CN109255079A - A kind of cloud service individual character recommender system and method based on sparse linear method - Google Patents
A kind of cloud service individual character recommender system and method based on sparse linear method Download PDFInfo
- Publication number
- CN109255079A CN109255079A CN201811345729.7A CN201811345729A CN109255079A CN 109255079 A CN109255079 A CN 109255079A CN 201811345729 A CN201811345729 A CN 201811345729A CN 109255079 A CN109255079 A CN 109255079A
- Authority
- CN
- China
- Prior art keywords
- cloud
- cloud service
- user
- individual demand
- scoring
- 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
- 238000000034 method Methods 0.000 title claims abstract description 67
- 238000011156 evaluation Methods 0.000 claims abstract description 15
- 239000011159 matrix material Substances 0.000 claims description 27
- 238000004458 analytical method Methods 0.000 claims description 21
- 230000000379 polymerizing effect Effects 0.000 claims description 6
- 229920000642 polymer Polymers 0.000 claims description 5
- 238000004891 communication Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 4
- 238000005516 engineering process Methods 0.000 abstract description 2
- 238000002474 experimental method Methods 0.000 description 11
- 238000004422 calculation algorithm Methods 0.000 description 8
- 238000013480 data collection Methods 0.000 description 8
- 238000001914 filtration Methods 0.000 description 6
- NJPPVKZQTLUDBO-UHFFFAOYSA-N novaluron Chemical compound C1=C(Cl)C(OC(F)(F)C(OC(F)(F)F)F)=CC=C1NC(=O)NC(=O)C1=C(F)C=CC=C1F NJPPVKZQTLUDBO-UHFFFAOYSA-N 0.000 description 6
- 230000008859 change Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 201000006549 dyspepsia Diseases 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 235000012054 meals Nutrition 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/51—Discovery or management thereof, e.g. service location protocol [SLP] or web services
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Tourism & Hospitality (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention is suitable for field of cloud computer technology, provides a kind of cloud service personalized recommendation system and method based on sparse linear method, this method comprises: S1, reception cloud user UiThe individual demand P of submissionk, that is, select several attributes valued the most；S2, it is based on cloud user UiHistory evaluation data in rating database obtain cloud user in individual demand PkIt is directed to the scoring of each cloud service J down；S3, to cloud user UiThe highest cloud service of recommendation score.Cloud service personalized recommendation method provided by the invention based on sparse linear method has following advantageous effects: 1. can precisely and neatly define individual demand；2. obtaining cloud user by the relevant historical data of study under each individual demand and without the deviation under individual needs, predicted based on deviation, so that recommendation results are more accurate.
Description
Technical field
The invention belongs to field of cloud computer technology, provide a kind of cloud service individual character recommendation system based on sparse linear method
System and method.
Background technique
It is proposed that " since cloud computing concept, cloud computing is easy to extend, store on demand, elastometer with it from Google in 2006
The advantages that calculation, rapidly develops in the whole world.In recent years, Ali, Baidu, Google etc. propose cloud service in succession, with major operator
Constantly by various computing resources and storage capacity etc. with infrastructure i.e. service (Infrastructure-as-a-Service,
IaaS), platform services (Platform-as-a-Service, PaaS) and software services (Software-as-a-
Service, SaaS) etc. forms be published in network, occur more and more cloud services in network, bring information overload problem.
And cloud service that non-functional attribute different identical in face of numerous functional attributes, user, which is difficult to be rapidly selected out, meets oneself
The cloud service of individual demand[1], it is user among the cloud service of magnanimity how under complicated cloud computing environment therefore
Quickly recommending the cloud service for meeting users ' individualized requirement to be one is worth the problem of furtheing investigate.
In recent years, the algorithm recommended about cloud service proposes very much, and common several proposed algorithms have: 1. collaborative filtering
Proposed algorithm, this is a kind of the most widely used recommended method.Such methods are divided into based on user/project collaborative filtering method
With the collaborative filtering based on model.It by user/project collaborative filtering method is utilized based on use experience of the user to project
It calculates to find similar user/project, recycles similar users/project to be recommended, such as gone out a kind of sketch-based user interface similar value
Calculation method, recommend by comparing the attribute value of the currently used service of user and similar services, it is such as a kind of to be based on user
The collaborative filtering of preference, the general advisory speed of such methods is fine but accuracy rate is inadequate；Collaborative filtering based on model is
Using machine learning and statistical method from existing scoring middle school's acquistion to model, this model is recycled to carry out project scoring pre-
It surveys, thus obtains project recommendation, such methods performance is relatively preferable, but indigestion or some are latent when its common problem
It cannot be explained well in factor；2. content-based recommendation method, thinking are: utilizing the description and use of project substantive characteristics
The historical record at family, for user recommend in its not used project with the highest project of historical record project similarity degree.It is such
Method disadvantage is that the extraction to cloud service attribute, analysis, quantization, expression expend more resource.3. based on correlation rule
Recommended method.Correlation rule proposed algorithm does not need domain knowledge, can find new point of interest, but this method rule extraction it is difficult,
Personalization level is low.
Summary of the invention
The embodiment of the present invention provides a kind of cloud service individual character recommended method based on sparse linear method, it is desirable to provide a kind of
It explains flexibly, and recommendation results more accurately cloud service personalized recommendation method.
To achieve the goals above, the present invention provides a kind of, and the cloud service individual character based on sparse linear method recommends system
System, the system include:
Cloud client, cloud assessment centers and cloud recommend center；Cloud client and cloud assessment centers and cloud recommend center
Communication, wherein
Cloud client: for recommending center to submit individual demand and evaluate based on used cloud service to cloud to cloud
Feed back cloud clothes evaluation in center；
Cloud assessment centers: the cloud service evaluation of cloud user feedback is parsed, cloud is obtained and takes device under individual demand
Scoring, and above-mentioned data are updated to rating database；
Cloud recommends center, for the individual demand of the submission of cloud user, based on the evaluation data in rating database to
Cloud user recommends cloud service.
To achieve the goals above, the present invention provides a kind of cloud service individual character recommendation side based on sparse linear method
Method, described method includes following steps:
S1, it receives and uses cloud user UiThe individual demand P of submissionk, that is, select several attributes valued the most；
S2, it is based on cloud user UiHistory evaluation data in rating database predict cloud user in individual demand PkUnder
For the scoring of each cloud service J；
S3, to cloud user UiThe highest cloud service of recommendation score.
Further, the step S2 specifically comprises the following steps:
S21, cloud user U is calculatediFor the cloud service J that scoredhIn individual demand PkIt descends and the scoring under no individual needs
Deviation calculates effort analysis matrix D；
S22, it is based on effort analysis matrix D and sparse polymer matrix W, to predict cloud user UiIn individual demand PkLower pair
The scoring of cloud service J, the scoring of the cloud service J include: the cloud service J that scoredhScoring and the cloud service J that do not scorewComment
Point.
Further, if cloud user U in step S221iMultiple cloud service J are directed under no individual demandhIt is commented
Point, then use multiple cloud service JhThe average value of scoring is as cloud user UiScore R under no individual demandi,h。
Further, in step S221, cloud user U if it does not existiTo the cloud service J that scored under no individual demandh
Scoring, using cloud user UiTo the cloud service J that scored under all individual demandshGrade average, as UiWithout a
To the cloud service J that scored under property demandhScoring Ri,h。
Further, user U is calculated based on formula (1) and formula (2)iIn individual demand PkUnder to j-th of cloud service
JjScoringFormula (1) and formula (2) are specific as follows:
Wherein, Ri,nFor cloud user UiCloud service J is directed under no individual demandnScoring, Di.kFor cloud user UiA
Property demand PkScoring and the effort analysis under no individual demand of knit stitch, K are the species number of personalized deviation, and N is cloud clothes
The number of business, pkFor users ' individualized requirement vector, Wn,jFor cloud service JnWith cloud service JjBetween polymerizing factor.
Cloud service personalized recommendation method provided by the invention based on sparse linear method is imitated with following Advantageous
Fruit:
1. individual demand precisely and can be defined neatly；
2. obtaining cloud user by the relevant historical data of study under each individual demand and under no individual needs
Deviation is predicted based on deviation, so that recommendation results are more accurate.
Detailed description of the invention
Fig. 1 is cloud service personalized recommendation system structural schematic diagram provided in an embodiment of the present invention；
Fig. 2 is the cloud service individual character recommended method flow chart provided in an embodiment of the present invention based on sparse linear method；
Fig. 3 is the MAP emulation experiment figure provided in an embodiment of the present invention on music data collection；
Fig. 4 is the MAP emulation experiment figure provided in an embodiment of the present invention on cinematic data collection；
Fig. 5 is the MAP emulation experiment figure on data set at the restaurant provided in an embodiment of the present invention；
Fig. 6 is the Recall emulation experiment figure provided in an embodiment of the present invention on music data collection；
Fig. 7 is the Recall emulation experiment figure provided in an embodiment of the present invention on cinematic data collection；
Fig. 8 is the Recall emulation experiment figure on data set at the restaurant provided in an embodiment of the present invention；
Fig. 9 is the Precision emulation experiment figure provided in an embodiment of the present invention on music data collection；
Figure 10 is the Precision emulation experiment figure provided in an embodiment of the present invention on cinematic data collection；
Figure 11 is the Precision emulation experiment figure on data set at the restaurant provided in an embodiment of the present invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
The relevant explanation for some professional terms that this patent is related to:
1) personalized recommendation: personalized recommendation is to be recommended according to the interest preference of user, and cloud user is selecting
When cloud service, what is usually considered first is the functional attributes of cloud service, and the individual demand of cloud user is namely to cloud service function
The requirement of energy attribute, because the invention provides at least five functional attributes of cloud service, 5 functional attributes are as follows:
1. reliability r (reliability): failure-free operation at the appointed time and after breaking down in the stipulated time
The interior ability for restoring service；
2. availability a (availability): persistently providing the ability of specific function under regulation environment；
3. safety s (security): providing the ability of data security and integrality；
4. real-time e (real-time): the ability of respond request at the appointed time；
5. maintainability m (maintainability): being easy to modification and perfect ability；
The individual demand of cloud user is defined as cloud user most concerned several clouds when using cloud service and taken by this patent
Business attribute, this several cloud service attribute are divided into different degree, form individual demand class example with two cloud service attributes and said
It is bright, such as: 1. 2. timeliness is a kind of individual demand for safety, and 1. 2. safety is that another personalization needs to timeliness
It asks, why is since it is considered that more cloud users' values with the mode combined in this way.
Fig. 1 is cloud service personalized recommendation system structural schematic diagram provided in an embodiment of the present invention, for ease of description, only
It is illustrated in the part that the embodiment of the present invention is first closed.
The cloud service personalized recommendation system includes: cloud client, and cloud assessment centers and cloud recommend center；Cloud client
Recommend center communication with cloud assessment centers and cloud, wherein
Cloud client is used to that center to be recommended to submit individual demand and based on used cloud service into cloud evaluation to cloud
The heart feeds back cloud clothes evaluation；
Cloud assessment centers: the cloud service evaluation of cloud user feedback is parsed, cloud is obtained and takes device in the individual demand
Under scoring, and above-mentioned data are updated to rating database；
Databases contain cloud user and score for the history of each cloud service, and history scoring includes: in no personalization
History scoring under demand and the history scoring under each individual demand, and based on cloud user currently under certain individual demand
History of the cloud user under same personalized demand for identical cloud service is updated for the newest scoring of certain cloud service to comment
Point.
Cloud recommends center, for the individual demand of the submission of cloud user, based on the evaluation data in rating database to
Cloud user recommends cloud service；
The individual demand that cloud recommends center to be used to submit based on cloud user, obtains the history of cloud user in database
Scoring.
Fig. 2 is the cloud service individual character recommended method flow chart provided in an embodiment of the present invention based on sparse linear method, cloud
The cloud service individual character recommended method at recommendation center specifically comprises the following steps:
S1, it receives and uses cloud user UiThe individual demand P of submissionk, that is, select several attributes valued the most；
S2, it is based on cloud user UiHistory evaluation data in rating database obtain cloud user in individual demand PkUnder
For the scoring of each cloud service J；
S3, to cloud user UiThe highest cloud service of recommendation score.
In embodiments of the present invention, step S2 specifically comprises the following steps:
S21, cloud user U is calculatediFor the cloud service J that scoredhIn individual demand PkIt descends and the scoring under no individual needs
Deviation calculates effort analysis matrix D；
Since the data set of cloud user's evaluation is than sparse, i.e., it cannot be guaranteed that the same cloud user is in identical individual character
Scoring is all made that other cloud services under change demand, thus has no idea to obtain cloud user under same personalized demand
In order to solve this problem scoring to other cloud services introduces personalized effort analysis.Personalized effort analysis is meant
Cloud user in no individual demand and has the deviation in the case of individual demand between scoring to cloud service, that is, recognizes
For for every a pair<cloud user, all there is deviation in the scoring of individual demand>and cloud user under no individual demand.
As shown in table 1, certain user is 4 to the scoring of certain cloud service in the case where no individual demand, but under
It is respectively 3 and 5 to the scoring of this cloud service under the individual demand of two kinds of face, this illustrates user in conditions of demand with no personalization
It is lower that there are deviations between the scoring when there is individual demand.
Whether there is or not user's scoring examples under individual demand for table 1
| User | Cloud service | Scoring | Individual demand |
| u1 | c1 | 4 | Nothing |
| u1 | c1 | 3 | 1. real-time 2. reliability |
| u1 | c1 | 5 | 1. safety 2. real-time |
Effort analysis matrix D is N × K two-dimensional matrix, for indicating personalized effort analysis, effort analysis matrix D
In every a line represent a cloud user under different individual demands scoring and the effort analysis under no individual demand, it is each
Column represent scoring of all users under this individual demand and the effort analysis under no individual demand, as shown in table 2；
The personalized effort analysis matrix D of table 2
In embodiments of the present invention, if cloud user U is not present in databasenIn individual demand PmLower historical data, i.e. cloud
User UnIn individual demand PkIt is lower not score any cloud service, then recommend each cloud service at random to user；
In embodiments of the present invention, if cloud user U in step S221iTo cloud service of having scored under no individual demand
JhScoring be it is multiple, then using multiple scorings average value as cloud user UiScore R under no individual demandi,h；If
There is no cloud user UiTo the J for cloud service of having scored under no individual demandhScoring, using cloud user UiIn all individual characteies
To the cloud service J that scored under change demandhGrade average, as UiTo the cloud service J that scored under no individual demandh's
Score Ri,h。
S22, it is based on effort analysis matrix D and sparse polymer matrix W, to predict cloud user UiIn individual demand PkLower pair
The scoring of cloud service J, the scoring of the cloud service J include: the cloud service J that scoredhScoring and the cloud service J that do not scorewComment
Point.
Sparse polymer matrix W is the nonnegative matrix of N*N, indicates the polymerizing factor between cloud service, is similar between project
Similarity matrix, every a line and each column in sparse polymer matrix represent the polymerizing factor between cloud service.
User U is calculated based on formula (1) and formula (2)iIn individual demand PkUnder to j-th of cloud service JjScoringFormula (1) and formula (2) are specific as follows:
Wherein, Ri,nFor cloud user UiCloud service J is directed under no individual demandnScoring, Di.kFor cloud user UiA
Property demand PkUnder scoring with the effort analysis under no individual demand, K be personalization deviation species number, N is cloud service
Number, pkFor users ' individualized requirement vector, Wn,jFor cloud service JnWith cloud service JjBetween polymerizing factor,For cloud user
UiIn individual demand PkIt is directed to cloud service J downnIntermediate scoring.
Understanding for formula (2), based on cloud user for scoring of a certain cloud service under sharp individual demand and
Property effort analysis score cloud user is calculated under same personalized demand the intermediate of other cloud services.Also,
In this patent, same still based on public affairs even if cloud user carried out scoring to part cloud service under a certain individual demand
Formula (2) scores to the intermediate of cloud service of having scored to calculate cloud user under same personalized demand.Because can make in this way
More data are learnt to obtain in matrix D, so that it is more accurate so that final prediction is scored, obtain better recommendation effect.
Understanding for formula (1) calculates cloud user UiIn individual demand PkUnder to cloud service JnScoring?
It is exactly to utilize cloud user UiIn individual demand PkUnder to remove cloud service JnScoring is multiplied with polymerizing factor W among cloud service in addition
It obtains, i.e., is calculated based on formula (1).
pkIndicate cloud users ' individualized requirement binary vector, be the combined situation according to individual demand, than if any
Then (0,0,0,0) p, this p (0,0,1,0) when individual demand is the third can become more with combination and increase by four kinds of situations,
Also being exactly can be as the variation of combination, p also has corresponding variation.Such as if it exists simultaneously comprising three or four kinds of person p (0,
0,1,1).
It is true with user since the row and column point of personalized deviation matrix represents user (u) and individual demand (p)
Personalization scoring is training set, from the parameter learnt in matrix D and W in data, only relies on four for the prediction scoring of cloud service
A factor: without rating matrix R, personalized deviation matrix D, project convergence factor matrix W and the user personality under individual demand
Change requirement vector p.
The solution of matrix D and W can be converted into the minimum problems of regularization optimization problem, can be counted with formula (4)
It calculates.Wherein | | W | |FIt is the Frobenius normal form (l of matrix WfNormal form), | | W | |1It is the l of matrix W1Normal form, | | D | |FWith | | W |
|1Similarly；θ and β is learning rate.
It is excellent that formula (3) can use stochastic gradient descent method (stochastic gradient descent, SGD) progress
Change and solve, wherein the update of parameter passes through formula (4)-(6):
Wherein parameter alpha1, β1, λ1, α2, β2, λ2It is learning rate, for Di,k, we only update the parameter of position k, and
It is to work as pkIt is just updated when=1；And Wh,jWe only update the parameter h of relevant position, the i.e. scored cloud service collection of user.
Cloud service personalized recommendation method provided by the invention based on sparse linear method is imitated with following Advantageous
Fruit: 1. can precisely and flexibly define individual demand；2. obtaining cloud user by the relevant historical data of study each
It under individual demand and without the deviation under individual needs, is predicted based on deviation, so that recommendation results are more accurate.
In order to verify the performance that this patent proposes method, emulation experiment is from rate of precision (precision), recall rate
(recall) it is compared with mean accuracy mean value (MAP) three standards with benchmark algorithm.Benchmark algorithm is more commonly used
Two kinds of proposed algorithms: CF and CASA.Rate of precision and recall rate are to assess module popular in recommender system, define C
It is the recommendation list obtained on test set, D is the recommendation list obtained on training set, and enabling I=C ∩ D, i.e. I is calculated based on recommendation
The correct service recommendation result that method obtains.Based on these, the calculation formula of precision and recall be respectively (7) and
(8).MAP is another common ranking index, it takes into account the recommendation grade of ranked items.MAP can pass through formula
(9) it calculates.
Wherein, M indicates the quantity of cloud user, and N indicates the length of recommendation list, and P (k) indicates to end at k in recommendation list
Precision, m indicate relevant item quantity.
Following multiple groups emulation experiment figure is by method proposed in this paper and common two kinds of proposed algorithms CF and CASA three
Performance on kind data set compares, and the standard of assessment is rate of precision mentioned above (precision), recall rate
(recall) and mean accuracy mean value (MAP).
User generally also has personalized demand, such as music, film and meal when selecting music, film and restaurant to it
The type in shop, so cloud service data set is simulated using music, film and three, restaurant data set herein, such as table 3.Use three
A different data set is to verify the performance of methods herein on different data sets.
3 data set of table
Based on Simulation results Fig. 3 to Figure 11 of data set in above-mentioned three, show that upper method proposed in this paper is usually excellent
In pedestal method.Although method proposed in this paper is not so good as pedestal method in certain situations in terms of some, such as in cinematic data collection
On accuracy for, process proposed herein in preceding 10 recommendation be not so good as pedestal method, but after 10 be better than base
Quasi- method；There are also the accuracys on data set at the restaurant to be not so good as CF method after 20, but is better than pedestal method in preceding 20,
In addition to this, context of methods is superior to pedestal method.Performance wherein on music data collection is especially prominent, this is because music
Data set is more intensive, that is,<user, and project>to presence is scored under different individual demands, so that deviation
Matrix can preferably learn to effort analysis, so that recommendation effect is more accurate.Context of methods shows very on MAP simultaneously
It is good, it is relatively more to be superior to pedestal method.Therefore model proposed in this paper effect in personalized recommendation is more preferable
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.
Claims (6)
1. a kind of cloud service personalized recommendation system based on sparse linear method, which is characterized in that the system comprises:
Cloud client, cloud assessment centers and cloud recommend center；Cloud client and cloud assessment centers and cloud recommend center communication,
Wherein,
Cloud client: for recommending center to submit individual demand to cloud and being based on used cloud service to cloud assessment centers
Feed back cloud clothes evaluation；
Cloud assessment centers: the cloud service evaluation of cloud user feedback is parsed, cloud service commenting under individual demand is obtained
Point, and above-mentioned data are updated to rating database；
Cloud recommends center, for the individual demand of the submission of cloud user, is used based on the evaluation data in rating database to cloud
Recommend cloud service in family.
2. the recommended method of the cloud service personalized recommendation system based on sparse linear method as described in claim 1, feature
It is, described method includes following steps:
S1, it receives and uses cloud user UiThe individual demand P of submissionk, that is, select several attributes valued the most；
S2, it is based on cloud user UiHistory evaluation data in rating database predict cloud user in individual demand PkUnder be directed to
The scoring of each cloud service J；
S3, to cloud user UiThe highest cloud service of recommendation score.
3. as claimed in claim 2 based on the cloud service personalized recommendation method of sparse linear method, which is characterized in that the step
Rapid S2 specifically comprises the following steps:
S21, cloud user U is calculatediFor the cloud service J that scoredhIn individual demand PkIt descends and the effort analysis under no individual needs
To calculate effort analysis matrix D；
S22, it is based on effort analysis matrix D and sparse polymer matrix W, to predict cloud user UiIn individual demand PkUnder to cloud take
The scoring of business J, the scoring of the cloud service J includes: the cloud service J that scoredhScoring and the cloud service J that do not scorewScoring.
4. as claimed in claim 3 based on the cloud service personalized recommendation method of sparse linear method, which is characterized in that if step
Cloud user U in S221iMultiple cloud service J are directed under no individual demandhIt scores, then uses multiple cloud services
JhThe average value of scoring is as cloud user UiScore R under no individual demandi,h。
5. as claimed in claim 3 based on the cloud service personalized recommendation method of sparse linear method, which is characterized in that in step
In S221, cloud user U if it does not existiTo the cloud service J that scored under no individual demandhScoring, using cloud user UiInstitute
Have under individual demand to the cloud service J that scoredhGrade average, as UiTo the cloud clothes that scored under no individual demand
Be engaged in JhScoring Ri,h。
6. as claimed in claim 3 based on the cloud service personalized recommendation method of sparse linear method, which is characterized in that based on public affairs
Formula (1) and formula (2) calculate user UiIn individual demand PkUnder to j-th of cloud service JjScoringFormula (1) and
Formula (2) is specific as follows:
Wherein, Ri,nFor cloud user UiCloud service J is directed under no individual demandnScoring, Di.kFor cloud user UiIn personalization
Demand PkScoring and the effort analysis under no individual demand of knit stitch, K are the species number of personalized deviation, and N is cloud service
Number, pkFor users ' individualized requirement vector, Wn,jFor cloud service JnWith cloud service JjBetween polymerizing factor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811345729.7A CN109255079B (en) | 2018-11-13 | 2018-11-13 | Cloud service personality recommendation system and method based on sparse linear method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811345729.7A CN109255079B (en) | 2018-11-13 | 2018-11-13 | Cloud service personality recommendation system and method based on sparse linear method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109255079A true CN109255079A (en) | 2019-01-22 |
| CN109255079B CN109255079B (en) | 2021-09-28 |
Family
ID=65043732
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811345729.7A Active CN109255079B (en) | 2018-11-13 | 2018-11-13 | Cloud service personality recommendation system and method based on sparse linear method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109255079B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111428040A (en) * | 2020-04-01 | 2020-07-17 | 中移雄安信息通信科技有限公司 | Method, device and equipment for determining target object and computer storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102523247A (en) * | 2011-11-24 | 2012-06-27 | 合肥工业大学 | Cloud service recommendation method and device based on multi-attribute matching |
| CN103051730A (en) * | 2013-01-15 | 2013-04-17 | 合肥工业大学 | Multi-source information service-resource allocating system and IA-Min allocating method in cloud-computing business environment |
| CN103237023A (en) * | 2013-04-16 | 2013-08-07 | 安徽师范大学 | Dynamic trust model establishing system |
| US20150154670A1 (en) * | 2013-12-04 | 2015-06-04 | Nec Laboratories America, Inc. | Online Optimization and Fair Costing for Dynamic Data Sharing in a Cloud Data Market |
| CN105574685A (en) * | 2016-02-02 | 2016-05-11 | 浙江工业大学 | Subjective and objective combination-based cloud service evaluation method |
| CN106775948A (en) * | 2016-12-23 | 2017-05-31 | 安徽师范大学 | A kind of cloud method for scheduling task and device based on priority |
| CN108182273A (en) * | 2018-01-23 | 2018-06-19 | 成都信达智胜科技有限公司 | Network data processing method based on cloud storage |
-
2018
- 2018-11-13 CN CN201811345729.7A patent/CN109255079B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102523247A (en) * | 2011-11-24 | 2012-06-27 | 合肥工业大学 | Cloud service recommendation method and device based on multi-attribute matching |
| CN103051730A (en) * | 2013-01-15 | 2013-04-17 | 合肥工业大学 | Multi-source information service-resource allocating system and IA-Min allocating method in cloud-computing business environment |
| CN103237023A (en) * | 2013-04-16 | 2013-08-07 | 安徽师范大学 | Dynamic trust model establishing system |
| US20150154670A1 (en) * | 2013-12-04 | 2015-06-04 | Nec Laboratories America, Inc. | Online Optimization and Fair Costing for Dynamic Data Sharing in a Cloud Data Market |
| CN105574685A (en) * | 2016-02-02 | 2016-05-11 | 浙江工业大学 | Subjective and objective combination-based cloud service evaluation method |
| CN106775948A (en) * | 2016-12-23 | 2017-05-31 | 安徽师范大学 | A kind of cloud method for scheduling task and device based on priority |
| CN108182273A (en) * | 2018-01-23 | 2018-06-19 | 成都信达智胜科技有限公司 | Network data processing method based on cloud storage |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111428040A (en) * | 2020-04-01 | 2020-07-17 | 中移雄安信息通信科技有限公司 | Method, device and equipment for determining target object and computer storage medium |
| CN111428040B (en) * | 2020-04-01 | 2023-08-18 | 中移雄安信息通信科技有限公司 | Target object determination method, device, equipment and computer storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109255079B (en) | 2021-09-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wu et al. | Contextual bandits in a collaborative environment | |
| US11216503B1 (en) | Clustering search results | |
| US10296849B2 (en) | Member clustering with equi-sized partitions | |
| CN108140075A (en) | User behavior is classified as exception | |
| CN106844407B (en) | Tag network generation method and system based on data set correlation | |
| CN109146626B (en) | Fashionable garment matching recommendation method based on user dynamic interest analysis | |
| CN106407349A (en) | Product recommendation method and device | |
| Zheng et al. | Integrating context similarity with sparse linear recommendation model | |
| US20200302398A1 (en) | Determination of profile match for job posting | |
| CA3059929A1 (en) | Text searching method, apparatus, and non-transitory computer-readable storage medium | |
| Na et al. | User identification based on multiple attribute decision making in social networks | |
| US9336330B2 (en) | Associating entities based on resource associations | |
| KR101346927B1 (en) | Search device, search method, and computer-readable memory medium for recording search program | |
| Lessa et al. | Filtering graduate courses based on LinkedIn profiles | |
| CN104572915B (en) | One kind is based on the enhanced customer incident relatedness computation method of content environment | |
| Wei et al. | Online education recommendation model based on user behavior data analysis | |
| US20150293988A1 (en) | System and Method for Opinion Sharing and Recommending Social Connections | |
| US9400789B2 (en) | Associating resources with entities | |
| CN114490923A (en) | Training method, device and equipment for similar text matching model and storage medium | |
| CN109255079A (en) | A kind of cloud service individual character recommender system and method based on sparse linear method | |
| US11238102B1 (en) | Providing an object-based response to a natural language query | |
| CN105095385B (en) | A kind of output method and device of retrieval result | |
| Wang et al. | Occupation recommendation based on student achievement mining in vocational skill training | |
| US11544332B2 (en) | Bipartite graph construction | |
| CN113657766A (en) | Tourist attraction joy index metering method based on tourist multi-metadata |
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 |