CN109784722A - Web service selection method and system based on user preference - Google Patents
Web service selection method and system based on user preference Download PDFInfo
- Publication number
- CN109784722A CN109784722A CN201910035858.4A CN201910035858A CN109784722A CN 109784722 A CN109784722 A CN 109784722A CN 201910035858 A CN201910035858 A CN 201910035858A CN 109784722 A CN109784722 A CN 109784722A
- Authority
- CN
- China
- Prior art keywords
- attribute
- weight
- qos
- user
- web service
- 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.)
- Pending
Links
Abstract
The present disclosure proposes web service selection methods and system based on user preference, improve order relation analytic approach according to user to QoS attribute fancy grade for several service quality QoS attributes by fuzzy number and determine subjective weight;The objective weight of QoS attribute is determined using entropy assessment, subjective weight is corrected in realization by Objective Weight;By the way of combination weighting, comprehensively considers subjective weights and Objective Weight finds out comprehensive QoS weight;Specific service is selected by COSINE similarity improved TOPSIS method on the basis of considering user preference.The disclosure improves the accuracy of service and decision-making, and the service selected is made to be more in line with the demand of user.
Description
Technical field
This disclosure relates to which Internet technical field, more particularly to the web service selection method based on user preference and is
System.
Background technique
On the internet with a large amount of Web service deployment functionally identical or similar, QoS becomes to distinguish these Web clothes
The key of business.However different users is often different for the significance level of QoS attribute, some users think price more
It is important, and some users then think that the response time is more important.So this is just needed in services selection, consider that user's is inclined
It is good.Weight is the scale for reflecting a criterion significance level, and the bigger criterion of weight illustrates that the significance level of the criterion is got over
Height more influences the result of decision.However, people are easier to provide " liking " to QoS attribute in real life, " do not like
Vigorously ", information as " price is more important than the response time " is difficult accurately provide the weight of each attribute, Ta Menwu
Method expresses the preference of oneself with a quantitative numerical value.
Existing research, the method for determining weight are roughly divided into two kinds: 1. subjective weights, 2. Objective Weights.Subjective weighting method
According to the hobby of user, each attribute is compared and carries out tax power.There are commonly AHP method, expert survey (Delphi method),
G1 method etc..Objective weighted model determines attribute weight by comparing different QoS attribute value information content and variation, there is Geordie system
Number method, entropy assessment etc..
Currently, the services selection based on service preferences has had some research achievements.Cloud model is by Chinese Academy of Engineering
What academician professor Li Deyi proposed.It is the uncertain transformation model for handling qualitativing concept and quantitative description.Document " Xie Hui, Li Song
Month, Sun Yonghe, Han Wei based on cloud model solve attribute weight DEMATEL technique study [J/OL] computer engineering with answer
With: author determines the subjective information of user by the method using cloud model in 1-7 ", is determined by qualitative be converted into of cloud model
The mode of amount expresses uncertain description of the user to preference very well.But completely from the angle of user, subjectivity is too strong,
Ignore existing association between QoS attribute value.
A kind of document " uncertain weights calculation method [J] software of Web service QoS attribute of Fan Zhiqiang, Li Ning, Hao Bo
Engineering, 2016,19 (07): 14-17. " indicates user's subjective uncertainty by the numerical characteristic of cloud model, and by with visitor
Sight weight, which combines, seeks comprehensive weight.But the method that the disclosure obtains subjective weight is needed similar to AHP method by QoS weight
The property wanted is done to be compared two-by-two, and consistency hardly results in guarantee." Sun Ruonan, Zhang Bin, Liu Tingting are a kind of to use improvement entropy weight to document
The small-sized microcomputer system of Web service quality evaluation sort method [J] of TOPSIS, 2017,38 (06): 1221-1226. "
The objective weight of each attribute of QoS is calculated by entropy assessment, but the subjective preferences of user is considered less.Document " grandson
The elegant court of a feudal ruler, button is pretty, a kind of Web service selection strategy [J] computer application research based on Evaluation formula of Gong Qingbo, Li Zhen,
2017,34 (08): 2408-2411+2428 " and document " web service composition method research of the Duan Jingshan based on user preference
[D] Southwest University solves uncertain problem by fuzzy logic in 2014. ".Pass through fuzzy AHP (analytic hierarchy process (AHP)) method
Come the user's erratic demand determined.The advantages that AHP method is practical because simple, structural is widely used.And AHP method
Qualitative question can be converted to quantitative problem, user is converted to quantitative mistake to the entitled process of attribute is in fact exactly qualitative
Journey.But AHP method needs to guarantee judgment matrix approach, increasing with order in practice, judgment matrix approach is often
It is difficult to ensure that.Although Fuzzy AHP solution user bring uncertain problem due to subjective factor, still needs
Consistency check is carried out to judgment matrix.
In conclusion people often indicate user to QoS (Quality of Service) attribute with weight at this stage
Preference, however due to the subjective judgement of user and to the ambiguity of preference description, the weight that traditional tax power method is found out is difficult standard
Really represent the preference of user.The main purpose of technical scheme is that the description of QoS attribute bias is fuzzy to ask in order to solve
Topic and the accuracy for improving services selection.
Summary of the invention
In order to solve the deficiencies in the prior art, embodiment of the disclosure provides the choosing of the Web service based on user preference
Selection method can accurately express user to the preference of attribute.
To achieve the goals above, the application uses following technical scheme:
Web service selection method based on user preference, comprising:
Sequence is improved by fuzzy number according to user to QoS attribute fancy grade for several service quality QoS attributes
Relation analysis determines subjective weight;
The objective weight of QoS attribute is determined using entropy assessment, subjective weight is corrected in realization by Objective Weight;
By the way of combination weighting, comprehensively considers subjective weights and Objective Weight finds out comprehensive QoS weight;
Specific service is selected by COSINE similarity improved TOPSIS method on the basis of considering user preference.
As the further technical solution of the application, order relation analytic approach is improved by fuzzy number and determines subjective weight, is adopted
The Assessment of Important that user provides is indicated with Triangular Fuzzy Number, by the corresponding ratio of Preference order to QoS attribute fancy grade
Triangular Fuzzy Number is converted into compared with language set.
As the further technical solution of the application, by the corresponding relatively language of the Preference order to QoS attribute fancy grade
Speech collection is converted into Triangular Fuzzy Number, and the when corresponding grade of the different degree of adjacent attribute is formed by specially in Preference order
Scale table corresponding with natural language description is converted, and is converted into Indexes Comparison grade and corresponding Triangular Fuzzy Number table, is calculated every
The weight of a attribute.
As the further technical solution of the application, scale table corresponding with natural language description are as follows:
One scale of table table corresponding with natural language description
Wherein, in Preference order, the ratio between the different degree of adjacent attribute is denoted as rk, rkIndicate Ck-1With CkThe ratio between importance, k
=2,3,4......n.
As the further technical solution of the application, Indexes Comparison grade and corresponding Triangular Fuzzy Number table are as follows:
Two comparison scale of table and corresponding Triangular Fuzzy Number
Grade | Rk |
Index Ck-1With index CkIt is of equal importance | (1.0,1.0,1.1) |
Index Ck-1With index CkIt is slightly important | (1.1,1.2,1.3) |
Index Ck-1With index CkIt is obvious important | (1.3,1.4,1.5) |
Index Ck-1With index CkIt is strong important | (1.5,1.6,1.7) |
Index Ck-1With index CkIt is extremely important | (1.7,1.8,1.8) |
Wherein, the ratio between different degree of adjacent attribute R in Preference orderk。
As the further technical solution of the application, Triangular Fuzzy Number and contacting is converted between number, passes through connection
It counts Triangular Fuzzy Number de-fuzzy.
As the further technical solution of the application, objective QoS weight is solved, specific steps are as follows:
QoS attribute value type according to QoS attribute is normalized, QoS attribute include two types, one
Kind is cost type, a kind of profit evaluation model;
For the data after normalization, specific gravity shared by i-th of service under j-th of QoS attribute is found out, according to comentropy
Solution formula finds out the information entropy of j-th of attribute;
The weight of each attribute is calculated according to information entropy.
As the further technical solution of the application, by the way of combination weighting, subjective weights and objective are comprehensively considered
Power is assigned, comprehensive QoS weight calculation is carried out, comprehensive QoS weight equation is as follows:
Wherein, βjFor the weight of j-th of attribute,For Confidence, n QoS attribute, λ ∈ [0,1] passes through the value tune of λ
Save subjective weight specific gravity shared in comprehensive weight.
As the further technical solution of the application, using the method improved TOPSIS of COSINE similarity come computation attribute
It is worth the distance of ideal solution, attribute data is mapped in vector space first, then the space by surveying between two vectors
Included angle cosine measures the similarity between them.
As the further technical solution of the application, specific step is as follows by improved TOPSIS:
QoS attribute matrix is normalized according to affiliated attribute type respectively;
Weighting matrix is established by each attribute synthesis weight;
Determine that optimal, most bad ideal solution, positive ideal solution are the maximum v of weighting matrixij, minus ideal result is that weighting matrix is minimum
ViIf positive ideal solution vector and minus ideal result vector respectively indicate;
Calculate the COSINE similarity and distance between candidate service and plus-minus ideal solutions vector;
It finds out to the distance of positive ideal solution vector and the distance of vector to minus ideal result vector according to vector and calculates each attribute
Compactness.
Embodiment of the present disclosure also discloses the selection system of the Web service based on user preference, including processor and calculating
Machine readable storage medium storing program for executing, processor is for realizing each instruction;Computer readable storage medium is for storing a plurality of instruction, the finger
It enables and is suitable for being loaded by processor and being executed the web service selection method based on user preference.
Compared with prior art, the beneficial effect of the disclosure is:
It assigns power mode for traditional user to be likely difficult to that accurately the preference of user is showed with weight, the disclosure mentions
It is a kind of out to improve G1 method with Triangular Fuzzy Number, with can more accurately giving expression to expressed by user by the characteristic of Triangular Fuzzy Number
Preference, while the brought subjectivity of G1 method is corrected by entropy assessment, keep weight more objective by combination weights, science.
The disclosure improves some inevitable disadvantages such as tradition TOPSIS such as backward simultaneously, improves service and decision-making
Accuracy, so that the service selected is more in line with the demand of user.
Detailed description of the invention
The accompanying drawings constituting a part of this application is used to provide further understanding of the present application, and the application's shows
Meaning property embodiment and its explanation are not constituted an undue limitation on the present application for explaining the application.
Fig. 1 be in embodiment of the present disclosure with the improved TOPSIS of COSINE similarity and traditional TOPSIS used into
Row comparison schematic diagram.
Specific embodiment
It is noted that following detailed description is all illustrative, it is intended to provide further instruction to the application.Unless another
It indicates, all technical and scientific terms that the disclosure uses have logical with the application person of an ordinary skill in the technical field
The identical meanings understood.
It should be noted that term used herein above is merely to describe specific embodiment, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singular
Also it is intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or their combination.
In a kind of typical embodiment of the application, the disclosure by fuzzy number improve order relation analytic approach (G1 method) come
The subjective weight for expressing user, QoS attribute objective weight is determined using entropy assessment, finally corrects subjective weight with objective weight
Find out comprehensive QoS weight.Service is finally selected by COSINE similarity improved TOPSIS method on the basis of considering user preference.
It is experimentally confirmed, disclosure effective solution user preference describes uncertain problem, is mentioned by improved TOPSIS method
The high accuracy of services selection, the service selected are more in line with the demand of user.
Be introduced separately below with regard to technology involved in technical scheme: about G1 method, G1 method is a kind of master
It sees and assigns Quan Fangfa, it is the shortcomings that improving to analytic hierarchy process (AHP), avoid traditional analytic hierarchy process (AHP), and use is simpler.G1
Method principle is first to be ranked up according to user's preferences to index, two neighboring index significance level is then judged, according to important journey
Degree finds out the weight of each index.This method is advantageous in that without carrying out consistency check, for possessing multiple QoS attributes
For Web service, AHP method can be because judge that order is too many and causes inconsistence problems.And for a user, category is provided
Property significance level sequence, it is more much easier two-by-two in than two attributes of judgement for carrying out adjacent index.So passing through G1 method
It is suitable for seeking user's subjectivity weight.
In the specific implementation, about G1 method, specific step is as follows:
If there is n QoS attribute, it is denoted as { C1, C2... ... .Cn}.User likes journey to attribute according to oneself
Degree, is ranked up attribute.If the disclosure is by attribute CiThan attribute CjIt is important to be denoted as Ci>Cj。
(1) a favorite attribute is selected in user elder generation dependence be denoted as C1。
(2) user selects again from remaining n-1 attribute oneself thinks that most important attribute is denoted as C2。
(3) successively go down, eventually form a Preference order C1>C2>.........Cn。
(4) it referring to table one, provides in Preference order, the ratio between the different degree of adjacent attribute is denoted as rk。rkIndicate Ck-1With CkWeight
The ratio between property wanted, k=2,3,4......n.It should be noted that r hereinkThe ratio between different degree when referring to real number.
One scale of table table corresponding with natural language description
(5) weight, W are calculated by following formulanRepresent the weight of n-th of attribute in Preference order, rkRepresent importance degree
Ratio.
Wn-1=rkWnFormula 2
Due in real life, the diversity due to QoS attribute and the ambiguity for the cognition of QoS attribute, than
Compared with being often difficult to provide accurately preference degree after two attributes, so going to indicate uncertain information frequently with fuzzy number.Interval number
It is common fuzzy number, but sometimes section may obtain excessive, be easy to cause error after carrying out interval arithmetic, and triangle
Fuzzy number, which had not only been able to maintain variable, is the section of value, but also provides the maximum median of possibility, can be very by Triangular Fuzzy Number
Accurate measurement can not be provided by solving object well, and but habit is evaluated with natural language.So the disclosure uses Triangle Module
Paste number indicates Assessment of Important that user provides.
Define 1: ifWhereinThen a is referred to as Triangular Fuzzy Number.IfWithFor two arbitrary Triangular Fuzzy Numbers, then Triangular Fuzzy Number algorithm is as follows:
M is any positive real number;
It defines 2: letting R be set of real numbers, then a+bi is referred to as connection number by a > 0, b ∈ R, i ∈ [- 1,1].A is known as determining that number, b are
Uncertain number, i is Uncertainty.Contacting number is that a mathematical tool is provided in Set Pair Analysis, can use Set Pair Analysis Theory
The amount of can determine and Uncertainty are connected.Triangular Fuzzy Number be exactly it is determining combined with uncertain, intermediate value be it is relatively determining,
And upper and lower two value is uncertain value, so the disclosure can by Triangular Fuzzy Number and contact number between convert, pass through connection
Coefficient is by Triangular Fuzzy Number de-fuzzy.If there is the Triangular Fuzzy Number to beIt is after being converted to connection numberThe wherein value range of i are as follows:It is former according to the value of i in the decision model of connection number
Reason,
Triangular Fuzzy Number is converted by the ratio between different degree referring to table one, as shown in Table 2, is denoted asIts
InRk,Rk Indicate that user thinks attribute Ck-1With CkThe ratio between importance is most conservative as a result, most probable as a result, most optimistic
As a result.Then user provides the Preference order C of attribute1>C2>.........Cn, refer again to table two and Preference order be converted into phase
The ratio between the different degree of adjacent attribute Rk, then further according to formula 1, formula 2 finds out the subjective weight of each attributeFinally according to definition
Two intermediate cam fuzzy numbers and several conversion formulas are contacted, is converted into real number.It should be noted that R hereinkRefer to referring to one turn of table
It is changed to the ratio between the different degree of Triangular Fuzzy Number.
Two comparison scale of table and corresponding Triangular Fuzzy Number
Grade | Rk |
Index Ck-1With index CkIt is of equal importance | (1.0,1.0,1.1) |
Index Ck-1With index CkIt is slightly important | (1.1,1.2,1.3) |
Index Ck-1With index CkIt is obvious important | (1.3,1.4,1.5) |
Index Ck-1With index CkIt is strong important | (1.5,1.6,1.7) |
Index Ck-1With index CkIt is extremely important | (1.7,1.8,1.8) |
It is solved about objective QoS weight:, can very big improvement user subjectivity although upper section improves G1 method by triangle fuzz method
Entitled randomness, but simple subjective weights do not reflect the connection between QoS attribute value, thus in order to allow weight more
Science, the disclosure correct subjective weight by Objective Weight.The concept source of entropy is in thermodynamics, after introducing information theory later,
It is widely applied in every field, Information Entropy is a kind of objective weighted model.Its principle is the comentropy of parameter, and comentropy is got over
Small, bigger comprising information content, proportion is higher in evaluation.It is applied to the weight of objective judgement QoS attribute by the disclosure, such as
Fruit attribute entropy is smaller, illustrates that the attribute value difference of the different candidate services of same attribute is larger, the disclosure assigns this attribute
Weigh relatively high weight.If entropy is bigger, illustrate that the attribute value of the different candidate services of same attribute is not much different, the disclosure
Assign this attribute to lesser weight.Specific step is as follows:
Step 1: QoS attribute value is normalized according to following formula, QoS attribute includes two types, a kind of
It is cost type, a kind of profit evaluation model.Cost type attribute, value is smaller more to be liked by requestor, such as the response time, price.Benefit
Type attribute, the value the high more is liked by requestor, such as reliability, stability.Attribute is divided into two types difference by the disclosure
It is normalized, formula is as follows:
Wherein qmaxAttribute maximum value thus, qminThe minimum value of attribute thus, qijAttribute value.
Step 2: the disclosure provides comentropy solution formula
Wherein, M is constant value, M=(- lnn)-1, SjFor the information entropy of j-th of attribute, m is the number of candidate service, Pij
Shared specific gravity is serviced for i-th under j-th of QoS attribute,CijTo normalize later data.
Step 3: calculating the weight of each attribute according to entropy, calculation formula is
Wherein, m is service number, SjFor the entropy of j-th of attribute.
About comprehensive QoS weight calculation: although subjective weighting method can be very good the preference for giving expression to user to attribute, but
It is there is certain subjectivity, and ignore qos value internal association.And the relationship between objective weighted model consideration value has ignored
The subjective preferences of user, so in order to make weight that can embody the subjective preferences of user, institute again while taking into account QoS Attribute Association
In such a way that the disclosure is using combination weighting, subjective weights and Objective Weight are comprehensively considered.Then combining weights formula is as follows:
Wherein, λ is Confidence, λ ∈ [0,1],For subjective weight.Subjective weight is adjusted in comprehensive weight by the value of λ
In shared specific gravity, the disclosure shows user to the confidence of the subjective weight specified by itself and to weight by Confidence
The intention of value.User's confidence is bigger, and the value of Confidence is bigger.When λ is 1, show weight completely by user's requirement.On the contrary,
When λ is 0, then show that assigning power by system assigns power according to entropy assessment.In general, the disclosure takes λ=0.5.
Finally, selecting QoS service: TOPSIS (Technique for Order Preference using improved TOPSIS
By Similarity to an Ideal Solution) method is common method in a kind of pair of multiobjective decision-making, due to this side
Method principle is simple, and convenient for calculating, so being widely used in services selection and multiobjective decision-making field.TOPSIS method
Principle is first to find the best and worst value of each attribute in m alternative objects, forms positive ideal solution and minus ideal result, so
The each attribute value of alternative services is regarded as each point in space afterwards, by calculate each point and positive ideal solution and minus ideal result it
Between Euclidean distance be ranked up.But traditional TOPSIS method has the following problems:
1. since index weights can not express completely the demand of user, to influence the result of TOPSIS.
2 ask the QoS attribute value of each alternative services to the distance of ideal solution by Euclidean distance method.When two differences are commented
, can not be more fine or not by exchange premium degree when valence scheme is equal to optimal, most inferior solution Euclidean distance, it is easy to appear error.
There is backward in 3 tradition TOPSIS methods, after increasing attribute or reducing attribute, as a result can generate error.
For first problem, the disclosure obscures G1 method by triangle and finds out subjective weight, and it is objective to be found out by entropy assessment
Weight finally integrates subjective and objective weight and finds out comprehensive weight, and weight can more express the demand of user.For Second Problem, if
Mahalanobis distance is taken to carry out improved TOPSIS method, but the application of mahalanobis distance needs Service Properties number to be less than candidate service number
Otherwise mesh can obtain the covariance matrix of not inverse matrix.This is clearly to set up always, so in actual use
It is bound too strong.
The disclosure is using the method improved TOPSIS of COSINE similarity come the distance of computation attribute value to ideal solution.
COSINE similarity is to calculate a kind of method of similarity between two high-dimensional vectors.This method first maps attribute data
Into vector space, the similarity between them is then measured by the space angle cosine surveyed between two vectors.Two
Individual vector angle is bigger to illustrate that similarity is lower at 0 ° to 180 °.If two vector A=[a1, a2 ..., an] and B=
[b1, b2 .., bn] then similarity between vector A and B
Wherein,For vector A, the mean value of vector B is insensitive for correcting COSINE similarity based method logarithm
Problem.
The COSINE distance of two vectors
D=1-sim (A, B) formula 9
Specific step is as follows by improved TOPSIS:
(1) to QoS attribute matrix according to formula 3, formula 4 is normalized;
(2) weighting matrix is established by each attribute synthesis weight;
(3) determine that optimal, most bad ideal solution, positive ideal solution are the maximum a of weighting matrixj, minus ideal result is weighting matrix
The smallest bjIf positive ideal solution vector and minus ideal result vector are indicated with A and B respectively:
A=(a1, a2..., an)
B=(b1, b2..., bn)
aj=max { vij, bj=min { vij, (i=1,2 ..., m;J=1,2 ... .., n) formula 10
(4) the COSINE similarity and distance between candidate service and plus-minus ideal solutions vector are calculated according to formula 8;
(5) compactness for calculating each candidate service is found out according to distance;
Wherein, di +It is distance of the vector to positive ideal solution vector, di _It is distance of the vector to minus ideal result vector.
The disclosure obscures G1 method and entropy assessment combination weighting method by triangle to determine that user requests the power of QoS attribute
Weight, and pass through the accuracy of COSINE similarity improvement tradition TOPSIS method.G1 method provides Importance of Attributes row by user
Sequence determines the weight of attribute, without carrying out consistency check.For a user, the sequence for providing attribute often belongs to than providing
Property weight wants much easier.When attribute value same for different Web services is not much different, although this attribute is critically important,
But what comparability it is without in fact, and the disclosure should give lesser weight, and the attribute that attribute value differs greatly then is answered
The weight is larger.Entropy assessment is exactly the entropy by seeking each index, the number of each indication information amount embodied, thus root
The weight of index is determined according to information content size.Obviously it is reasonable for solving objective weight using entropy assessment.Finally by combination
Synthesis weight after assigning power can accurately express user to the preference of attribute.
Embodiment of the present disclosure also discloses the selection system of the Web service based on user preference, including processor and calculating
Machine readable storage medium storing program for executing, processor is for realizing each instruction;Computer readable storage medium is for storing a plurality of instruction, the finger
It enables and is suitable for being loaded by processor and being executed the web service selection method based on user preference.
In order to enable those skilled in the art can clearly understand the technical solution of the application, below with reference to tool
The technical solution of the application is described in detail in the embodiment and comparative example of body.
This section illustrates the explanation of the disclosure by emulation experiment, and proves its feasibility.It suppose there is 6 and meet user's function
It can demand and QoS different candidate service.Wherein, there are price, response time, reliability, 4 QoS attributes of accuracy.Such as table three
It is shown.Wherein price, response time are cost type attribute, and the disclosure is normalized by formula 3;
Three candidate service group of table
True property is profit evaluation model attribute, is normalized by formula 4.Treated normalization matrix FijFollowing institute
Show.
First seek the subjective preferences of user, 4 attributes are successively ordered as price > sound according to oneself hobby by user first
Between seasonable > reliability > accuracy.Referring next to table two, the importance degree of more two neighboring index.Comparison result is
R2=(1.3,1.4,1.5)
R3=(1.5,1.6,1.7)
R4=(1.3,1.4,1.5)
Then the disclosure finds out price according to formula 1 and formula 2, and the weight of response time, reliability, accuracy are respectively
(0.37,0.40,0.43), (0.29,0.29,0.29), (0.17,0.18,0.19), (0.11,0.13,0.15), by Triangle Module
Paste weight is α=(0.34,0.29,0.21,0.16) by contacting number de-fuzzy to obtain weight.
The disclosure seeks the objective weight of user by entropy assessment, according to the matrix after normalization, data in disclosure matrix
It is updated to formula 5, the objective weight that formula 6 finds out attribute is β=(0.29,0.19,0.38,0.14).The last disclosure is comprehensive
Objective weight and subjective weight, find out combining weights W according to formula 7j=(0.32,0.24,0.29,0.15), in the disclosure,
λ=0.5 is taken, in practice, user can change comprehensive weight size by adjusting λ value according to their own needs.
Weighted decision matrix R is acquired by the weight disclosureij
The disclosure determines positive ideal solution R according to formula 10+With minus ideal result R,.Then positive ideal solution and minus ideal result are
R+=(0.320 0.240 0.290 0.150)
R-=(0.000 0.000 0.000 0.000)
Each scheme is calculated to positive ideal solution R by formula 8 and formula 9+With minus ideal result R-COS distance DR+=
(0.11,0.19,0.24,0.02,0.17,0.16), DR-=(0.9,0.91,0.91,0.89,0.91,0.9)
If the distance of service to a positive ideal solution is closer, the distance to minus ideal result is remoter, then services and more meet use
The demand at family is C according to the compactness that 11 disclosure of formula finds out each servicei=(0.89,0.82,0.80,0.98,0.84,
0.85).W can be learnt by the compactness disclosure4>W1>W6>W5>W2>W3。
Experimental analysis: by this experimental result and document, " Sun Ruonan, Zhang Bin, Liu Tingting are a kind of to use improvement entropy weight TOPSIS
The small-sized microcomputer system of Web service quality evaluation sort method [J], 2017,38 (06): 1221-1226. ", document
" Sun Xiuting, button is pretty, and a kind of Web service selection strategy [J] computer application based on Evaluation formula of Gong Qingbo, Li Zhen is ground
Study carefully, 2017,34 (08): 2408-2411+2428 " method, which is obtained a result, to be compared, as shown in Table 4:
Table four
Service ranking result | Weight | |
Methods herein | W4>W1>W6>W5>W2>W3 | 0.32 0.24 0.29 0.15 |
Document [9] method | W1>W2>W3>W6>W4>W5 | 0.29 0.19 0.38 0.14 |
Document [10] method | W1>W2>W5>W3>W4>W6 | 0.34 0.18 0.27 0.21 |
Document [9] Sun Ruonan, Zhang Bin, Liu Tingting is a kind of to be sorted using the Web service quality evaluation for improving entropy weight TOPSIS
The small-sized microcomputer system of method [J], 2017,38 (06): 1221-1226.
Document [10] Sun Xiuting, button is pretty, a kind of Web service selection strategy based on Evaluation formula of Gong Qingbo, Li Zhen
[J] computer application research, 2017,34 (08): 2408-2411+2428.
Disclosure user is ordered as price > response time > reliability > accuracy to attribute hobby, passes through comparison, discovery text
Offer " Sun Ruonan, Zhang Bin, Liu Tingting it is a kind of using improve entropy weight TOPSIS Web service quality evaluation sort method [J] it is small-sized
Microcomputer system, 2017,38 (06): 1221-1226. " using entropy assessment no matter weight and result all with the expectation of user
Gap is larger, and " Sun Xiuting, button is pretty, a kind of Web service selection strategy [J] based on Evaluation formula of Gong Qingbo, Li Zhen for document
2017,34 (08): computer application research analyses method using based on fuzzy AHP method and principal component in 2408-2411+2428 "
Combination, the result of weight and user it is expected some gaps.This experiment no matter each attribute weight and service ranking results on all
The demand for more meeting user, by not only can satisfy the subjective demand of user, while using based on fuzzy G1 method and entropy assessment again
Objective approach goes to correct subjective weight, and weight is more scientific and reasonable, and the service of selection is more in line with the request of user.
The disclosure is in terms of selection services accuracy with the improved TOPSIS of COSINE similarity and document " Sun Xiuting, button
Person of outstanding talent, a kind of Web service selection strategy [J] computer application research based on Evaluation formula of Gong Qingbo, Li Zhen, 2017,34
(08): traditional TOPSIS's used in 2408-2411+2428 " is compared as shown in Figure 1.It is obviously seen by the picture disclosure
Out, it is significantly improved by the improved TOPSIS method accuracy of COSINE similarity, avoids traditional TOPSIS and difference occur
Object is to positive ideal solution and minus ideal result apart from error identical to be judged.
The foregoing is merely preferred embodiment of the present application, are not intended to limit this application, for the skill of this field
For art personnel, various changes and changes are possible in this application.Within the spirit and principles of this application, made any to repair
Change, equivalent replacement, improvement etc., should be included within the scope of protection of this application.
Claims (10)
1. the web service selection method based on user preference, characterized in that include:
Order relation is improved by fuzzy number according to user to QoS attribute fancy grade for several service quality QoS attributes
Analytic approach determines subjective weight;
The objective weight of QoS attribute is determined using entropy assessment, subjective weight is corrected in realization by Objective Weight;
By the way of combination weighting, comprehensively considers subjective weights and Objective Weight finds out comprehensive QoS weight;
Specific service is selected by COSINE similarity improved TOPSIS method on the basis of considering user preference.
2. as described in claim 1 based on the web service selection method of user preference, characterized in that improved by fuzzy number
Order relation analytic approach determines subjective weight, and the Assessment of Important that user provides is indicated using Triangular Fuzzy Number, will be to QoS attribute
The corresponding relatively language set of the Preference order of fancy grade is converted into Triangular Fuzzy Number.
3. as claimed in claim 2 based on the web service selection method of user preference, characterized in that will like QoS attribute
The corresponding relatively language set of the Preference order of degree is converted into Triangular Fuzzy Number, and adjacent attribute is important specially in Preference order
The when corresponding grade of degree is formed by scale table corresponding with natural language description and is converted, and is converted into Indexes Comparison grade
And corresponding Triangular Fuzzy Number table, calculate the weight of each attribute.
4. as claimed in claim 3 based on the web service selection method of user preference, characterized in that scale and natural language
Corresponding table is described are as follows:
One scale of table table corresponding with natural language description
Wherein, in Preference order, the ratio between the different degree of adjacent attribute is denoted as rk, rkIndicate Ck-1With CkThe ratio between importance, k=2,
3,4......n。
5. as claimed in claim 3 based on the web service selection method of user preference, characterized in that Indexes Comparison grade and
Corresponding Triangular Fuzzy Number table are as follows:
Two comparison scale of table and corresponding Triangular Fuzzy Number
Wherein, the ratio between different degree of adjacent attribute R in Preference orderk。
6. as claimed in claim 2 based on the web service selection method of user preference, characterized in that by Triangular Fuzzy Number with
It is converted between connection number, by contacting number for Triangular Fuzzy Number de-fuzzy.
7. as described in claim 1 based on the web service selection method of user preference, characterized in that objective QoS weight is asked
Solution, specific steps are as follows:
QoS attribute value type according to QoS attribute is normalized, QoS attribute includes two types, one is
Cost type, a kind of profit evaluation model;
For the data after normalization, specific gravity shared by i-th of service under j-th of QoS attribute is found out, is solved according to comentropy
Formula finds out the information entropy of j-th of attribute;
The weight of each attribute is calculated according to entropy.
8. as described in claim 1 based on the web service selection method of user preference, characterized in that using combination weighting
Mode comprehensively considers subjective weights and Objective Weight, carries out comprehensive QoS weight calculation, and comprehensive QoS weight equation is as follows:
Wherein, βjFor the objective weight of j-th of attribute,For the subjective weight of j-th of attribute, λ is Confidence, n QoS attribute,
λ ∈ [0,1] adjusts subjective weight specific gravity shared in comprehensive weight by the value of λ.
9. as described in claim 1 based on the web service selection method of user preference, characterized in that similar using COSINE
The method improved TOPSIS of degree carrys out computation attribute value to the distance of ideal solution, and attribute data is mapped in vector space first,
Then the similarity between them is measured by the space angle cosine surveyed between two vectors;
Specific step is as follows by improved TOPSIS:
QoS attribute matrix is normalized according to affiliated attribute type respectively;
Weighting matrix is established by each attribute synthesis weight;
Determine that optimal, most bad ideal solution, positive ideal solution are the maximum v of weighting matrixij, minus ideal result is that weighting matrix is the smallest
viIf positive ideal solution vector and minus ideal result vector respectively indicate;
Calculate the COSINE similarity and distance between candidate service and plus-minus ideal solutions vector;
The patch for calculating each attribute is found out according to vector to the distance of positive ideal solution vector and the distance of vector to minus ideal result vector
It is right.
10. the Web service based on user preference selects system, including processor and computer readable storage medium, processor is used
In each instruction of realization;Computer readable storage medium is for storing a plurality of instruction, characterized in that described instruction is suitable for by processor
It loads and perform claim requires any web service selection method based on user preference of 1-9.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910035858.4A CN109784722A (en) | 2019-01-15 | 2019-01-15 | Web service selection method and system based on user preference |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910035858.4A CN109784722A (en) | 2019-01-15 | 2019-01-15 | Web service selection method and system based on user preference |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109784722A true CN109784722A (en) | 2019-05-21 |
Family
ID=66499403
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910035858.4A Pending CN109784722A (en) | 2019-01-15 | 2019-01-15 | Web service selection method and system based on user preference |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109784722A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110348747A (en) * | 2019-07-15 | 2019-10-18 | 齐鲁工业大学 | A kind of Intensive Design method and system of product design |
CN110584605A (en) * | 2019-09-10 | 2019-12-20 | 贾英 | Similarity-matched diagnosis and monitoring comprehensive medical system and matching method thereof |
CN110691000A (en) * | 2019-10-15 | 2020-01-14 | 山东理工大学 | Web service combination method based on fusion of FAHP and planning graph |
CN111242428A (en) * | 2019-12-31 | 2020-06-05 | 华为技术有限公司 | Microservice processing method, microservice processing device, microservice processing apparatus, and storage medium |
CN112885471A (en) * | 2021-03-12 | 2021-06-01 | 上海中医药大学附属岳阳中西医结合医院 | Psoriasis curative effect evaluation system based on Bayesian network maximum entropy self-learning extension set pair analysis |
CN115964570A (en) * | 2023-03-17 | 2023-04-14 | 湖南师范大学 | Cloud service recommendation method and device based on QoS multi-period change characteristic prediction |
CN117273238A (en) * | 2023-11-16 | 2023-12-22 | 四川省致链数字科技有限公司 | Wooden furniture service combination method and system based on QoS constraint |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106961356A (en) * | 2017-04-26 | 2017-07-18 | 中国人民解放军信息工程大学 | Web service choosing method and its device based on dynamic QoS and subjective and objective weight |
CN107070704A (en) * | 2017-03-22 | 2017-08-18 | 东南大学 | A kind of Trusted Web services combined optimization method based on QoS |
-
2019
- 2019-01-15 CN CN201910035858.4A patent/CN109784722A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107070704A (en) * | 2017-03-22 | 2017-08-18 | 东南大学 | A kind of Trusted Web services combined optimization method based on QoS |
CN106961356A (en) * | 2017-04-26 | 2017-07-18 | 中国人民解放军信息工程大学 | Web service choosing method and its device based on dynamic QoS and subjective and objective weight |
Non-Patent Citations (1)
Title |
---|
MAOYING WU等: "Research on Web Service Selection Based on User Preference", 《SPRINGERLINK》 * |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110348747A (en) * | 2019-07-15 | 2019-10-18 | 齐鲁工业大学 | A kind of Intensive Design method and system of product design |
CN110348747B (en) * | 2019-07-15 | 2021-11-26 | 齐鲁工业大学 | Intensive design method and system for product design |
CN110584605A (en) * | 2019-09-10 | 2019-12-20 | 贾英 | Similarity-matched diagnosis and monitoring comprehensive medical system and matching method thereof |
CN110691000A (en) * | 2019-10-15 | 2020-01-14 | 山东理工大学 | Web service combination method based on fusion of FAHP and planning graph |
CN110691000B (en) * | 2019-10-15 | 2021-12-21 | 山东理工大学 | Web service combination method based on fusion of FAHP and planning graph |
CN111242428A (en) * | 2019-12-31 | 2020-06-05 | 华为技术有限公司 | Microservice processing method, microservice processing device, microservice processing apparatus, and storage medium |
CN112885471A (en) * | 2021-03-12 | 2021-06-01 | 上海中医药大学附属岳阳中西医结合医院 | Psoriasis curative effect evaluation system based on Bayesian network maximum entropy self-learning extension set pair analysis |
CN112885471B (en) * | 2021-03-12 | 2023-01-24 | 上海中医药大学附属岳阳中西医结合医院 | Psoriasis curative effect evaluation system based on Bayesian network maximum entropy self-learning extension set pair analysis |
CN115964570A (en) * | 2023-03-17 | 2023-04-14 | 湖南师范大学 | Cloud service recommendation method and device based on QoS multi-period change characteristic prediction |
CN115964570B (en) * | 2023-03-17 | 2023-06-02 | 湖南师范大学 | Cloud service recommendation method and device based on QoS multi-period change feature prediction |
CN117273238A (en) * | 2023-11-16 | 2023-12-22 | 四川省致链数字科技有限公司 | Wooden furniture service combination method and system based on QoS constraint |
CN117273238B (en) * | 2023-11-16 | 2024-02-13 | 四川省致链数字科技有限公司 | Wooden furniture service combination method and system based on QoS constraint |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109784722A (en) | Web service selection method and system based on user preference | |
Du et al. | Managing noncooperative behaviors in large-scale group decision-making: Integration of independent and supervised consensus-reaching models | |
Peng et al. | User preferences based software defect detection algorithms selection using MCDM | |
US10019442B2 (en) | Method and system for peer detection | |
CN108092798A (en) | A kind of cloud service preferred method, Cloud Server based on change granularity | |
Yang et al. | SMAA-PO: project portfolio optimization problems based on stochastic multicriteria acceptability analysis | |
CN109783738A (en) | A kind of double extreme learning machine mixing collaborative filtering recommending methods based on more similarities | |
Benouaret et al. | Ws-sky: An efficient and flexible framework for qos-aware web service selection | |
Li et al. | Collaborative filtering algorithm with social information and dynamic time windows | |
Fu et al. | Collaborative filtering recommendation algorithm towards intelligent community | |
Fan et al. | Optimal selection of design scheme in cloud environment: A novel hybrid approach of multi-criteria decision-making based on F-ANP and F-QFD | |
CN107066328A (en) | The construction method of large-scale data processing platform | |
Chen et al. | DPM-IEDA: dual probabilistic model assisted interactive estimation of distribution algorithm for personalized search | |
Gao et al. | A user-knowledge dynamic pattern matching process and optimization strategy based on the expert knowledge recommendation system | |
Ruiz et al. | Ranking decision making units: The cross-efficiency evaluation | |
Sarkar | Fuzzy decision making and its applications in cotton fibre grading | |
Lin et al. | A dynamic grey target evaluation method with multiple reference points for new R&D institution performance | |
Zhao et al. | A memory-efficient approach to the scalability of recommender system with hit improvement | |
Kou et al. | A Dynamic Assessment Method for Urban Eco‐environmental Quality Evaluation | |
CN109919219A (en) | A kind of Xgboost multi-angle of view portrait construction method based on Granule Computing ML-kNN | |
Chalermpornpong et al. | Rating pattern formation for better recommendation | |
Zhang et al. | Building Energy Consumption Prediction Based on Temporal-Aware Attention and Energy Consumption States | |
CN108388911A (en) | A kind of mobile subscriber's Dynamic Fuzzy Clustering Algorithm method towards mixed attributes | |
Ran | Influence of government subsidy on high-tech enterprise investment based on artificial intelligence and fuzzy neural network | |
Liu et al. | Rank factor granules with fuzzy collaborative clustering and factor space theory |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190521 |
|
RJ01 | Rejection of invention patent application after publication |