CN110472191A - A kind of the service evaluation calculation method and device of dynamic self-adapting - Google Patents
A kind of the service evaluation calculation method and device of dynamic self-adapting Download PDFInfo
- Publication number
- CN110472191A CN110472191A CN201910591389.4A CN201910591389A CN110472191A CN 110472191 A CN110472191 A CN 110472191A CN 201910591389 A CN201910591389 A CN 201910591389A CN 110472191 A CN110472191 A CN 110472191A
- Authority
- CN
- China
- Prior art keywords
- scoring
- service
- dynamic window
- current
- dynamic
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0282—Rating or review of business operators or products
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Physics & Mathematics (AREA)
- Game Theory and Decision Science (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Educational Administration (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Complex Calculations (AREA)
Abstract
The invention discloses the service evaluation calculation methods and device of a kind of dynamic self-adapting, the dynamic window Q={ r including being cached with history scoringi‑n…,ri‑2,ri‑1, n is the current scoring number in the dynamic window Q, which comprises step S1: being directed to a service S, is receiving i-th of scoring riWhen, according to the current n history scoring r in the Qi‑n…,ri‑2,ri‑1, estimate current scoring f (i)；Step S2: the f (i) and the ri, determine the scoring r of the service SiIt is whether just；Step S3: according to the fairness of scoring, the range of the dynamic window Q is adjusted；Step S4: in the range of the dynamic window Q after the adjustment, the service evaluation value R of the service S is calculatedi.The present invention may be implemented to calculate the dynamic of service evaluation value, obtain more accurate and reliable service evaluation value, supplier or developer as service provide more effective and accurate feedback data, select service to provide more reliable reference and foundation for the user of service.
Description
Technical field
The present invention relates to service evaluation technical fields, more particularly to a kind of service evaluation calculation method of dynamic self-adapting
And a kind of service evaluation computing device of dynamic self-adapting.
Background technique
In the service systems such as e-commerce, the crowdsourcing of extensive Open Dynamic, service user, which is difficult to obtain, explicitly may be used
With the feature of service, service evaluation model is introduced into describe service characteristic.Jingdone district, Taobao, Amazon etc. at present
On the one hand the service system evaluation model that extensive e-commerce platform uses will receive the influence of injustice scoring, another party
Face timeliness is insufficient, that is, for the dynamic change of service, there is certain hysteresis quality.
Summary of the invention
The present invention provide a kind of dynamic self-adapting service evaluation calculation method and a kind of service evaluation of dynamic self-adapting
Computing device improves the validity to service evaluation to overcome above-mentioned technical problem.
To solve the above-mentioned problems, the invention discloses a kind of service evaluation calculation methods of dynamic self-adapting, including slow
There is the dynamic window Q={ r of history scoringi-n,…,ri-2,ri-1, n is the current scoring number in the dynamic window Q, institute
The method of stating includes:
Step S1: it is directed to a service S, is receiving i-th of scoring riWhen, it is scored according to the current n history in the Q
ri-n,…,ri-2,ri-1, estimate current scoring f (i)；
Step S2: the f (i) and the ri, determine the scoring r of the service SiIt is whether just；
Step S3: according to the fairness of scoring, the range of the dynamic window Q is adjusted；
Step S4: in the range of the dynamic window Q after the adjustment, the service evaluation value R of the service S is calculatedi。
Optionally, the step S1 includes:
According to the current n history scoring r in the dynamic window Qi-n,…,ri-2,ri-1, estimated using formula (1)-(5)
Calculate the current scoring f (i)；
Wherein:
The f (i) is expressed as the multinomial of scoring serial number, and calculation formula includes:
In (1) formula, M is order of a polynomial, wjIt is weight；
Using Bayesian model to the Q={ ri-n,…,ri-2,ri-1Be trained, the training step includes:
For the set r={ r of current n history scoring1,r2,…,rn, weight w=is found using Bayesian model
{w1,w2,…,wnMaximum-likelihood estimation, specific formula is as follows:
For meeting the prior distribution p being just distributed very much (w | α), lnp (w | r) meet following formula:
By EM algorithm, the calculated result of parameter alpha and β are as follows:
In (4), (5) formula, a and C be respectively indicate the output w of sample (w | r) about input feature vector r, model parameter α with
And the mean vector and covariance matrix of the probability density function p (W | r, α, β) of sample distribution variance β,It is polynomial basis letter
NumberEmpirical covariance matrix, Trace (C) be Matrix C mark.
Optionally, the step S2 includes:
Compare the f (i) and the riDifference whether meet | f (i)-ri|~N (0, σ2)；
If the difference meets, the r is determinediIt is just；
Otherwise, it is determined that the riIt is injustice.
Optionally, the step S2 further include:
The r is verified using hypothesis testing methodiIt is whether just；
Wherein, the verification step includes:
The threshold value of hypothesis testing is set as τ, the τ calculation formula is as follows:
τ=σ Θ-1(x)(δ/2) (6)；
Wherein, Θ (x)=1- Φ (x), Φ (x) are the cumulative distribution function of (0,1) standardized normal distribution N, and δ is confidence
Value；
Compare described | f (i)-ri| with the τ；
If | f (i)-ri|≤τ, the then f (i) and the riDifference in confidence interval, inspection result be the ri
" just scoring "；
If | f (i)-ri| > τ, the then f (i) and the riDifference not in confidence interval, inspection result be the ri
" doubtful injustice scoring ", and by the riIt is put into buffer area H.
Optionally, if the inspection result is the ri" just scoring ", the step S3 include:
By the riThe dynamic window Q is written, the dynamic window Q is updated to { ri-n,…,ri-2,ri-1,ri}。
Optionally, if the inspection result is the ri" doubtful injustice scoring ", the step S3 includes:
If judging the scoring number for servicing S variation for K；
By the scoring number L of current cache area H compared with the K；
If 1) L < K, confirms that " the doubtful injustice scoring " that caches in the H is " inequity scoring ", empties the H,
Keep the Q constant；
If 2) L=K, determine service S change cause continuously score variation, confirm cached in the H it is " doubtful
Inequity scoring " is " just scoring ", empties dynamic window Q, and dynamic window Q is written in all scorings in the H.
Optionally, in the step S4, the RiCalculation formula include:
Ri=ρ f (t)+(1- ρ) ri(7)；
Wherein, n ' is the current scoring number in the dynamic window Q adjusted.
Optionally, the RiIncluding following situations:
(A) RiFor constant, the opinion rating that expression service S is maintained at high or low is constant；
(B) RiBe it is linear, indicate service S service evaluation with input Grading sequence be linearly raised and lowered；
(C) RiFor segmented, indicate that the variation of service S causes service evaluation substantially to change；
(D) RiFor fluctuation, indicate the service evaluation of service S in interior fluctuation by a small margin.
Optionally, the method also includes:
The service evaluation value R of the service S is measured using mean absolute error MAEiError rate；
Wherein, described
In (9) formula, N is the evaluation number of the service S.
To solve the above-mentioned problems, the invention also discloses a kind of service evaluation computing devices of dynamic self-adapting, including
It is cached with the dynamic window Q={ r of history scoringi-n,…,ri-2,ri-1, n is the current scoring number in the dynamic window Q,
Described device includes:
The estimation block that scores is receiving i-th of scoring r for being directed to a service SiWhen, according to the current n in the Q
A history scoring ri-n,…,ri-2,ri-1, estimate current scoring f (i)；
Score determination module, for the f (i) and the ri, determine the scoring r of the service SiIt is whether just；
Window adjustment module adjusts the range of the dynamic window Q for the fairness according to scoring；
Computing module is evaluated, in the range of the dynamic window Q after the adjustment, calculates the service of the service S
Evaluation of estimate Ri。
Compared with prior art, the present invention includes the following advantages:
The present invention judges the fairness of the scoring, after obtaining current scoring according to scoring for a service
Fairness, adjustment is cached with the dynamic window of history scoring, the then range of the dynamic window Q after the adjustment in real time
It is interior, calculate the service evaluation value R of the service Si,；Calculation method through the invention may be implemented to move service evaluation value
State calculates, and obtains the service and more accurately provides more with reliable service evaluation value, supplier or developer as service
Effective and accurate feedback data, so that the supplier or developer of service accurately understand the advantage and disadvantage entirely serviced, to service S
It more efficiently improves, searches loophole, optimization service；It is also provided simultaneously for user's selection service of service more reliable
With reference to and foundation.
Detailed description of the invention
Fig. 1 is a kind of step flow chart of the service evaluation calculation method of dynamic self-adapting of the embodiment of the present invention；
Fig. 2 is the experimental result comparison diagram of the embodiment of the present invention Yu average algorithm and OACR algorithm；
Fig. 3 is a kind of structural schematic diagram of the service evaluation computing device of dynamic self-adapting of the embodiment of the present invention；
Fig. 4 be the embodiment of the present invention judge scoring whether just flow diagram.
Specific embodiment
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, with reference to the accompanying drawing and specific real
Applying mode, the present invention is described in further detail.
Referring to Fig.1, a kind of step process of the service evaluation calculation method of dynamic self-adapting of the embodiment of the present invention is shown
Figure, the dynamic window Q={ r including being cached with history scoringi-n,…,ri-2,ri-1, n is currently commenting in the dynamic window Q
Divide number, the method may include following steps:
Step S1: it is directed to a service S, is receiving i-th of scoring riWhen, it is scored according to the current n history in the Q
ri-n,…,ri-2,ri-1, estimate current scoring f (i)；
Step S2: the f (i) and the ri, determine the scoring r of the service SiIt is whether just；
Step S3: according to the fairness of scoring, the range of the dynamic window Q is adjusted；
Step S4: in the range of the dynamic window Q after the adjustment, the service evaluation value R of the service S is calculatedi。
It calculates in service under cloud computing application mode, all kinds of heterogeneous resources being distributed on internet can be all encapsulated as
Web services use for outside.Service is the fundamental as building application system, is a kind of towards the basic of business entities
Service, basic function service and tool-class service with minimum functional unit, usually have characteristics that 1) affairs are complete
Property；2) reusability；3) stability；4) validity of system development and maintenance.In embodiments of the present invention, service S can refer to
The extensive e-commerce platform of Jingdone district, Taobao, Amazon etc. is the function services that user provides, for example commodity are swept and swept
Service, goods links sharing service, shopping group establish sharing service etc., can also refer to the business that certain crowdsourcing service platforms provide
Service, such as Ali's cloud storage service etc..Since the type and the criteria for classifying of service S are not the emphasis of the embodiment of the present invention, In
This is not repeated excessively, can form a dynamic Service, and service evaluation calculation method of the invention can be used and evaluated.
Scoring can refer to that user scores to the usage experience sense of entire service S, can be taken by e-commerce platform or crowdsourcing
Product feedback interface, user experience interface or the product downloading evaluation interface that business platform is released are collected.
It is the personal behavior of user due to scoring, there is very big subjectivity and uncontrollable factor, there are injustice scorings
Possibility, cause one service S service characteristic cannot relatively accurately be reflected so that service supplier or developer without
Method accurately understands the advantage and disadvantage of entire service S, can not more efficiently improve to service S, presence service optimizes loophole.Together
When, current service is usually dynamic change, has certain hysteresis quality when being supplied to user, also leads to the impression of each user
It is inconsistent.Therefore, for a service S, obtaining the service is more effectively particularly important with accurate user scoring.
In view of the above-mentioned problems, the embodiment of the present invention proposes a kind of service evaluation calculation method of dynamic self-adapting, can obtain
Service S is more accurate and reliable service evaluation value, provided for the supplier of service or developer effectively and accurately anti-
Present data.Specific calculation method is referring to step S1~step S4.
Firstly, the embodiment of the present invention devises a dynamic window Q, it include current n history scoring in dynamic window
ri-n,…,ri-2,ri-1, n is the current scoring number in the dynamic window Q, for example n is 100 or 50, refers to current dynamic window
Inside there are 100 or 50 scorings.By the setting of dynamic window, some abnormal or malice scorings can be removed, to score data
Effective selection is carried out, to improve service evaluation value RiThe validity and reliability of calculating.Specifically, scoring for history, pass through
Weight and EM algorithm estimate a current scoring, so as to the scoring r that just receivesiIt is compared.In step
In S1, estimate that the concrete methods of realizing of current scoring f (i) is as follows:
According to the current n history scoring r in the dynamic window Qi-n,…,ri-2,ri-1, estimated using formula (1)-(5)
Calculate the current scoring f (i)；
Wherein:
The f (i) is expressed as the multinomial of scoring serial number, and calculation formula includes:
In (1) formula, M is order of a polynomial, wjIt is weight；
Using Bayesian model to the Q={ ri-n,…,ri-2,ri-1Be trained, the training step includes:
For the set r={ r of current n history scoring1,r2,…,rn, weight w=is found using Bayesian model
{w1,w2,…,wnMaximum-likelihood estimation, specific formula is as follows:
For meeting the prior distribution p being just distributed very much (w | α), lnp (w | r) meet following formula:
By EM algorithm, the calculated result of parameter alpha and β are as follows:
In (4), (5) formula, a and C be respectively indicate the output w of sample (w | r) about input feature vector r, model parameter α with
And the mean vector and covariance matrix of the probability density function p (w | r, α, β) of sample distribution variance β,It is polynomial basis letter
NumberEmpirical covariance matrix, Trace (C) be Matrix C mark.
Secondly, the f (i) and the ri, determine the riIt is whether just.Referring to Fig. 4, implementation of the present invention is shown
Whether just flow diagram judges whether scoring is just including tentatively judging and testing in step s 2 for the scoring that judges of example
The process of card.
In embodiments of the present invention, the process tentatively judged is as follows:
Compare the f (i) and the riDifference whether meet | f (i)-ri|~N (0, σ2)；
If the difference meets, the r is determinediIt is just；
Otherwise, it is determined that the riIt is injustice.
Usual most users should be similar for the usage experience of the same service, it is, everybody is universal
It will be considered that a service " good " or " bad ", give service scoring thus, it is supposed that user can be just, then service is commented
Sub-sequence should meet normal distribution.In addition, small part user may be different to service view, provide very big with public difference
Scoring, alternatively, there is malice brush the phenomenon that dividing, this partial evaluation does not meet the normal distribution of just Grading sequence.Therefore, it adopts
Screening and judgement can be carried out to malice brush point with the preliminary judgement judgment method of the embodiment of the present invention, further increase the present invention
The fairness and validity of embodiment service evaluation.
For specific implementation, according to existing scoring, estimation meets the scoring f (i) of normal distribution 1, calculating it is just
The difference of scoring ri and estimation scoring f (i), difference should meet another normal distribution 2, and the mean value of normal distribution 2 should
It is 0.
Next, verifying the r using hypothesis testing methodiScoring it is whether just.It is described to be come using hypothesis testing method
Verify the riScoring whether just step can specifically include:
The threshold value of hypothesis testing is set as τ, the τ calculation formula is as follows:
τ=σ Θ-1(x)(δ/2) (6)；
Wherein, Θ (x)=1- Φ (x), Φ (x) are the cumulative distribution function of (0,1) standardized normal distribution N, and δ is confidence
Value；
Compare described | f (i)-ri| with the τ；
If | f (i)-ri|≤τ, the then f (i) and the riDifference in confidence interval, inspection result be the ri
" just scoring "；
If | f (i)-ri| > τ, the then f (i) and the riDifference not in confidence interval, inspection result be the ri
" doubtful injustice scoring ", and by the riIt is put into buffer area H.
Then, according to the fairness of scoring, i.e. inspection result, the range of the dynamic window Q is adjusted.It can specifically include
Several situations below:
Situation one: if the inspection result is the ri" just scoring ", then directly by the riThe dynamic window is written
Q, the dynamic window Q are updated to { ri-n,…,ri-2,ri-1,ri}。
Situation two: if the inspection result is the ri" doubtful injustice scoring ", then further sentence in the following ways
It is disconnected:
If judging the scoring number for servicing S variation for K；
By the scoring number L of current cache area H compared with the K；
If 1) L < K, confirms that " the doubtful injustice scoring " that caches in the H is " inequity scoring ", empties the H,
Keep the Q constant, i.e. maintenance Q={ ri-n,…,ri-2,ri-1}；
If 2) L=K, determine service S change cause continuously score variation, confirm cached in the H it is " doubtful
Inequity scoring " is " just scoring ", empties dynamic window Q, and dynamic window Q is written in all scorings in the H.
Finally, executing step S4, the service evaluation value Ri of service S is calculated using formula (7) (8)；Wherein: RiCalculating it is public
Formula includes:
Ri=ρ f (t)+(1- ρ) ri(7)；
Wherein, n ' is the current scoring number in the dynamic window Q adjusted.
Further, in order to more clearly explain that the embodiment of the present invention can follow the dynamic change of service, service is realized
The dynamic of evaluation calculates, and the supplier or developer as service provide more effective and accurate feedback data, below with Ah
In for cloud storage service, a kind of service evaluation calculation method of dynamic self-adapting of the embodiment of the present invention is illustrated.
Assuming that some cloud storage service, scoring window before this is { 8,7.9,8.1,8,7.5,7.8,8,7.7 }, setting
Parameter is as follows:
1) just it is distributed very much N (0, σ2) in, μ=0, σ=6.5；
2) in formula (3), α=0.005, for meeting the scoring being just distributed very much,
3) in formula (6), δ=5；
4) contingent window full-size K=10；
It is { 8,7.9,8.1,8,7.5,7.8,8,7.7 } and above-mentioned parameter according to scoring window before this, estimation is currently gone through
Commentary on historical events or historical records point, and calculate and examine threshold tau.
The case where current user's scoring that arrives, which scores, may be as follows:
1) scoring is 7.6, makees absolute value of the difference compared with τ with the scoring of current history for 7.6, and 7.6, which have passed through scoring, examines,
Then directly by 7.6 write-in scoring windows, window becomes { 8,7.9,8.1,8,7.5,7.8,8,7.7,7.6 }, in the range,
New service evaluation value is calculated by formula (7) (8).
2) scoring is 3, this scoring, which is examined, not to be passed through, and is first put into buffer area for 3, scoring window be still 8,7.9,8.1,
8,7.5,7.8,8,7.7 }, service evaluation value is calculated according to the scoring of estimation, then the scoring next to arrive is judged.
If 2.1) the Y user scoring next newly arrived, (Y < K at this time is arranged in experiment, K=10), scoring inspection is not
Pass through, is then first put into buffer area, scoring window is still { 8,7.9,8.1,8,7.5,7.8,8,7.7 }, according to the scoring of estimation
Calculate service evaluation value；
If the Y+1 scoring is 8.1, scoring is passed through and has examined, then, Y scoring in buffer area is all judged as injustice
Positive scoring, is deleted；Window becomes { 8,7.9,8.1,8,7.5,7.8,8,7.7,8.1 }, in the range, passes through formula (7) (8)
Calculate new service evaluation value.
If 2.2) next newly arrive the Y scoring, (Y=K at this time, experiment in K=10 is arranged), scoring inspection is still not
Pass through, then, it is believed that the cloud service has occurred and that variation, and window reforms into this 10 scorings, in this 10 scoring ranges, passes through
Formula (7) (8) calculates new service evaluation value.
In embodiments of the present invention, the calculated result of service evaluation value Ri includes following several situations:
(A) RiFor constant, the opinion rating that expression service S is maintained at high or low is constant；
(B) RiBe it is linear, indicate service S service evaluation with input Grading sequence be linearly raised and lowered；
(C) RiFor segmented, indicate that the variation of service S causes service evaluation substantially to change；
(D) RiFor fluctuation, indicate that the service evaluation of service S in interior fluctuation by a small margin, generallys use SIN function mould
It is quasi-.
In addition, the embodiment of the present invention also proposed the method measured to the error rate of the service evaluation value Ri of service S,
Specifically, measuring the service evaluation value R of the service S preferably by mean absolute error MAEiError rate；
Wherein, described
In (9) formula, N is the evaluation number of the service S.
The embodiment of the present invention is measured by the error rate to service evaluation value Ri, can find service of the present invention in time
Loophole existing for evaluation calculation method and the device manufactured according to the method.
In order to verify the embodiment of the present invention calculation method reliability, the embodiment of the present invention also with existing service evaluation
Algorithm is compared, wherein existing comparison algorithm includes average algorithm and OACR algorithm.Average algorithm service calculated is commented
Value is the arithmetic mean of instantaneous value of all user's scorings in window.OACR algorithm is a kind of Two Phase Method, uses changeable weight first
Formula estimates latest services evaluation, then removes injustice use with based on " olfactory fatigue phenomenon " derivative " olfaction reaction formula "
Family scoring bring influences.
Specific control experiment is as follows:
Firstly, inputting 150 clothes respectively into the calculation method and average algorithm of the embodiment of the present invention and OACR algorithm
Business evaluation sequence.150 meet the user's Grading sequence being just distributed very much, wherein the method for the injustice scoring of simulation is, by section
[20,26) scoring be changed to r=r-20, section [110,130) scoring be changed to r=r+20.It calculates and takes by respective calculation
Business evaluation of estimate, measures the error rate of three kinds of calculation methods further according to formula (9).The embodiment of the present invention and average algorithm and
The experimental result comparison diagram of OACR algorithm as shown in Fig. 2, from figure it can be seen that, dynamic self-adapting service evaluation meter proposed by the present invention
Calculation method mean error is minimum, and effect is best.
It should be noted that for simple description, therefore, it is stated as a series of action groups for embodiment of the method
It closes, but those skilled in the art should understand that, embodiment of that present invention are not limited by the describe sequence of actions, because according to
According to the embodiment of the present invention, some steps may be performed in other sequences or simultaneously.Secondly, those skilled in the art also should
Know, the embodiments described in the specification are all preferred embodiments, and the related movement not necessarily present invention is implemented
Necessary to example.
Referring to Fig. 3, a kind of structural representation of the service evaluation computing device of dynamic self-adapting of the embodiment of the present invention is shown
Figure, the dynamic window Q={ r including being cached with history scoringi-n,…,ri-2,ri-1, n is currently commenting in the dynamic window Q
Divide number, the apparatus may include following modules:
The estimation block 301 that scores is receiving i-th of scoring r for being directed to a service SiWhen, according to working as in the Q
Preceding n history scoring ri-n,…,ri-2,ri-1, estimate current scoring f (i)；
Score determination module, for the f (i) and the ri, determine the scoring r of the service SiIt is whether just；
Window adjustment module 302 adjusts the range of the dynamic window Q for the fairness according to scoring；
Computing module 303 is evaluated, in the range of the dynamic window Q after the adjustment, calculates the clothes of the service S
Be engaged in evaluation of estimate Ri。
For device embodiment, since it is basically similar to the method embodiment, related so being described relatively simple
Place illustrates referring to the part of embodiment of the method.
The embodiment of the invention also provides a kind of devices, comprising:
One or more processors；With
One or more machine readable medias of instruction are stored thereon with, are executed when by one or more of processors
When, so that described device executes a kind of service evaluation meter of one or more dynamic self-adaptings as described in the embodiments of the present invention
Calculation method.
The embodiment of the invention also provides the computer programs of a kind of computer readable storage medium, storage to handle
Device executes a kind of service evaluation calculation method of dynamic self-adapting as described in the embodiments of the present invention.
All the embodiments in this specification are described in a progressive manner, the highlights of each of the examples are with
The difference of other embodiments, the same or similar parts between the embodiments can be referred to each other.
Above to a kind of the service evaluation calculation method and device of dynamic self-adapting provided by the present invention, carry out in detail
It introduces, used herein a specific example illustrates the principle and implementation of the invention, the explanation of above embodiments
It is merely used to help understand method and its core concept of the invention；At the same time, for those skilled in the art, according to this
The thought of invention, there will be changes in the specific implementation manner and application range, in conclusion the content of the present specification is not answered
It is interpreted as limitation of the present invention.
Claims (10)
1. a kind of service evaluation calculation method of dynamic self-adapting, which is characterized in that the dynamic window including being cached with history scoring
Mouth Q={ ri-n..., ri-2, ri-1, n is the current scoring number in the dynamic window Q, which comprises
Step S1: it is directed to a service S, is receiving i-th of scoring riWhen, it is scored according to the current n history in the Q
ri-n..., ri-2, ri-1, estimate current scoring f (i)；
Step S2: the f (i) and the ri, determine the scoring r of the service SiIt is whether just；
Step S3: according to the fairness of scoring, the range of the dynamic window Q is adjusted；
Step S4: in the range of the dynamic window Q after the adjustment, the service evaluation value R of the service S is calculatedi。
2. the method according to claim 1, wherein the step S1 includes:
According to the current n history scoring r in the dynamic window Qi-n..., ri-2, ri-1, institute is estimated using formula (1)-(5)
State current scoring f (i)；
Wherein:
The f (i) is expressed as the multinomial of scoring serial number, and calculation formula includes:
In (1) formula, M is order of a polynomial, wjIt is weight；
Using Bayesian model to the Q={ ri-n..., ri-2, ri-1Be trained, the training step includes:
For the set r={ r of current n history scoring1, r2..., rn, weight w={ w is found using Bayesian model1,
w2..., wnMaximum-likelihood estimation, specific formula is as follows:
For meeting the prior distribution p being just distributed very much (w | α), ln p (w | r) meet following formula:
By EM algorithm, the calculated result of parameter alpha and β are as follows:
In (4), (5) formula, a and C are to indicate the output w of sample (w | r) about input feature vector r, model parameter α and sample respectively
The mean vector and covariance matrix of the probability density function p (w | r, α, β) of this distribution variance β,It is polynomial basis functionEmpirical covariance matrix, Trace (C) be Matrix C mark.
3. the method according to claim 1, wherein the step S2 includes:
Compare the f (i) and the riDifference whether meet | f (i)-ri|~N (0, σ2)；
If the difference meets, the r is determinediIt is just；
Otherwise, it is determined that the riIt is injustice.
4. according to the method described in claim 3, it is characterized in that, the step S2 further include:
The r is verified using hypothesis testing methodiIt is whether just；
Wherein, the verification step includes:
The threshold value of hypothesis testing is set as τ, the τ calculation formula is as follows:
τ=σ Θ-1(x)(δ/2) (6)；
Wherein, Θ (x)=1- Φ (x), Φ (x) are the cumulative distribution function of (0,1) standardized normal distribution N, and δ is the value of the confidence；
Compare described | f (i)-ri| with the τ；
If | f (i)-ri|≤τ, the then f (i) and the riDifference in confidence interval, inspection result be the riIt is " just
Scoring "；
If | f (i)-ri| > τ, the then f (i) and the riDifference not in confidence interval, inspection result be the riIt " doubts
Score like inequity ", and by the riIt is put into buffer area H.
5. according to the method described in claim 4, it is characterized in that, if the inspection result is the ri" just scoring ", it is described
Step S3 includes:
By the riThe dynamic window Q is written, the dynamic window Q is updated to { ri-n..., ri-2, ri-1, ri}。
6. according to the method described in claim 4, it is characterized in that, if the inspection result is the ri" doubtful inequity is commented
Point ", the step S3 includes:
If judging the scoring number for servicing S variation for K；
By the scoring number L of current cache area H compared with the K；
If 1) L < K, confirms that " the doubtful injustice scoring " that caches in the H is " inequity scoring ", empty the H, keep
The Q is constant；
If 2) L=K, determines to service S and change to cause continuously to score to change, confirm " the doubtful injustice cached in the H
Positive scoring " is " just scoring ", empties dynamic window Q, and dynamic window Q is written in all scorings in the H.
7. method according to claim 1 or 6, which is characterized in that in the step S4, the RiCalculation formula packet
It includes:
Ri=ρ f (t)+(1- ρ) ri(7)；
Wherein, n ' is the current scoring number in the dynamic window Q adjusted.
8. the method according to claim 1, wherein the RiIncluding following situations:
(A) RiFor constant, the opinion rating that expression service S is maintained at high or low is constant；
(B) RiBe it is linear, indicate service S service evaluation with input Grading sequence be linearly raised and lowered；
(C) RiFor segmented, indicate that the variation of service S causes service evaluation substantially to change；
(D) RiFor fluctuation, indicate the service evaluation of service S in interior fluctuation by a small margin.
9. the method according to claim 1, wherein the method also includes:
The service evaluation value R of the service S is measured using mean absolute error MAEiError rate；
Wherein, described
In (9) formula, N is the evaluation number of the service S.
10. a kind of service evaluation computing device of dynamic self-adapting, which is characterized in that the dynamic window including being cached with history scoring
Mouth Q={ ri-n..., ri-2, ri-1, n is the current scoring number in the dynamic window Q, and described device includes:
The estimation block that scores is receiving i-th of scoring r for being directed to a service SiWhen, it is gone through according to current n in the Q
Commentary on historical events or historical records divide ri-n..., ri-2, ri-1, estimate current scoring f (i)；
Score determination module, for the f (i) and the ri, determine the scoring r of the service SiIt is whether just；
Window adjustment module adjusts the range of the dynamic window Q for the fairness according to scoring；
Computing module is evaluated, in the range of the dynamic window Q after the adjustment, calculates the service evaluation of the service S
Value Ri。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910591389.4A CN110472191B (en) | 2019-07-02 | 2019-07-02 | Dynamic self-adaptive service evaluation calculation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910591389.4A CN110472191B (en) | 2019-07-02 | 2019-07-02 | Dynamic self-adaptive service evaluation calculation method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110472191A true CN110472191A (en) | 2019-11-19 |
| CN110472191B CN110472191B (en) | 2021-03-12 |
Family
ID=68507042
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910591389.4A Active CN110472191B (en) | 2019-07-02 | 2019-07-02 | Dynamic self-adaptive service evaluation calculation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110472191B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117196734A (en) * | 2023-09-14 | 2023-12-08 | 长沙理工大学 | Value evaluation method, system, equipment and storage medium for crowdsourcing task |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1945644A (en) * | 2006-10-13 | 2007-04-11 | 福建师范大学 | Multilayer grading data processing system and method for navigation mark remote measuring terminal |
| CN101296128A (en) * | 2007-04-24 | 2008-10-29 | 北京大学 | Method for monitoring abnormal state of internet information |
| US7655895B2 (en) * | 1992-05-05 | 2010-02-02 | Automotive Technologies International, Inc. | Vehicle-mounted monitoring arrangement and method using light-regulation |
| CN101908065A (en) * | 2010-07-27 | 2010-12-08 | 浙江大学 | On-line attribute abnormal point detecting method for supporting dynamic update |
| US20110099294A1 (en) * | 2009-10-23 | 2011-04-28 | Ajay Kapur | Dynamically rehosting web content |
| CN102685145A (en) * | 2012-05-28 | 2012-09-19 | 西安交通大学 | Domain name server (DNS) data packet-based bot-net domain name discovery method |
| CN104463548A (en) * | 2014-12-25 | 2015-03-25 | 南京大学 | Carriage quantitative selection method influenced by multiple factors |
| CN107092582A (en) * | 2017-03-31 | 2017-08-25 | 江苏方天电力技术有限公司 | One kind is based on the posterior exceptional value on-line checking of residual error and method for evaluating confidence |
| CN107944602A (en) * | 2017-11-09 | 2018-04-20 | 重庆大学 | The evaluation of cloud manufacturing service and matching process based on trust model |
| CN108881194A (en) * | 2018-06-07 | 2018-11-23 | 郑州信大先进技术研究院 | Enterprises user anomaly detection method and device |
| CN109816196A (en) * | 2018-12-04 | 2019-05-28 | 平安科技（深圳）有限公司 | Evaluation value calculation method, device, equipment and the readable storage medium storing program for executing of prediction model |
-
2019
- 2019-07-02 CN CN201910591389.4A patent/CN110472191B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7655895B2 (en) * | 1992-05-05 | 2010-02-02 | Automotive Technologies International, Inc. | Vehicle-mounted monitoring arrangement and method using light-regulation |
| CN1945644A (en) * | 2006-10-13 | 2007-04-11 | 福建师范大学 | Multilayer grading data processing system and method for navigation mark remote measuring terminal |
| CN101296128A (en) * | 2007-04-24 | 2008-10-29 | 北京大学 | Method for monitoring abnormal state of internet information |
| US20110099294A1 (en) * | 2009-10-23 | 2011-04-28 | Ajay Kapur | Dynamically rehosting web content |
| CN101908065A (en) * | 2010-07-27 | 2010-12-08 | 浙江大学 | On-line attribute abnormal point detecting method for supporting dynamic update |
| CN102685145A (en) * | 2012-05-28 | 2012-09-19 | 西安交通大学 | Domain name server (DNS) data packet-based bot-net domain name discovery method |
| CN104463548A (en) * | 2014-12-25 | 2015-03-25 | 南京大学 | Carriage quantitative selection method influenced by multiple factors |
| CN107092582A (en) * | 2017-03-31 | 2017-08-25 | 江苏方天电力技术有限公司 | One kind is based on the posterior exceptional value on-line checking of residual error and method for evaluating confidence |
| CN107944602A (en) * | 2017-11-09 | 2018-04-20 | 重庆大学 | The evaluation of cloud manufacturing service and matching process based on trust model |
| CN108881194A (en) * | 2018-06-07 | 2018-11-23 | 郑州信大先进技术研究院 | Enterprises user anomaly detection method and device |
| CN109816196A (en) * | 2018-12-04 | 2019-05-28 | 平安科技（深圳）有限公司 | Evaluation value calculation method, device, equipment and the readable storage medium storing program for executing of prediction model |
Non-Patent Citations (1)
| Title |
|---|
| 冯耀东 等: "一种自适应软件体系结构建模及其实施方法", 《北京大学学报（自然科学版）》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117196734A (en) * | 2023-09-14 | 2023-12-08 | 长沙理工大学 | Value evaluation method, system, equipment and storage medium for crowdsourcing task |
| CN117196734B (en) * | 2023-09-14 | 2024-03-22 | 长沙理工大学 | Value evaluation method, system, equipment and storage medium for crowdsourcing task |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110472191B (en) | 2021-03-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20110090811A1 (en) | Available Bandwidth Estimation in a Packet-Switched Communication Network | |
| CN104539601B (en) | Dynamic network attack process analysis method for reliability and system | |
| CN108683564B (en) | Network simulation system reliability evaluation method based on multidimensional decision attributes | |
| Dong et al. | Round trip time prediction using recurrent neural networks with minimal gated unit | |
| CN101404591B (en) | Self-adapting dynamic trust weight estimation method | |
| CN113537788B (en) | Urban traffic jam recognition method based on virus propagation theory | |
| CN112541639A (en) | Recommendation system scoring prediction method based on graph neural network and attention mechanism | |
| CN109961172A (en) | A kind of CPS rare event probability forecasting method examined based on statistical model | |
| CN110083518B (en) | AdaBoost-Elman-based virtual machine software aging prediction method | |
| CN109766281A (en) | A kind of imperfect debugging software reliability model of fault detection rate decline variation | |
| CN114301935A (en) | Reputation-based method for selecting edge cloud collaborative federated learning nodes of Internet of things | |
| CN114679310A (en) | Network information security detection method | |
| CN116781343A (en) | Terminal credibility assessment method, device, system, equipment and medium | |
| CN117041017A (en) | Intelligent operation and maintenance management method and system for data center | |
| CN115567405A (en) | Network flow gray prediction method based on self-adaptive feedback regulation mechanism | |
| CN110472191A (en) | A kind of the service evaluation calculation method and device of dynamic self-adapting | |
| Yoo et al. | Control Channel Isolation in SDN Virtualization: A Machine Learning Approach | |
| CN113887748A (en) | Online federal learning task allocation method and device, and federal learning method and system | |
| Bektemyssova et al. | Time series forecasting by the arima method | |
| CN108647897A (en) | A kind of method and system of product reliability analysis | |
| CN116401372A (en) | Knowledge graph representation learning method and device, electronic equipment and readable storage medium | |
| CN114184211B (en) | Method for judging consistency of performance change mechanism in inertial navigation reliability test | |
| Lian et al. | Sliding mode control for non‐linear networked control systems subject to packet disordering via prediction method | |
| CN114925750A (en) | Information recommendation method and device, computer readable storage medium and electronic equipment | |
| Tu et al. | Internet time-delay prediction based on unbiased grey model |
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 |