obtain the knowledge sort closely related with flow tasks as a result, realizing knowledge to the more quasi- of flow tasks
Really push.
It is described to refer to the knowledge money numerous out of enterprise using text information extractive technique from extraction knowledge in knowledge resource
Required knowledge content is extracted in source;The text information extractive technique refer to extracted from one section of text as defined in it is a certain
Class or a few class specific informations, and carried out the process of structuring processing.
The knowledge vector is the vector space for describing knowledge.
The source of the knowledge vector (KNOWLEDGE VECTORS, WV) includes the theme of text, author, keyword, plucks
It wants and the time;The knowledge vector is a five-tuple:
KV=SF, AF, KF, ABF, TIF
Wherein:
SF is the theme the factor (TOPIC FACTORS), SF={ sfi: I=1,2 ..., | SF | }, indicate the affiliated of text
Theme, by the theme lexical representation of finite number, same piece document can belong to one or more themes;
AF be author's factor (AUTHOR FACTORS), AF=AFJ:J=1,2 ..., | AF | }, indicate the work of knowledge
Person;
KF be the keyword factor (KEYWORD FACTORS), KF=KFK:K=1,2 ..., | KF | }, indicate text
Keyword;
ABF be abstract the factor (ABSTRACT FACTORS), ABF=ABFM:M=1,2 ..., | ABF | }, by limited
The several vocabulary from text snippet indicates;
TIF is time factor (TIME FACTORS), by six coded representations of YY-MM-DD, respectively represents document generation
Year, month and day.
The process vector comes from flow tasks, and process vector (WORKFLOW VECTORS, WV) is a triple:
WV=FF, WTF, WPF
Wherein:
FF is the field factor (FIELD FACTORS), FF={ FFI: I=1,2 ..., | FF | }, indicate business scope
Finite aggregate;WTF is the flow type factor (WORKFLOW TYPE FACTORS), WTF={ WTFJ: J=1,2 ..., | WTF | },
It indicates in orchestration instance operation phase, the flow type finite aggregate proposed according to specific example tasks and situation;WPF
For the saddlebag factor (WORKPACKAGE FACTORS), WPF={ WPFK: K=1,2 ..., | WPF | }, indicate flow tasks institute
The saddlebag characteristic item that is included finite aggregate;
Field factor FF referred in the Business Process Modeling stage, was based on tissue experience and best practices, for the activity of its composition
The classification of specified business scope is the finite aggregate being made of domain name;Flow type factor WTF refers to the affiliated class of process
Type, such as program planning class process, calculating class process, test class process;The source of saddlebag factor WPF may include saddlebag
Title, keyword, input, output, binding target and resource;
The field factor, the flow type factor in process vector, weight can be specified by user as needed in advance;
The weight of characteristic item in saddlebag is calculated by the calculation of Features weight in knowledge vector.
The user property vector refers to the vector space to characterize individual consumer's feature.
The user property vector is a triple:
UV=MF, UIF, EF
Wherein:
MF is the professional factor (MAJOR FACTORS), MF=MF I:I=1,2 ..., | MF | }, by the special of finite number
Industry title lexical representation;
UIF is the user interest factor, UIF=UIF I:I=1,2 ..., | UIF | }, by several feature lexical representations;
EF is experience factor (EXPERIENCE FACTORS), EF=EF I:I=1,2 ..., | EF | }, from user's mistake
It extracts into project experiences, is indicated by the professional technicality of finite number;Characteristic item in user property vector, can by with
Family is previously according to needing to specify its weight.
The vector set institute table that the knowledge requirement (being indicated by KR) is made of process vector WV, user property vector UV
Sign, it may be assumed that KR=WV ∪ UV.
The relatedness computation method of the knowledge requirement and knowledge are as follows:
Wherein, W1K, W2K respectively indicate the weight of knowledge requirement KR and knowledge vector KV k-th characteristic item, and 1≤K≤
N。
It is described that knowledge score value method is calculated by user behavior variable are as follows:
It is calculated based on the scoring to knowledge that user behavior vector generates.The result that knowledge score value calculates is to correlation calculations
As a result there is corrective action, act in knowledge sort and embody.
The user behavior vector be based on specific knowledge is browsed by user, is downloaded, collecting, recommending data is calculated,
It is indicated respectively with FL, FD, FF, FR.
The formula that knowledge score value (KNOWLEDGE SCORE, KS) calculates are as follows:
Wherein, FDIIt is respectively represented in the number and the A to Z of that the knowledge is downloaded with FD and is downloaded most knowledge
Download time, FLIThe browsing that most knowledge is browsed in the number and the A to Z of that the knowledge is browsed is respectively represented with FL
Number, FFIThe collection number that most knowledge is collected in the number and the A to Z of that the knowledge is collected is respectively represented with FF,
FRIThe recommendation number that most knowledge is recommended in the knowledge recommended number and the A to Z of is respectively represented with FR.By upper
Definition is stated it is found that similarly,
The calculation method of the knowledge sort result are as follows:
On the basis of correlation calculations, knowledge score value calculated result intervenes sequence of the correlation calculations to knowledge, obtains
Knowledge score after correction is ranked up knowledge according to the sequence of score from high to low, and result is pushed to and completes a certain spy
The employee of task is determined, to realize the push with flow tasks with the knowledge being closely connected of the present invention;It calculates public
Formula are as follows:
KCS=β × Sim (KR, KV)+(1- β) × KS
Wherein, the knowledge score after KCS (KNOWLEDGE CORRECTION SCORE, KCS) representative correction, SIM (KR,
It KV) is correlativity calculation result, KS is knowledge score value calculated result, by summing after the two ranking operation, after obtaining correction
Knowledge score KCS;β is the operator for adjusting correlativity calculation result and knowledge score value calculated result.
Advantages of the present invention: it is considered herein that the essence of knowledge push is in the correct time, in the form of correct, just
True knowledge is transmitted to correct people.Only knowledge, business events flow path and people's triplicity are got up, is just able to achieve and effectively knows
Know push.Work efficiency is high, accuracy for the domain knowledge extraction of aforementioned present invention offer Process-Oriented and supplying system and method
It is good.
Detailed description of the invention
Fig. 1 shows the frameworks of the extraction of Process-Oriented domain knowledge and supplying system
Fig. 2 indicates the implementation of knowledge push layer
Fig. 3 indicates knowledge operation and the sequencer procedure of Process-Oriented
Specific embodiment
Illustrate a specific embodiment of the invention with reference to the accompanying drawing:
The present invention proposes a kind of domain knowledge supplying system of Flow driving, including following module:
Line module, user are the targets of knowledge push, and the knowledge of user's frame of reference push executes flow tasks, goes forward side by side
Other information management activities of row;
Knowledge resource module provides a memory space for Company Knowledge, and carries out Classification Management to it, and responding to knowledge pushes away
Request is sent, the employee of line module is assisted to complete flow tasks;
Knowledge pushing module calculates reconstruction of layer by representation of knowledge abstraction, layer and knowledge and constitutes, for reducing knowledge matching model
It encloses, matching symbol closes the domain knowledge of flow characteristic, and is pushed to user, improves matching efficiency and precision;
Process module is the source of knowledge push, and process is generated by the specific requirements of enterprise, can be divided into multiple opposite
The attribute of knowledge can be arranged in independent sub-process task, each sub-process according to the relevant knowledge that it is related to.
Meanwhile the domain knowledge of Process-Oriented extracts and method for pushing, includes the following steps:
First, knowledge is extracted from knowledge resource and is indicated with knowledge vector, and process vector is extracted from flow tasks, from
User property vector is extracted in user's description;
Second, by carrying out matching primitives between process vector, user property vector sum knowledge vector, show that knowledge needs
The degree of correlation for knowledge of summing;
Knowledge score value is calculated by user behavior variable in third, finally with knowledge score value to knowledge relatedness computation
Result be corrected, obtain the knowledge sort closely related with flow tasks as a result, realizing knowledge to the more quasi- of flow tasks
Really push.
Architecture diagram in Fig. 1 describes the logical relation between at all levels.Major function at all levels is described as follows:
(1) client layer
User mostlys come from the employee of enterprises, is the main body and knowledge push of knowledge creation and application
Target.User can also carry out other information management activities other than the knowledge that frame of reference pushes executes flow tasks,
The implicit knowledge of oneself is such as evaluated the knowledge of push and fed back, issued, interested knowledge is customized, is pushed away to other users
Recommend the activities such as knowledge.
(2) knowledge resource layer
Knowledge resource layer is one and discloses net by Company Knowledge library, enterprise staff implicit knowledge, customer knowledge, enterprise external
One knowledge space of the knowledge composition of network etc..The main task of knowledge resource layer is that a storage is provided for Company Knowledge
Space, and Classification Management is carried out to it, responding to knowledge push request assists the employee of client layer to complete flow tasks.
(3) knowledge pushes layer
Knowledge push layer is to realize the intermediate level that is mutually related between client layer, knowledge resource layer and process layer.
Knowledge push layer can reduce knowledge matching range, and matching symbol closes the domain knowledge of flow characteristic, and is pushed to user, improve
Matching efficiency and precision.Knowledge push layer is emphasis place of the invention.
(4) process layer
Process layer is the starting point of knowledge push, is the source of knowledge push.Process is generated by the specific requirements of enterprise, can
To be divided into several relatively independent sub-process tasks, each employee is distributed to undertake and complete task.Each sub-process can
The attribute of knowledge to be arranged according to the relevant knowledge that it is related to.
2, knowledge pushes layer
As shown in Fig. 2, the knowledge push layer calculates reconstruction of layer by representation of knowledge abstraction, layer and knowledge and constitutes.
The representation of knowledge abstraction, layer includes Knowledge Extraction layer, representation of knowledge layer.
Needed for Knowledge Extraction layer is referred to and is extracted from knowledge resource numerous in enterprise using text information extractive technique
Knowledge content.
The text information extractive technique refers to extracts certain defined a kind of or a few class specific information from one section of text,
And carried out the process of structuring processing.The present invention extracts model realization Knowledge Extraction using rule-based text information.
Its treatment process includes, first the rule set that extracts of tectonic information, these rules is recycled to extract specified letter from text
Breath.
Representation of knowledge layer is that the result for extracting text information carries out information norm in the form of tables of data of database etc..
Representation of knowledge layer is corresponding with knowledge vector.About knowledge vector, it is discussed further below.
Knowledge calculates reconstruction of layer and refers on the basis of the representation of knowledge, the demand according to flow tasks to knowledge, by right
The matching operation of knowledge obtains qualified knowledge, and obtained knowledge is pushed to the employee for executing flow tasks.
3, knowledge calculating process
Knowledge calculating process by correlation calculations, knowledge score value calculate, knowledge sort and etc. form.
1) correlation calculations, refer to demand (i.e. process vector) according to operation flow to knowledge and knowledge from
The feature (i.e. knowledge vector) of body and profession, preference and the experience (i.e. user property vector) of user, to meeting knowledge requirement
Knowledge carries out correlation calculations;
In the present invention, process vector comes from flow tasks, is the main source of knowledge requirement.Process vector
(WORKFLOW VECTORS, WV) is a triple:
WV=FF, WTF, WPF
Wherein:
FF is the field factor (FIELD FACTORS), FF={ FFI: I=1,2 ..., | FF | }, indicate business scope
Finite aggregate;WTF is the flow type factor (WORKFLOW TYPE FACTORS), WTF={ WTFJ: J=1,2 ..., | WTF | },
It indicates in orchestration instance operation phase, the flow type finite aggregate proposed according to specific example tasks and situation;WPF
For the saddlebag factor (WORKPACKAGE FACTORS), WPF={ WPFK: K=1,2 ..., | WPF | }, indicate flow tasks institute
The saddlebag characteristic item that is included finite aggregate.
Field factor FF referred in the Business Process Modeling stage, was based on tissue experience and best practices, for the activity of its composition
The classification of specified business scope is the finite aggregate being made of domain name;Flow type factor WTF refers to the affiliated class of process
Type, such as program planning class process calculate class process, test class process, etc.;The source of saddlebag factor WPF may include saddlebag
Title, keyword, input, output, binding target and resource.(explanation: the saddlebag refers to work breakdown structure (WBS) (WORK
BREAKDOWN STRUCTURE, WBS) lowest level project deliverable achievement, with its structuring representation.) example
Such as, some process vector WV, field factor FF={ general field, pneumatic field, construction applications }, flow type factor WTF
={ program planning class process calculates class process }, saddlebag factor WPF={ title inputs, output, resource, binding target }.
The characteristic item general field of designated field factor FF, pneumatic field, construction applications, weight are respectively 0.3,0.3,0.2.It is specified
The characteristic item program planning class process of flow type factor WTF calculates class process, and weight is respectively 0.25,0.15.Saddlebag
The weight of characteristic item title, input, output, resource, binding target in factor WPF, by TF-IDF characteristic item in knowledge vector
The calculation of weight calculates, and weight is respectively 0.3,0.2,0.2,0.15,0.1.
Then the vector of process vector WV be expressed as WV=FF (0.3,0.3,0.2), WTF (0.25,0.15), WPF (0.3,
0.2,0.2,0.15,0.1) }.
The knowledge vector (KNOWLEDGE VECTORS, KV) is the vector space for describing knowledge, and source includes
Theme, author, keyword, abstract and the time of text.The acquisition of knowledge vector is completed by Knowledge Extraction process above-mentioned, i.e.,
Rule-based text information extracts.As described above, knowledge vector is a five-tuple:
KV=SF, AF, KF, ABF, TIF
Wherein:
SF is the theme the factor (TOPIC FACTORS), SF={ sfI: I=1,2 ..., | SF | }, indicate the affiliated of text
Theme, by the theme lexical representation of finite number, same piece document can belong to one or more themes;AF is author's factor
(AUTHOR FACTORS), AF={ AFJ: J=1,2 ..., | AF | }, indicate the author of knowledge;KF is the keyword factor
(KEYWORD FACTORS), KF={ KFK: K=1,2 ..., | KF | }, indicate the keyword of text;ABF is the abstract factor
(ABSTRACT FACTORS), ABF={ ABFM: M=1,2 ..., | ABF | }, by the vocabulary from text snippet of finite number
To indicate;TIF is time factor (TIME FACTORS), by six coded representations of YY-MM-DD, respectively represents document generation
Year, month and day, such as 13-06-01 indicate on June 1st, 2013.
For example, some knowledge vector KV, subject gene SF={ aerofoil design }, author's factors A F={ Wang Li } is crucial
Word factor K F={ aerofoil shape } makes a summary factors A BF={ a kind of method of aerofoil configuration design }, time factor TIF={ 13-
06-01}.The weight of characteristic item aerofoil design in subject gene SF, is calculated by the calculation of TF-IDF Features weight,
Weight is 0.3.The weight of characteristic item Wang Li in author's factors A F is calculated by the calculation of TF-IDF Features weight,
Weight is 0.2.The weight of characteristic item aerofoil shape in keyword factor K F, based on the calculation of TF-IDF Features weight
It calculates, weight 0.2.A kind of weight of the method for aerofoil configuration design of characteristic item in abstract factors A BF, by TF-IDF feature
The calculation of item weight calculates, weight 0.1.The weight of characteristic item 13-06-01 in time factor TIF, by TF-IDF
The calculation of Features weight calculates, weight 0.15.
Then the vector of knowledge vector KV is expressed as KV={ SF (0.3), AF (0.2), KF (0.2), ABF (0.1), TIF
(0.15)}。
The user property vector refers to the vector space to characterize individual consumer's feature.User property vector is knowledge
Another source of demand.User property vector is a triple:
UV=MF, UIF, EF
Wherein:
MF is the professional factor (MAJOR FACTORS), MF={ MFI: I=1,2 ..., | MF | }, by the profession of finite number
Title lexical representation;UIF is the user interest factor, UIF={ UIFI: I=1,2 ..., | UIF | }, by several feature vocabularies
Show;EF is experience factor (EXPERIENCE FACTORS), EF={ EFI: I=1,2 ..., | EF | }, from the passing project warp of user
It extracts in testing, is indicated by the professional technicality of finite number.
For example, some user property vector UV, profession factor M F={ master-plan }, user interest factor UIF={ fly
Action mechanics, flight flutter test }, experience factor EF={ aerofoil design, improve remodeling }.The characteristic item of specified profession factor M F
Master-plan, weight 0.2.The characteristic item flight dynamics of designated user's interest factor UIF, flight flutter test, power
Value is respectively 0.1,0.15.The characteristic item aerofoil design of specified experience factor EF improves remodeling, weight is respectively 0.15,
0.15。
Then the vector of user property vector UV be expressed as UV=MF (0.2), UIF (0.1,0.15), EF (0.15,
0.15)}。
Knowledge requirement (KNOWLEDGE REQUIREMENT, KR) from process vector WV, user property vector UV constitute to
Duration set is characterized, it may be assumed that
KR=WV ∪ UV
Example is connected, the vector of process vector WV is expressed as WV={ FF (0.3,0.3,0.2), WTF (0.25,0.15), WPF
(0.3,0.2,0.2,0.15,0.1) }, the vector of user property vector UV be expressed as UV=MF (0.2), UIF (0.1,0.15),
EF (0.15,0.15) }.
Then the vector of knowledge requirement KR be expressed as KR=FF (0.3,0.3,0.2), WTF (0.25,0.15), WPF (0.3,
0.2,0.2,0.15,0.1), (0.2) MF, UIF (0.1,0.15), EF (0.15,0.15) }.
The weight computing of characteristic item:
A. the field factor in process vector, the flow type factor, weight can be specified by user as needed in advance;Work
The weight of characteristic item in wrapping is calculated by the calculation of Features weight in knowledge vector;
B. the characteristic item in user property vector, such as professional (major name), interest (technical term), experience (technology art
Language), it can be by user previously according to needing to specify its weight;
C. in knowledge vector, Features weight is calculated, and quotes industry formula, and the present invention refers to TF-IDF function:
Wherein, TFT, DIt is characterized the frequency that a T occurs in document D, N is the number of all documents, DTFor the text containing T
Gear number mesh.The it is proposed of the function is based on such a hypothesis: the word significant to difference document should be those in document
The middle frequency of occurrences is sufficiently high, but the frequency of occurrences word few enough in other documents of entire collection of document.Such as " undercarriage ",
Occur in document D 10 times, total frequency of word is 200 in document D, then TFT, D=10/200, i.e., 0.05;N is 100000,
I.e. number of files is 100000, and the document containing " undercarriage " is 10, then Then feature weight=the 0.05*4=0.2 of " undercarriage ".
Correlation calculations are the processes that correlation is acquired between knowledge requirement (KR) and knowledge vector (KV).Specifically:
Wherein, W1K、W2KRespectively indicate the weight of knowledge requirement KR and knowledge vector KV k-th characteristic item, 1≤K≤N.
Such as the characteristic item of knowledge requirement KR is A, B, C, D, weight is respectively 30,20,20,10, and the characteristic item of knowledge vector KV is A,
C, D, E, weight are respectively 40,30,20,10, then the vector of KR is expressed as KR (30,20,20,10,0), and the vector of KV is expressed as
The degree of correlation of KV (40,0,30,20,10), then the knowledge requirement KR calculated according to above formula and knowledge vector KV are 0.86.
2) the knowledge score value calculates, and refers to that the scoring to knowledge generated based on user behavior vector is calculated.Knowledge point
Being worth the result calculated has corrective action to correlativity calculation result, acts in knowledge sort and embodies.
The user behavior vector be based on specific knowledge is browsed by user, is downloaded, collecting, recommending data is calculated,
It is indicated respectively with FL, FD, FF, FR.
The formula that knowledge score value (KNOWLEDGE SCORE, KS) calculates are as follows:
Wherein, FDIIt is respectively represented in the number and the A to Z of that the knowledge is downloaded with FD and is downloaded most knowledge
Download time, FLIThe browsing that most knowledge is browsed in the number and the A to Z of that the knowledge is browsed is respectively represented with FL
Number, FFIThe collection number that most knowledge is collected in the number and the A to Z of that the knowledge is collected is respectively represented with FF,
FRIThe recommendation number that most knowledge is recommended in the knowledge recommended number and the A to Z of is respectively represented with FR.By upper
Definition is stated it is found that similarly,The present invention couple
Downloading, browsing behavior merge processing, are assigned to a weight α to the arithmetic average of the two, to collection, behavior are recommended to carry out
Merging treatment is assigned to weight 1- α to the arithmetic average of the two.On ordinary meaning, 0.1 α≤0.4 ≦.That is, the downloading of user and
Browsing behavior influences to be lower than the influence of collection and recommendation behavior to knowledge score value for knowledge score value.
For example, FDI、FD、FLI、FL、FFI、FF、FRI, FR be respectively 40,50,80,100,30,40,15,20, if α=
0.3, then KS=0.765 in this example.
3) knowledge sort refers to that on the basis of correlation calculations, knowledge score value calculated result intervenes correlation meter
The sequence to knowledge is calculated, the knowledge score after being corrected is ranked up knowledge according to the sequence of score from high to low, will tie
Fruit is pushed to the employee for completing a certain particular task, to realize of the present invention have what is be closely connected to know with flow tasks
The push of knowledge.Its calculation formula is:
KCS=β × Sim (KR, KV)+(1- β) × KS
Wherein, the knowledge score after KCS (KNOWLEDGE CORRECTION SCORE, KCS) representative correction, SIM (KR,
It KV) is correlativity calculation result, KS is knowledge score value calculated result, by summing after the two ranking operation, after obtaining correction
Knowledge score KCS.β is the operator for adjusting correlativity calculation result and knowledge score value calculated result.Correlation calculations pair
Knowledge sort plays a major role, therefore, on ordinary meaning, 0.7 β≤0.9 ≦.For example, it is assumed that the correlation calculations of two knowledge
It as a result is respectively 0.86 and 0.80, corresponding respective knowledge score value KS is 0.72 and 0.9, if β is 0.7, then the KCS of the two distinguishes
For 0.818 and 0.83, therefore, in final knowledge sort, the sorting position of the latter will prior to the former, as with knowledge requirement
The higher knowledge recommendation of matching degree is to the user for executing flow tasks.
Claims (8)
1. a kind of domain knowledge of Process-Oriented extracts and method for pushing, which comprises the steps of:
First, knowledge is extracted from knowledge resource and is indicated with knowledge vector, process vector is extracted from flow tasks, from user
User property vector is extracted in description;
Second, by carrying out matching primitives between process vector, user property vector sum knowledge vector, obtain knowledge requirement and
The degree of correlation of knowledge, wherein the vector set institute table that the knowledge requirement KR is made of process vector WV, user property vector UV
Sign, KR=WV ∪ UV, the knowledge be from the five-tuple knowledge that the theme of text, author, keyword, abstract and time are constituted to
Measure KV, the degree of correlation expression of the knowledge requirement and knowledge are as follows:
Wherein, W1k, W2k respectively indicate the weight of knowledge requirement KR and knowledge vector KV k-th characteristic item, 1≤k≤N,
In, the characteristic item being not present in the set of knowledge requirement or knowledge, weight of this feature item in corresponding set is 0;
Knowledge score value is calculated by user behavior variable in third, final to use knowledge score value to the knot of knowledge relatedness computation
Fruit is corrected, and obtains the knowledge sort closely related with flow tasks as a result, realizing that knowledge is pushed away to the more acurrate of flow tasks
It send.
2. the domain knowledge of Process-Oriented according to claim 1 extracts and method for pushing, which is characterized in that described from knowing
Know resource in extract knowledge refer to extracted from knowledge resource numerous in enterprise using text information extractive technique it is required
Knowledge content;The text information extractive technique refers to extracts certain defined a kind of or a few specific letter of class from one section of text
Breath, and carried out the process of structuring processing.
3. the domain knowledge of Process-Oriented according to claim 1 extracts and method for pushing, which is characterized in that the knowledge
Vector is the vector space for describing knowledge.
4. the domain knowledge of Process-Oriented according to claim 1 extracts and method for pushing, which is characterized in that the knowledge
The source of vector (Knowledge Vectors, WV) includes theme, author, keyword, abstract and the time of text;It is described to know
Knowing vector is a five-tuple:
KV=SF, AF, KF, ABF, TIF
Wherein:
SF is the theme the factor (Topic Factors), SF={ sfi: i=1,2 ..., | SF | }, indicate the affiliated theme of text, by
The theme lexical representation of finite number, same piece document can belong to one or more themes;
AF be author's factor (Author Factors), AF=afj:j=1,2 ..., | AF |, indicate the author of knowledge;
KF be the keyword factor (Keyword Factors), KF=kfk:k=1,2 ..., | KF |, indicate the key of text
Word;
ABF is to make a summary the factor (Abstract Factors), ABF=abfm:m=1,2 ..., | ABF | }, by finite number
Vocabulary from text snippet indicates;
TIF is time factor (Time Factors), by six coded representations of YY-MM-DD, respectively represent document generation year,
The moon and day.
5. the domain knowledge of Process-Oriented according to claim 1 extracts and method for pushing, which is characterized in that the process
Vector comes from flow tasks, and process vector (Workflow Vectors, WV) is a triple:
WV=FF, WTF, WPF
Wherein:
FF is the field factor (Field Factors), FF={ ffi: i=1,2 ..., | FF |, indicate the finite aggregate of business scope;
WTF is the flow type factor (Workflow Type Factors), WTF={ wtfj: j=1,2 ..., | WTF |, it indicates in industry
It is engaged in the flow instance operation phase, the flow type finite aggregate proposed according to specific example tasks and situation;WPF is saddlebag
The factor (Workpackage Factors), WPF={ wpfk: k=1,2 ..., | WPF |, indicate the work where flow tasks
Wrap the finite aggregate of included characteristic item;
Field factor FF referred in the Business Process Modeling stage, was based on tissue experience and best practices, signified for the activity of its composition
The classification of fixed business scope is the finite aggregate being made of domain name;Flow type factor WTF refers to the affiliated type of process,
Such as program planning class process calculates class process, test class process;The source of saddlebag factor WPF may include the name of saddlebag
Title, keyword, input, output, binding target and resource;
The field factor, the flow type factor in process vector, weight can be specified by user as needed in advance;
The weight of characteristic item in saddlebag is calculated by the calculation of Features weight in knowledge vector.
6. the domain knowledge of Process-Oriented according to claim 1 extracts and method for pushing, which is characterized in that described
User property vector refers to the vector space to characterize individual consumer's feature.
7. the domain knowledge of Process-Oriented according to claim 1 extracts and method for pushing, which is characterized in that the user
Attribute vector is a triple:
UV=MF, UIF, EF
Wherein:
MF is the professional factor (Major Factors), MF=mfi:i=1,2 ..., | MF |, by the major name of finite number
Lexical representation;
UIF be the user interest factor, UIF=uifi:i=1,2 ..., | UIF | }, by several feature lexical representations;
EF be experience factor (Experience Factors), EF=efi:i=1,2 ..., | EF | }, from the passing project of user
It extracts in experience, is indicated by the professional technicality of finite number;Characteristic item in user property vector, can be preparatory by user
Its weight is specified as needed.
8. the domain knowledge of Process-Oriented according to claim 1 extracts and method for pushing, which is characterized in that described to pass through
Knowledge score value method is calculated in user behavior variable are as follows:
It is calculated based on the scoring to knowledge that user behavior vector generates, the result that knowledge score value calculates is to correlativity calculation result
With corrective action, acts in knowledge sort and embodies,
The user behavior vector be based on specific knowledge is browsed by user, is downloaded, collecting, recommending data is calculated, respectively
It is indicated with FL, FD, FF, FR,
The formula that knowledge score value (Knowledge Score, KS) calculates are as follows:
Wherein, FDiThe downloading time that most knowledge is downloaded in the number and the A to Z of that the knowledge is downloaded is respectively represented with FD
Number, FLiThe browsing time that most knowledge is browsed in the number and the A to Z of that the knowledge is browsed, FF are respectively represented with FLi
The collection number that most knowledge is collected in the number and the A to Z of that the knowledge is collected, FR are respectively represented with FFiAnd FR
The recommendation number that most knowledge is recommended in the knowledge recommended number and the A to Z of is respectively represented, it can by above-mentioned definition
Know, Similarly,
The calculation method of the knowledge sort result are as follows:
On the basis of correlation calculations, knowledge score value calculated result intervenes sequence of the correlation calculations to knowledge, is corrected
Knowledge score afterwards is ranked up knowledge according to the sequence of score from high to low, and result is pushed to and completes a certain specific
The employee of business, to realize the push with flow tasks with the knowledge being closely connected of the present invention;Its calculation formula is:
KCS=β × Sim (KR, KV)+(1- β) × KS
Wherein, KCS (Knowledge Correction Score, KCS) represents the knowledge score after correction, and Sim (KR, KV) is
Correlativity calculation result, KS are knowledge score value calculated result, by summing after the two ranking operation, the knowledge after being corrected
Score KCS;β is the operator for adjusting correlativity calculation result and knowledge score value calculated result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610117829.9A CN105787072B (en) | 2013-11-04 | 2013-11-04 | A kind of domain knowledge of Process-Oriented extracts and method for pushing |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610117829.9A CN105787072B (en) | 2013-11-04 | 2013-11-04 | A kind of domain knowledge of Process-Oriented extracts and method for pushing |
CN201310542024.5A CN103617481B (en) | 2013-11-04 | 2013-11-04 | A kind of domain knowledge extraction and supplying system and the method for Process-Oriented |
Related Parent Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310542024.5A Division CN103617481B (en) | 2013-11-04 | 2013-11-04 | A kind of domain knowledge extraction and supplying system and the method for Process-Oriented |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105787072A CN105787072A (en) | 2016-07-20 |
CN105787072B true CN105787072B (en) | 2019-06-28 |
Family
ID=56387792
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610117829.9A Active CN105787072B (en) | 2013-11-04 | 2013-11-04 | A kind of domain knowledge of Process-Oriented extracts and method for pushing |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105787072B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106952051A (en) * | 2017-04-06 | 2017-07-14 | 北京仿真中心 | A kind of knowledge method for pushing based on design cycle |
CN110020134B (en) * | 2017-11-09 | 2021-08-13 | 北京国双科技有限公司 | Knowledge service information pushing method and system, storage medium and processor |
CN109684537A (en) * | 2018-10-29 | 2019-04-26 | 昆明理工大学 | A kind of the knowledge resource intelligently pushing system and its method for pushing of Business Process-oriented |
CN113721897A (en) * | 2021-08-03 | 2021-11-30 | 中国航空工业集团公司沈阳飞机设计研究所 | Modeling method based on OPM |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101026802A (en) * | 2007-03-16 | 2007-08-29 | 华为技术有限公司 | Information push method and device |
CN101458732A (en) * | 2008-12-25 | 2009-06-17 | 浙江师范大学 | Knowledge application method orienting ornament innovation design |
CN101515309A (en) * | 2009-04-07 | 2009-08-26 | 华中科技大学 | City emergency evacuation simulation system based on multi intelligent agent |
CN101894351A (en) * | 2010-08-09 | 2010-11-24 | 北京邮电大学 | Multi-agent based tour multimedia information personalized service system |
CN102063433A (en) * | 2009-11-16 | 2011-05-18 | 华为技术有限公司 | Method and device for recommending related items |
-
2013
- 2013-11-04 CN CN201610117829.9A patent/CN105787072B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101026802A (en) * | 2007-03-16 | 2007-08-29 | 华为技术有限公司 | Information push method and device |
CN101458732A (en) * | 2008-12-25 | 2009-06-17 | 浙江师范大学 | Knowledge application method orienting ornament innovation design |
CN101515309A (en) * | 2009-04-07 | 2009-08-26 | 华中科技大学 | City emergency evacuation simulation system based on multi intelligent agent |
CN102063433A (en) * | 2009-11-16 | 2011-05-18 | 华为技术有限公司 | Method and device for recommending related items |
CN101894351A (en) * | 2010-08-09 | 2010-11-24 | 北京邮电大学 | Multi-agent based tour multimedia information personalized service system |
Also Published As
Publication number | Publication date |
---|---|
CN105787072A (en) | 2016-07-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103617481B (en) | A kind of domain knowledge extraction and supplying system and the method for Process-Oriented | |
Padhi et al. | Quantifying potential tourist behavior in choice of destination using Google Trends | |
CN106250438B (en) | Zero-citation article recommendation method and system based on random walk model | |
KR101426765B1 (en) | System and method for supplying collaboration partner search service | |
CN105787072B (en) | A kind of domain knowledge of Process-Oriented extracts and method for pushing | |
Ozcan et al. | Human resources mining for examination of R&D progress and requirements | |
Ayvaz et al. | A scalable streaming big data architecture for real-time sentiment analysis | |
Song et al. | Enterprise knowledge recommendation approach based on context-aware of time-sequence relationship | |
CN104794010B (en) | Information exchange optimization method in a kind of system | |
Sangeetha et al. | Modelling of E-governance framework for mining knowledge from massive grievance redressal data | |
Scrivner et al. | XD Metrics on Demand Value Analytics: Visualizing the Impact of Internal Information Technology Investments on External Funding, Publications, and Collaboration Networks | |
CN110750572A (en) | Adaptive method and device for heuristic evaluation of scientific and technological achievements | |
Clever et al. | Towards auto-suggested process modeling–prototypical development of an auto-suggest component for process modeling tools | |
Anand | ETL and its impact on Business Intelligence | |
Olszak et al. | Big Data Approach to Analyzing the IT Job Market | |
Andriansyah et al. | The Application of Power Business Intelligence in Analyzing the Availability of Rental Units | |
Shiramatsu et al. | A goal matching service for facilitating public collaboration using linked open data | |
Avdeenko et al. | Modeling information space for decision-making in the interaction of higher education system with regional labor market | |
Hsu et al. | Similarity search over personal process description graph | |
Zhang et al. | Tourism Forecast Based on Web Search Data and Sentiment Analysis of Social Network | |
Chen | Application of big data technology in smart Tourism | |
Du et al. | Feasibility Analytical Mapping (FAM) for the Bidding Decision: A Graphic Bidding Decision Making Model Based on Multidimensional Scaling and Discriminant Analysis | |
El Seddawy et al. | A proposed data mining technique to improve decision support system in an uncertain situation | |
Salhi et al. | Sentiment Analysis Application on Twitter for E-reputation | |
Dussoye et al. | Sentiment analytics framework integrating Twitter and Odoo ERP |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |