CN109255079B - Cloud service personality recommendation system and method based on sparse linear method - Google Patents

Cloud service personality recommendation system and method based on sparse linear method Download PDF

Info

Publication number
CN109255079B
CN109255079B CN201811345729.7A CN201811345729A CN109255079B CN 109255079 B CN109255079 B CN 109255079B CN 201811345729 A CN201811345729 A CN 201811345729A CN 109255079 B CN109255079 B CN 109255079B
Authority
CN
China
Prior art keywords
cloud
personalized
cloud service
user
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.)
Active
Application number
CN201811345729.7A
Other languages
Chinese (zh)
Other versions
CN109255079A (en
Inventor
张佩云
叶金勇
徐鸽
谢杰敏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Anhui Normal University
Original Assignee
Anhui Normal University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Anhui Normal University filed Critical Anhui Normal University
Priority to CN201811345729.7A priority Critical patent/CN109255079B/en
Publication of CN109255079A publication Critical patent/CN109255079A/en
Application granted granted Critical
Publication of CN109255079B publication Critical patent/CN109255079B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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 invention is suitable for the technical field of cloud computing, and provides a cloud service personalized recommendation system and method based on a sparse linear method, wherein the method comprises the following steps: s1 cloud user U for receivingiSubmitted personalized requirements PkNamely, selecting a plurality of most important attributes; s2 cloud user U-basediHistorical evaluation data in an evaluation database are used for acquiring the personalized demand P of the cloud userkA score for each cloud service J is given; s3 cloud user UiRecommending the highest-scoring cloud service. The cloud service personalized recommendation method based on the sparse linear method has the following beneficial technical effects: 1. personalized requirements can be accurately and flexibly defined; 2. the deviation of the cloud user under each personalized demand and the deviation without the personalized demand are obtained through the learned related historical data, and prediction is carried out based on the deviation, so that the recommendation result is more accurate.

Description

Cloud service personality recommendation system and method based on sparse linear method
Technical Field
The invention belongs to the technical field of cloud computing, and provides a cloud service personality recommendation system and method based on a sparse linear method.
Background
Since Google proposed the concept of "cloud computing" in 2006, cloud computing has rapidly developed globally with its advantages of easy expansion, on-demand storage, flexible computing, and the like. In recent years, cloud services have been proposed successively by ari, hundredth, Google, and the like, and as each large operator distributes various computing resources, storage capacities, and the like to a network in the form of Infrastructure as a Service (IaaS), Platform as a Service (PaaS), Software as a Service (SaaS), and the like, more and more cloud services appear in the network, which brings about a problem of information overload. In the face of a plurality of cloud services with the same functional attributes but different functional attributes, a user is difficult to quickly select the cloud service meeting the personalized requirements of the user[1]Therefore, how to quickly recommend and satisfy the requirements of the user in the massive cloud services in the complex cloud computing environmentCloud services for user personalization requirements are a problem worthy of intensive research.
In recent years, many algorithms for cloud service recommendation are proposed, and several common recommendation algorithms are: the method is a recommendation method which is most widely used. Such methods are classified into user/item-based collaborative filtering methods and model-based collaborative filtering. The collaborative filtering method based on the users/projects is to use the experience calculation of the users to the projects to find similar users/projects, and then use the similar users/projects to recommend, for example, a calculation method based on the ratio similarity value is provided, and the recommendation is carried out by comparing the service currently used by the users and the attribute value of the similar service, for example, a collaborative filtering algorithm based on the preference of the users, and the method generally has good recommendation speed but insufficient accuracy; the collaborative filtering based on the model is to learn a model from the existing scores by utilizing a machine learning and statistical method, and predict the item scores by utilizing the model so as to obtain item recommendations, wherein the method has relatively good performance, but common problems are difficult to understand or some potential factors cannot be well explained; secondly, a recommendation method based on contents, which has the following idea: and recommending the items which are similar to the history record items to the user to the highest degree in the unused items by using the description of the essential characteristics of the items and the history record of the user. The method has the defect that more resources are consumed for extracting, analyzing, quantifying and expressing the cloud service attributes. And thirdly, a recommendation method based on the association rule. The association rule recommendation algorithm does not need domain knowledge and can discover new interest points, but the method is difficult in rule extraction and low in personalization degree.
Disclosure of Invention
The embodiment of the invention provides a cloud service personalized recommendation method based on a sparse linear method, and aims to provide a cloud service personalized recommendation method which is flexible in explanation and accurate in recommendation result.
In order to achieve the above object, the present invention provides a cloud service personality recommendation system based on a sparse linear method, including:
the system comprises a cloud client, a cloud evaluation center and a cloud recommendation center; the cloud client communicates with the cloud evaluation center and the cloud recommendation center, wherein,
the cloud client side: the cloud service recommendation system is used for submitting personalized requirements to the cloud recommendation center and feeding back cloud service evaluation to the cloud evaluation center based on used cloud service;
cloud evaluation center: analyzing the cloud service evaluation fed back by the cloud user, acquiring the score of the cloud service device under the personalized requirement, and updating the data to an evaluation database;
and the cloud recommendation center is used for recommending cloud services to the cloud users based on the evaluation data in the evaluation database aiming at the personalized demands submitted by the cloud users.
In order to achieve the above object, the present invention provides a cloud service personality recommendation method based on a sparse linear method, including the following steps:
s1 cloud user U for receivingiSubmitted personalized requirements PkNamely, selecting a plurality of most important attributes;
s2 cloud user U-basediPredicting the personalized demand P of the cloud user in the historical evaluation data in the evaluation databasekA score for each cloud service J is given;
s3 cloud user UiRecommending the highest-scoring cloud service.
Further, the step S2 specifically includes the following steps:
s21, computing cloud user UiFor scored cloud service JhIn the personalized demand PkCalculating a scoring deviation matrix D according to the scoring deviation between the lower scoring matrix D and the scoring deviation without individual requirements;
s22, predicting cloud users U based on the scoring deviation matrix D and the sparse aggregation matrix WiIn the personalized demand PkScoring cloud service J, the scoring of cloud service J comprising: scored cloud service JhScored and unscored cloud service JwThe score of (1).
Further, if the cloud user U in step S221 is detectediTargeting multiple cloud services J without personalization requirementshScoring, and then adopting a plurality of cloud services JhAverage of scoresMean value as cloud user UiRating R without individualized requirementi,h。
Further, in step S221, if there is no cloud user UiScoring cloud service J without personalized demandhThe scoring adopts a cloud user UiScoring cloud service J under all personalized demandshThe average value of the scores of (1) is taken as UiScoring cloud service J without personalized demandhScore R ofi,h。
Further, the user U is calculated based on the formula (1) and the formula (2)iIn the personalized demand PkDown to jth cloud service JjIs scored
Figure BDA0001863693350000031
The equations (1) and (2) are as follows:
Figure BDA0001863693350000032
Figure BDA0001863693350000033
wherein R isi,nIs a cloud user UiFor cloud service J without personalized requirementsnScore of (D)i.kIs a cloud user UiIn the personalized demand PkThe deviation between the grading of the next needle and the grading without individualized requirement, K is the number of the types of the individualized deviation, N is the number of the cloud services, pkPersonalizing a demand vector, W, for a usern,jServing cloud JnWith cloud service JjThe polymerization coefficient of (2).
The cloud service personalized recommendation method based on the sparse linear method has the following beneficial technical effects:
1. personalized requirements can be accurately and flexibly defined;
2. the deviation of the cloud user under each personalized demand and the deviation without the personalized demand are obtained through the learned related historical data, and prediction is carried out based on the deviation, so that the recommendation result is more accurate.
Drawings
Fig. 1 is a schematic structural diagram of a cloud service personalized recommendation system provided by an embodiment of the present invention;
fig. 2 is a flowchart of a cloud service personality recommendation method based on a sparse linear method according to an embodiment of the present invention;
FIG. 3 is a diagram of a MAP simulation experiment on a music data set according to an embodiment of the present invention;
FIG. 4 is a diagram of a MAP simulation experiment on a cine data set according to an embodiment of the present invention;
FIG. 5 is a diagram of a MAP simulation experiment on a restaurant dataset according to an embodiment of the present invention;
FIG. 6 is a simulation experiment diagram of Recall on a music data set according to an embodiment of the present invention;
FIG. 7 is a diagram of a Recall simulation experiment on a movie data set according to an embodiment of the present invention;
FIG. 8 is a graph of a Recall simulation experiment on a restaurant dataset according to an embodiment of the present invention;
FIG. 9 is a diagram of a Precision simulation experiment on a music data set according to an embodiment of the present invention;
FIG. 10 is a diagram of a Precision simulation experiment on a movie data set according to an embodiment of the present invention;
FIG. 11 is a diagram of a Precision simulation experiment on a restaurant data set according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The patent refers to the field of 'electric digital data processing'.
1) Personalized recommendation: the personalized recommendation is carried out according to the interest preference of the user, and when the cloud user selects the cloud service, the functional attribute of the cloud service is generally considered at first, and the personalized requirement of the cloud user is the requirement on the functional attribute of the cloud service, so that at least 5 functional attributes of the cloud service are provided, wherein the 5 functional attributes are as follows:
reliability r (reliability) the ability to operate without failure within a prescribed time and to recover service within a prescribed time after failure;
availability a (availability) the ability to continue to provide specific functionality in a defined environment;
security s (security) the ability to provide data confidentiality and integrity;
real-time e (real-time) ability to respond to requests within a specified time;
maintainability m (maintainability) ability to be easily modified and improved;
the method defines the personalized requirements of the cloud user as a plurality of cloud service attributes which are most concerned by the cloud user when the cloud user uses the cloud service, the cloud service attributes have importance degrees, and the personalized requirements are described by using two cloud service attributes to form a personalized requirement class example, for example: the combination mode is used because the emphasis of more cloud users is considered.
Fig. 1 is a schematic structural diagram of a cloud service personalized recommendation system provided in an embodiment of the present invention, and for convenience of description, only relevant parts of the embodiment of the present invention are shown.
The cloud service personalized recommendation system comprises: the system comprises a cloud client, a cloud evaluation center and a cloud recommendation center; the cloud client communicates with the cloud evaluation center and the cloud recommendation center, wherein,
the cloud client is used for submitting personalized requirements to the cloud recommendation center and feeding back cloud service evaluation to the cloud evaluation center based on used cloud service;
cloud evaluation center: analyzing the cloud service evaluation fed back by the cloud user, acquiring the score of the cloud service device under the personalized requirement, and updating the data to an evaluation database;
historical scores of cloud users for the cloud services are stored in the database, and the historical scores comprise: and updating the historical scores of the cloud user for the same cloud service under the same personalized requirement based on the latest scores of the cloud user for the same cloud service under the certain personalized requirement at present.
The cloud recommendation center is used for recommending cloud services to the cloud users based on the evaluation data in the evaluation database aiming at the personalized demands submitted by the cloud users;
the cloud recommendation center is used for acquiring the historical scores of the cloud users in the database based on the personalized requirements submitted by the cloud users.
Fig. 2 is a flowchart of a cloud service personality recommendation method based on a sparse linear method according to an embodiment of the present invention, where the cloud service personality recommendation method of a cloud recommendation center specifically includes the following steps:
s1 cloud user U for receivingiSubmitted personalized requirements PkNamely, selecting a plurality of most important attributes;
s2 cloud user U-basediHistorical evaluation data in an evaluation database are used for acquiring the personalized demand P of the cloud userkA score for each cloud service J is given;
s3 cloud user UiRecommending the highest-scoring cloud service.
In the embodiment of the present invention, step S2 specifically includes the following steps:
s21, computing cloud user UiFor scored cloud service JhIn the personalized demand PkCalculating a scoring deviation matrix D according to the scoring deviation between the lower scoring matrix D and the scoring deviation without individual requirements;
because the data set evaluated by the cloud user is sparse, that is, the same cloud user cannot be guaranteed to score other cloud services under the same personalized requirement, so that no method is available for obtaining the scores of the cloud user on other cloud services under the same personalized requirement, and personalized score deviation is introduced to solve the problem. The personalized score deviation means a deviation between scores of cloud users without personalized demands and scores of cloud users with personalized demands, namely, a deviation is considered to exist between each pair of < cloud users, personalized demands > and scores of cloud users without personalized demands.
As shown in table 1, the score of a certain cloud service without personalized demand of a certain user is 4, but when the scores of the cloud service with the following two personalized demands are 3 and 5, respectively, this indicates that there is a deviation between the scores of the user without personalized demand and with personalized demand.
TABLE 1 user Scoring example with and without personalization requirements
User' s Cloud service Scoring Personalized requirements
u1 c1 4 Is free of
u1 c1 3 Real-time and reliable
u1 c1 5 Safety and real-time performance
The scoring deviation matrix D is an nxk two-dimensional matrix and is used for representing personalized scoring deviations, each row in the scoring deviation matrix D represents the scoring deviation of a cloud user under different personalized requirements and without personalized requirements, and each column represents the scoring deviation of all users under the personalized requirements and without personalized requirements, as shown in table 2;
TABLE 2 personalized Scoring deviation matrix D
Figure BDA0001863693350000071
In the embodiment of the invention, if the cloud user U does not exist in the databasenIn the personalized demand PmLower history data, i.e. cloud user UnIn the personalized demand PkIf no cloud service is scored, randomly recommending each cloud service to the user;
in the embodiment of the present invention, if the cloud user U in step S221 is detectediJ for scored cloud service without personalized demandhIf the number of the scores is multiple, the average value of the multiple scores is used as the cloud user UiRating R without individualized requirementi,h(ii) a If the cloud user U does not existiJ for scored cloud service without personalized demandhThe scoring adopts a cloud user UiScoring cloud service J under all personalized demandshThe average value of the scores of (1) is taken as UiScoring cloud service J without personalized demandhScore R ofi,h。
S22, predicting cloud users U based on the scoring deviation matrix D and the sparse aggregation matrix WiIn the personalized demand PkScoring cloud service J, the scoring of cloud service J comprising: scored cloud service JhScored and unscored cloud service JwThe score of (1).
The sparse aggregation matrix W is a non-negative matrix of N × N, and represents aggregation coefficients between cloud services, similar to a similarity matrix between items, and each row and each column in the sparse aggregation matrix represents an aggregation coefficient between cloud services.
Calculating user U based on formula (1) and formula (2)iIn the personalized demand PkDown to jth cloud service JjIs scored
Figure BDA0001863693350000072
The equations (1) and (2) are as follows:
Figure BDA0001863693350000073
Figure BDA0001863693350000074
wherein R isi,nIs a cloud user UiFor cloud service J without personalized requirementsnScore of (D)i.kIs a cloud user UiIn the personalized demand PkThe score is different from the score without individual requirement, K is the variety number of the individual difference, N is the number of the cloud service, pkPersonalizing a demand vector, W, for a usern,jServing cloud JnWith cloud service JjThe polymerization coefficient of (a) to (b),
Figure BDA0001863693350000081
is a cloud user UiIn the personalized demand PkDown-to-cloud service JnThe median score of (1).
For the understanding of the formula (2), the intermediate scores of the cloud user on other cloud services under the same personalized demand are calculated based on the scores and personalized score deviations of the cloud user on a certain cloud service under the personalized demand. In addition, in the patent, even if the cloud user scores part of the cloud services under a certain personalized requirement, the intermediate score of the cloud user on the scored cloud services under the same personalized requirement is calculated based on the formula (2). Because more data in the matrix D can be learned, the final prediction score is more accurate, and better recommendation effect is obtained.
Understanding formula (1), compute cloud user UiIn the personalized demand PkDown-to-cloud service JnIs scored
Figure BDA0001863693350000083
Namely, by using the cloud user UiIn the personalized demand PkPair-down cloud removal service JnThe extra cloud service intermediate score is obtained by multiplying the aggregation coefficient W, that is, calculated based on the formula (1).
pkThe binary vector representing the personalized requirements of the cloud user is p (0, 0, 0, 0) according to the combination condition of the personalized requirements, for example, if there are four conditions, when the personalized requirements are the third condition, p (0, 0, 1, 0) will increase with the increase of the combinations, and if the combination mode changes, p will also change correspondingly. For example, if three or four p (0, 0, 1, 1) are present.
Since the row and column of the personalized deviation matrix represent the user (u) and the personalized demand (p), the real personalized score of the user is used as a training set, the parameters in the matrixes D and W are learned from data, and the predicted score for the cloud service only depends on four factors: the system comprises a scoring matrix R, an individualized deviation matrix D, a project aggregation coefficient matrix W and a user individualized demand vector p under the condition of no individualized demand.
The solution of matrices D and W can be transformed into a minimum problem of the regularized optimization problem, which can be calculated using equation (4). Wherein | | W | | ceilingFIs the Frobenius paradigm (l) of the matrix WfNormal form), | W | count1Is l of the matrix W1Normal form, | D | luminanceFAnd W1The same process is carried out; θ and β are learning rates.
Figure BDA0001863693350000082
Equation (3) can be optimized and solved by using a Stochastic Gradient Descent (SGD), in which the parameters are updated according to equations (4) to (6):
Figure BDA0001863693350000091
Figure BDA0001863693350000092
Figure BDA0001863693350000093
wherein the parameter alpha1，β1，λ1，α2，β2，λ2Are all learning rates for Di,kWe only update the parameters at position k, and when pkUpdating when the value is 1; and Wh,jWe only update the parameter h of the relevant location, i.e. the set of cloud services that the user has scored.
The cloud service personalized recommendation method based on the sparse linear method has the following beneficial technical effects: 1. personalized requirements can be accurately and flexibly defined; 2. the deviation of the cloud user under each personalized demand and the deviation without the personalized demand are obtained through the learned related historical data, and prediction is carried out based on the deviation, so that the recommendation result is more accurate.
In order to verify the performance of the method proposed by the patent, a simulation experiment is compared with a reference algorithm from three standards of precision (precision), recall (recall) and average precision average (MAP). The benchmark algorithm is two recommended algorithms that are more commonly used: CF and CASA. The precision rate and the recall rate are relatively popular measurement standards in an evaluation recommendation system, C is defined as a recommendation list obtained on a test set, D is defined as a recommendation list obtained on a training set, and I is equal to C and D, namely I is a correct service recommendation result obtained based on a recommendation algorithm. Based on these, the calculation formulas of precision and recall are (7) and (8), respectively. MAP is another commonly used ranking indicator that takes into account the recommendation level of the ranked items. The MAP can be calculated by equation (9).
Figure BDA0001863693350000094
Figure BDA0001863693350000095
Figure BDA0001863693350000096
Where M represents the number of cloud users, N represents the length of the recommendation list, p (k) represents the precision at k in the recommendation list, and M represents the number of related items.
The following sets of simulation graphs compare the method presented herein with the performance of the two common recommendation algorithms CF and CASA on three data sets, evaluated on the criteria of precision (precision), recall (recall) and mean precision average (MAP) mentioned above.
Users also typically have personalized needs for them when selecting music, movies, and restaurants, such as the type of music, movies, and restaurants, so three data sets of music, movies, and restaurants are used herein to simulate a cloud service data set, as in table 3. The three different data sets are used in order to verify the performance of the method herein on the different data sets.
TABLE 3 data set
Figure BDA0001863693350000101
Simulation test results based on the above three datasets fig. 3-11 show that the method proposed herein above is generally superior to the baseline method. While the method presented herein is inferior in some respects to the baseline method, such as accuracy on a movie data set, the method presented herein is inferior to the baseline method in the top 10 recommendations, but superior to the baseline method after 10; also, the accuracy on the restaurant dataset is less than the CF method after 20, but better than the baseline method in the first 20, except that the method herein is better than the baseline method. The performance on the music data set is particularly outstanding, because the music data set is dense, that is, scoring exists on the < user, item > pairs under different personalized requirements, so that the scoring deviation can be better learned by the deviation matrix, and the recommendation effect is more accurate. Meanwhile, the method performs well on the MAP and is relatively more superior to the reference method. The model proposed herein therefore works better in personalized recommendations
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.

Claims (4)

1. A cloud service personalized recommendation system based on a sparse linear method is characterized by comprising the following steps:
the system comprises a cloud client, a cloud evaluation center and a cloud recommendation center; the cloud client communicates with the cloud evaluation center and the cloud recommendation center, wherein,
the cloud client side: the cloud service recommendation system is used for submitting personalized requirements to the cloud recommendation center and feeding back cloud service evaluation to the cloud evaluation center based on used cloud service;
cloud evaluation center: analyzing the cloud service evaluation fed back by the cloud user, acquiring the score of the cloud service under the personalized demand, and updating the data to an evaluation database;
the cloud recommendation center is used for recommending cloud services to the cloud users based on the evaluation data in the evaluation database aiming at the personalized demands submitted by the cloud users;
the method comprises the following steps:
s1 cloud user U for receivingiSubmitted personalized requirements PkNamely, selecting a plurality of most important attributes;
s2 cloud user U-basediPredicting the personalized demand P of the cloud user in the historical evaluation data in the evaluation databasekFor each cloud service JGrading;
s3 cloud user UiRecommending the highest-grade cloud service;
the step S2 specifically includes the following steps:
s21, computing cloud user UiFor scored cloud service JhIn the personalized demand PkCalculating a scoring deviation matrix D according to the scoring deviation between the lower scoring matrix D and the scoring deviation without individual requirements;
s22, predicting cloud users U based on the scoring deviation matrix D and the sparse aggregation matrix WiIn the personalized demand PkScoring cloud service J, the scoring of cloud service J comprising: scored cloud service JhScored and unscored cloud service JwThe score of (1).
2. The cloud service personalized recommendation system based on the sparse linear method as claimed in claim 1, wherein if the cloud user U in step S221 is the cloud user UiTargeting multiple cloud services J without personalization requirementshScoring, and then adopting a plurality of cloud services JhTaking the average value of the scores as the U of the cloud useriRating R without individualized requirementi,h。
3. The cloud service personalized recommendation system based on the sparse linear method as claimed in claim 2, wherein in step S221, if there is no cloud user UiScoring cloud service J without personalized demandhThe scoring adopts a cloud user UiScoring cloud service J under all personalized demandshThe average value of the scores of (1) is taken as UiScoring cloud service J without personalized demandhScore R ofi,h。
4. The cloud service personalized recommendation system based on the sparse linear method as claimed in claim 1, wherein the user U is calculated based on formula (1) and formula (2)iIn the personalized demand PkDown to jth cloud service JjIs scored
Figure FDA0003073313690000021
The equations (1) and (2) are as follows:
Figure FDA0003073313690000022
Figure FDA0003073313690000023
wherein R isi,nIs a cloud user UiFor cloud service J without personalized requirementsnScore of (D)i.kIs a cloud user UiIn the personalized demand PkThe deviation between the grading of the next needle and the grading without individualized requirement, K is the number of the types of the individualized deviation, N is the number of the cloud services, pkPersonalizing a demand vector, W, for a usern,jServing cloud JnWith cloud service JjThe polymerization coefficient of (2).
CN201811345729.7A 2018-11-13 2018-11-13 Cloud service personality recommendation system and method based on sparse linear method Active CN109255079B (en)

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 CN109255079A (en) 2019-01-22
CN109255079B true 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)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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

Citations (6)

* Cited by examiner, † Cited by third party
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
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

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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

Patent Citations (6)

* Cited by examiner, † Cited by third party
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
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

Also Published As

Publication number Publication date
CN109255079A (en) 2019-01-22

Similar Documents

Publication Publication Date Title
US11748379B1 (en) Systems and methods for generating and implementing knowledge graphs for knowledge representation and analysis
US20200257543A1 (en) Aggregate Features For Machine Learning
Kumar et al. Social popularity based SVD++ recommender system
US10380498B1 (en) Platform services to enable one-click execution of the end-to-end sequence of modeling steps
US9390142B2 (en) Guided predictive analysis with the use of templates
CN110580649B (en) Method and device for determining commodity potential value
CN109033101B (en) Label recommendation method and device
CN107895038B (en) Link prediction relation recommendation method and device
US10157351B1 (en) Persona based data mining system
CN105740312A (en) Clustering database queries for runtime prediction
US10769175B1 (en) Real-time hosted system analytics
US20140006166A1 (en) System and method for determining offers based on predictions of user interest
US10331681B1 (en) Crowdsourced evaluation and refinement of search clusters
CN104077357A (en) User based collaborative filtering hybrid recommendation method
US10089675B1 (en) Probabilistic matrix factorization system based on personas
CN109460519B (en) Browsing object recommendation method and device, storage medium and server
US11762819B2 (en) Clustering model analysis for big data environments
CN111966886A (en) Object recommendation method, object recommendation device, electronic equipment and storage medium
CN103309869A (en) Method and system for recommending display keyword of data object
CN108920647B (en) Low-rank matrix filling TOP-N recommendation method based on spectral clustering
WO2023284516A1 (en) Information recommendation method and apparatus based on knowledge graph, and device, medium, and product
CN115344698A (en) Label processing method, label processing device, computer equipment, storage medium and program product
US8510307B1 (en) Systems and methods for automatic item classification
US11109085B2 (en) Utilizing one hash permutation and populated-value-slot-based densification for generating audience segment trait recommendations
EP4298560A1 (en) Automated machine learning to generate recommendations for websites or applications

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