CN111553401A - QoS prediction method based on graph model and applied to cloud service recommendation - Google Patents

QoS prediction method based on graph model and applied to cloud service recommendation Download PDF

Info

Publication number
CN111553401A
CN111553401A CN202010322193.8A CN202010322193A CN111553401A CN 111553401 A CN111553401 A CN 111553401A CN 202010322193 A CN202010322193 A CN 202010322193A CN 111553401 A CN111553401 A CN 111553401A
Authority
CN
China
Prior art keywords
qos
similarity
user
services
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.)
Granted
Application number
CN202010322193.8A
Other languages
Chinese (zh)
Other versions
CN111553401B (en
Inventor
丁丁
畅振华
李浥东
夏有昊
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.)
Beijing Jiaotong University
Original Assignee
Beijing Jiaotong 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 Beijing Jiaotong University filed Critical Beijing Jiaotong University
Priority to CN202010322193.8A priority Critical patent/CN111553401B/en
Publication of CN111553401A publication Critical patent/CN111553401A/en
Application granted granted Critical
Publication of CN111553401B publication Critical patent/CN111553401B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides a graph model-based QoS prediction method applied to cloud service recommendation. The method comprises the following steps: constructing a full graph model containing multi-source information, wherein the full graph model comprises nodes representing users and services, and edges using the similarity among the users, the similarity among the services and the similarity among the users and the services as weights; dividing the full graph model into a plurality of sub graph models; respectively carrying out optimized probability matrix decomposition algorithm on the full graph model and the sub-graph model to obtain globally and locally predicted QoS; and carrying out self-adaptive fusion processing on the globally and locally predicted QoS to obtain the final predicted QoS. The method of the invention fully considers the influence of multi-source information on QoS, and adaptively fuses local and global characteristics to improve the QoS prediction precision. The missing QoS value can be accurately predicted, the sparse matrix is filled, the density degree of the matrix is improved, and the problem of sparse QoS in the field of cloud service recommendation is solved to a certain extent.

Description

QoS prediction method based on graph model and applied to cloud service recommendation
Technical Field
The invention relates to the technical field of computer application, in particular to a graph model-based QoS prediction method applied to cloud service recommendation.
Background
SOA (Service-Oriented Architecture) is widely used in distributed computing environments, especially in the field of cloud computing, and as a core part of the SOA, services (services) is a popular way to provide configurable functions. In this case, many redundant services with similar functions but different QoS (Quality of service) appear. Facing these huge amounts of services, users encounter great problems in how to select a suitable cloud service. The cloud service recommendation system aims to solve the problems that the number of cloud services is large, and users cannot quickly select appropriate services due to information overload. With the same functionality, the key to how to provide the proper service to the user is QoS. QoS is a set of attributes used to describe cloud service non-functionality including response time, throughput, reputation, failure rate of service invocation, stability, etc. QoS values may vary widely among different Web services due to changes in network environmental conditions and the different network environments in which users using the services are located. Due to the large number of services and the high calling cost, the QoS obtained by the personal user is extremely sparse, and the cloud service recommendation directly according to the QoS is not feasible. Therefore, the core of a non-functional based cloud service recommendation system can be regarded as prediction of QoS, and a proper service is recommended to a user according to the QoS lacking in prediction.
Currently, there are many scholars working on QoS prediction. Initially, scholars use static methods to predict QoS. Static methods use arithmetic mean for prediction, calculating the average QoS values including user and service from global respectively. The methods are simple and easy to implement, the situation perception factors of users and services do not need to be considered, but the static methods cannot reflect the dynamic characteristics of QoS (quality of service), and the obtained results are often inaccurate.
Inspired by the traditional recommendation system, many scholars apply a Collaborative Filtering (CF) algorithm in the recommendation system to the cloud service recommendation. The CF algorithm utilizes the history records of the calling service of the user, finds the similarity of experience among users or the similarity among services through the records, and predicts the QoS by utilizing the similarity. CF algorithms fall into two categories, memory-based and model-based. The researchers propose to use the collaborative filtering based on the user and the collaborative filtering based on the service to predict the QoS, and the prediction method based on the memory is applied to the field of web services service recommendation; however, these are only studied on QoS, and do not consider context information of users and services, so the predicted result has a certain limitation; then, the scholars begin to consider using the geographical position information to cluster the services and the users, and the QoS prediction is carried out by using a mixed matrix fusion mode after clustering, so that the accuracy of the QoS prediction is improved to a certain extent. Although the above studies made some improvements to the QoS prediction model, much context information was added; but does not consider how to efficiently integrate the multi-source information on the basis of maintaining the original contact of the user and the service. Meanwhile, when local information and global information are processed, a linear parameter adjusting mode is often adopted for fusion, and the method usually takes more energy on determining proper parameters and is not robust. Likewise, they are not sufficiently concerned with solving the cold start problem, and often choose to ignore these problems.
Fig. 1 shows an example of four users invoking five services: previous research is to extract unilateral characteristics to perform clustering operation on users and services, so that four users are divided into two types, five services are divided into two piles, and the users and services in the group represent neighbors closely related to the users and services. However, this practice often ignores the call records generated by the user calling the service, which is the most important relationship between the user and the service (as shown by the dashed lines in fig. 1).
Therefore, on the premise of considering how to keep close contact between users and services in the process of predicting the QoS, the method deeply excavates the contact among the multi-source information, and fully utilizes the contact to improve the accuracy of QoS prediction, thereby forming a problem to be solved urgently.
Disclosure of Invention
The embodiment of the invention provides a graph model-based QoS prediction method applied to the field of cloud service recommendation, and aims to overcome the problems in the prior art.
In order to achieve the purpose, the invention adopts the following technical scheme.
A graph model-based QoS prediction method applied to cloud service recommendation comprises the following steps:
constructing a full graph model containing multi-source information, wherein the full graph model comprises nodes representing users and services, and edges using the similarity among the users, the similarity among the services and the similarity among the users and the services as weights;
dividing the full graph model into a plurality of sub graph models;
respectively carrying out optimized probability matrix decomposition algorithm on the full graph model and the sub-graph model to obtain globally and locally predicted QoS;
and carrying out self-adaptive fusion processing on the globally and locally predicted QoS to obtain the final predicted QoS.
Preferably, the building of a full graph model containing multi-source information, the full graph model including nodes representing users and services, and edges using similarity between users, similarity between services, and similarity between users and services as weights includes:
calculating the similarity W between the users on the geographical position according to the longitude and latitude information of the userslo(ui，uj)；
Calculating the similarity W between the user and the user in terms of network position by using the autonomous system information of the userAS(ui，uj)；
Obtaining user u by comprehensively considering longitude and latitude information of user and autonomous system informationiAnd user ujFinal similarity between them
Figure RE-RE-GDA0002535305760000037
As shown in equation (1):
Figure RE-RE-GDA0002535305760000031
η therein1Is a given threshold value, when the final similarity value is greater than η1Then it will be at user uiAnd user ujConstruct a weight of
Figure RE-RE-GDA0002535305760000033
Edge of (1), other cases user uiAnd user ujThere is no edge connection between them; lambda [ alpha ]1And λ2Respectively representing the weight of the geographical position information and the autonomous system information of the user in the final similarity;
calculating semantic similarity W between services according to WSDL information of the servicesws(si，sj)；
Computing similarity W between services and services in terms of network location using autonomous system information of the serviceAS(si，sj)；
Obtaining service s by comprehensively considering WSDL information and autonomous system information of serviceiAnd service sjFinal similarity between them
Figure RE-RE-GDA0002535305760000034
As shown in equation (2):
Figure RE-RE-GDA0002535305760000035
η therein2Is a given threshold value, when the final similarity value is greater than η2Is served by the time siAnd service sjConstruct a weight of
Figure RE-RE-GDA0002535305760000036
Edge of (1), other case service siAnd service sjThere is no edge connection between them; gamma ray1And gamma2And respectively representing the weight of the WSDL information and the autonomous system information of the service in the final similarity.
The similarity between the user and the service is obtained by using the QoS prediction matrix through the following transformation:
Eui，sj＝rti，j
rtmaxj＝max(rtij|i＝1，2，...，m)
rtminj＝min(rtij|i＝1，2，...，m)
rtmax＝(rtmax1，rtmax2，...，rtmaxn)
rtmin＝(rtmin1，rtmin2，...，rtminn)
Figure RE-RE-GDA0002535305760000032
the service call is first normalized and then the normalized RT is converted to a similarity using equation (3), where rtmaxjAnd rtminjRespectively representing services sjAnd rtmax and rtmin represent the maximum and minimum values of all services, respectively, Im×1A one-dimensional column vector representing m users;
respectively taking the similarity between users, the similarity between services and the similarity between users and services as the weight of three edges, taking the users and the services as nodes, and constructing a full graph model containing the users, the services and the three edges: g ═ V, E }, where V ═ U, S }, U ═ U }, and1，u2，...，umis and S ═ S1，s2，...，snRespectively representing that m users and n services are contained; the set of edges is represented as: e ═ Euu，Ess，EusIn which Euu，Ess，EusThe edges using the inter-user, inter-service, and user-service similarities as weights are respectively indicated.
Preferably, the building of a full graph model containing multi-source information, the full graph model including nodes representing users and services, and edges using similarity between users, similarity between services, and similarity between users and services as weights includes:
calculating the similarity W between the users on the geographical position according to the longitude and latitude information of the userslo(ui，uj)：
Figure RE-RE-GDA0002535305760000041
Figure RE-RE-GDA0002535305760000042
dis(ui，uj) Representing user uiAnd user ujThe Euclidean distance between the users is converted into the similarity W of the users on the geographical positions by using the formula (5)lo(ui，uj)，xiAnd yiRespectively represent users uiLongitude and latitude information of the geographical location, xjAnd yjRespectively represent users ujLongitude and latitude information of the geographic location.
Calculating user u by formula (6) using the autonomous system information of the useriAnd user ujSimilarity W in network location betweenAS(ui，uj)：
Figure RE-RE-GDA0002535305760000043
The method comprises the steps of performing text removal processing on WSDL information, removing structured format information, extracting feature words in the WSDL information, and converting special keywords in the WSDL information into semantic similarity between services by using a tf-idf algorithm in the field of natural language processing through the following two formulas:
Figure RE-RE-GDA0002535305760000044
Figure RE-RE-GDA0002535305760000045
in formula (7), M is the total number of web pages searched from Google using the feature words x and y, and logf (x) and logf (y) are the number of clicks searched using the feature words x and y, respectively; f (x, y) represents the number of web pages using both the feature words x and y, in equation (8)
Figure RE-RE-GDA0002535305760000046
And
Figure RE-RE-GDA0002535305760000047
respectively representing services si，sjThe feature word vectors of (a) are,
Figure RE-RE-GDA0002535305760000048
representing the cardinality of the vector. Wws(si，sj) Representation service siAnd service sjSemantic similarity in WSDL information.
Calculating service s by formula (9) using autonomous system information where service is locatediAnd service sjSimilarity in network location WAS(si，sj)：
Figure RE-RE-GDA0002535305760000051
Preferably, the dividing the full graph model into a plurality of sub-graph models includes:
converting the full graph model G into a similarity matrix containing the similarity between the user and the service;
constructing an adjacent matrix according to the similarity matrix, and constructing a degree matrix at the same time;
constructing a Laplace matrix according to the adjacent matrix and the degree matrix;
normalizing the laplacian matrix;
calculating minimum K eigenvalues and eigenvectors of the Laplace matrix;
combining K eigenvectors into an eigenvector matrix, and standardizing the eigenvector matrix according to rows;
clustering the normalized feature matrix to obtain a set of the segmented K sub-graphs
Figure RE-RE-GDA0002535305760000052
Where K denotes the number of subgraphs, Sub1Denotes the 1 st Sub-figure, SubKRepresenting the K-th sub-graph.
Preferably, the obtaining global and local predicted QoS by the probability matrix decomposition algorithm for optimizing the full graph model and the sub-graph model respectively includes:
the optimized probability matrix decomposition algorithm is set as follows:
Figure RE-RE-GDA0002535305760000053
r represents the final predicted QoS value,
Figure RE-RE-GDA0002535305760000054
represents RijObedience mean 0 and variance σRIs normally distributed.
μBMean deviation of all QoS, BU mean deviation of QoS generated by a user invoking a service, BS mean deviation of QoS generated by a service invoked by a user, and three mean deviations are defined as follows:
Figure RE-RE-GDA0002535305760000055
Figure RE-RE-GDA0002535305760000056
Figure RE-RE-GDA0002535305760000061
where | R | represents the number of all Qos, R(i)Representing user uiSet of invoked services, R(j)Indicating that a service s has been invokedjA set of users of (1);
converting QoS in the full graph model into a matrix form, decomposing the matrix according to the optimized probability matrix decomposition algorithm to respectively obtain a user hidden factor vector and a service hidden factor vector, multiplying the user hidden factor vector and the service hidden factor vector to obtain predicted QoS, and multiplying the predicted QoS and the muBBU and BS are added to obtain globally predicted QoS;
converting QoS in the sub-graph model into a matrix form, decomposing the matrix according to the optimized probability matrix decomposition algorithm to respectively obtain a user hidden factor vector and a service hidden factor vector, multiplying the user hidden factor vector and the service hidden factor vector to obtain predicted QoS, and multiplying the predicted QoS and muBThe BU and the BS are added to obtain the locally predicted QoS.
Preferably, the adaptively fusing the globally and locally predicted QoS to obtain the final predicted QoS includes:
based on locally and globally predicted QoS, a gaussian mixture model is constructed, obeying the gaussian model, using the following equations (12) and (13):
Figure RE-RE-GDA0002535305760000062
Figure RE-RE-GDA0002535305760000063
where p (x | t) ═ N (x | u)t，∑t) A function of a gaussian model representing the t-th gaussian distribution, the parameters satisfying:
Figure RE-RE-GDA0002535305760000064
and is
Figure RE-RE-GDA0002535305760000065
Figure RE-RE-GDA0002535305760000066
And
Figure RE-RE-GDA0002535305760000067
respectively representing the proportion of the local predicted QoS and the global predicted QoS in the final predicted QoS;
Figure RE-RE-GDA0002535305760000068
and
Figure RE-RE-GDA0002535305760000069
respectively representing the mean and variance in Gaussian models in local and global matrix decomposition;
and adaptively fusing local and global predicted QoS based on the Gaussian mixture model to obtain final predicted QoS:
Figure RE-RE-GDA00025353057600000610
Figure RE-RE-GDA00025353057600000611
Figure RE-RE-GDA00025353057600000612
wherein
Figure RE-RE-GDA00025353057600000613
And
Figure RE-RE-GDA00025353057600000614
respectively represent locally predicted QoS and globally predicted QoS, and1and2local and global predicted QoS respectively represent the respective proportions of the local and global predicted QoS in the final predicted QoS,
Figure RE-RE-GDA0002535305760000071
SubBU,SubBSrespectively represents the average deviation of QoS, user uiMean deviation of invoked local services and service sjAverage deviation of all local user calls; gμB,GBU,GBSRespectively expressed in the global matrix, average deviation of QoS, user uiMean deviation of all services invoked and service sjIs calculated from the average deviation of all local user invocations.
It can be seen from the technical solutions provided by the embodiments of the present invention that the present invention provides a concept of a full map to integrate various multi-source information. The full-graph model not only reflects the relationship between users, but also extracts the implicit relationship between services, and simultaneously fully utilizes the QoS matrix information to enhance the relationship between the users and the services. By cutting through the whole graph, the noise information which is not closely related is weakened, and the cold start problem can be relieved. The influence of multi-source information on the QoS is fully considered, and local and global features are adaptively fused to improve the QoS prediction accuracy. The embodiment of the invention can accurately predict the missing QoS value, further fill the sparse matrix, improve the density degree of the matrix and solve the problem of sparse QoS in the field of cloud service recommendation to a certain extent.
Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a schematic diagram illustrating a four-user invocation of five services provided in the prior art;
fig. 2 is a schematic diagram illustrating an implementation principle of a graph model-based QoS prediction method in cloud service recommendation according to an embodiment of the present invention;
fig. 3 is a processing flow chart of a QoS prediction method based on a graph model in cloud service recommendation according to an embodiment of the present invention;
fig. 4 is a schematic diagram of obtaining sub-graphs by segmenting a full graph by using a spectral clustering idea according to an embodiment of the present invention;
fig. 5 is a schematic diagram of performing QoS prediction on a sub-graph and a full graph by using an optimized probability matrix decomposition algorithm according to an embodiment of the present invention;
FIG. 6 is a diagram illustrating a comparison between QoS prediction results of GMF models and other methods according to an embodiment of the present invention;
FIG. 7 is a diagram illustrating a comparison between QoS prediction results of a GMF model and two-part modules of the GMF model according to an embodiment of the present invention;
fig. 8 is a schematic diagram illustrating an analysis of an influence of the number K of subgraphs on a result when the sparsity is 5% according to the embodiment of the present invention;
fig. 9 is a schematic diagram illustrating an analysis of an influence of the number D of hidden factors of PMF matrix decomposition on a full graph and a sub graph on a result when sparsity is 5% according to an embodiment of the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention.
As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or coupled. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
For the convenience of understanding the embodiments of the present invention, the following description will be further explained by taking several specific embodiments as examples in conjunction with the drawings, and the embodiments are not to be construed as limiting the embodiments of the present invention.
The embodiment of the invention aims at the existing QoS prediction method, researches the QoS prediction problem based on matrix decomposition, and provides a novel graph model-based QoS prediction model GMF (graph based matrix factorization) based on adaptive matrix decomposition, wherein the model effectively integrates multi-source information of users and services, and meanwhile, the graph cutting algorithm adopted by the model can reduce the cold start problem. Based on the model, multi-source information can be effectively integrated, a matrix decomposition algorithm is further improved, and a better prediction result is obtained through the self-adaptive mixed matrix decomposition method on the graph model.
Fig. 2 is a schematic diagram illustrating an implementation principle of a graph model-based QoS prediction method applied to cloud service recommendation according to an embodiment of the present invention, where a processing flow of the method is shown in fig. 3, and the method includes the following processing steps:
step S1: constructing a full graph model containing multi-source information;
on the basis of the original QoS matrix, multi-source information which has a significant influence on QoS prediction is fully considered, and the multi-source information comprises structured WSDL (Web Services Description Language) text information, geographical location information of users and Services, network location information of the users and the Services, and the like. The multi-source information with significant influence on QoS prediction is quantized and converted into the similarity between users, the similarity between services and the similarity between users and services. And using the similarity as the weight of the edge, and using the user and the service as nodes to construct a full graph model.
Calculating the similarity of the user and the user on the geographical position by using the latitude and longitude information provided by the user:
Figure RE-RE-GDA0002535305760000091
Figure RE-RE-GDA0002535305760000092
wherein x and y respectively represent longitude and latitude information of the geographic position of the user; calculating the distance between users using the Euclidean formula through formula (1), and then converting the distance information into the similarity, dis (u) using formula (2)i，uj) Representing user uiAnd user ujOf between, islo(ui，uj) Representing user uiAnd user ujThe similarity in geographical location between them. In the formula (2), all values satisfy 0 < Wlo(ui，uj) 1, the larger the value, the higher the similarity.
Calculating user u by formula (3) using autonomous system information of useriAnd user ujSimilarity between them in terms of network location:
Figure RE-RE-GDA0002535305760000093
obtaining user u by comprehensively considering longitude and latitude information of user and autonomous system informationiAnd user ujFinal similarity between them
Figure RE-RE-GDA0002535305760000094
As shown in equation (4):
Figure RE-RE-GDA0002535305760000095
η therein1Is a given threshold value, when the final similarity value is greater than η1Then it will be at user uiAnd user ujConstruct a weight of
Figure RE-RE-GDA0002535305760000096
There is no edge connection between the user u _ i and the user u _ j in other cases, λ1And λ2Respectively representing the weight of the geographical position information and the autonomous system information of the user in the final similarity.
Semantic similarity between services is calculated using WSDL information:
firstly, the WSDL information is subjected to text removal processing to remove structured format information, and then feature words in the WSDL information are extracted, wherein the feature words comprise information such as port information (portType), service name (name), used data type (types), communication protocol (binding), message (message) and the like. Then, a termfrequency-inverse document frequency (tf-idf) algorithm in the natural language processing field is used for converting the special keywords in the WSDL information into the semantic similarity information of the service through the following two formulas.
Figure RE-RE-GDA0002535305760000101
Figure RE-RE-GDA0002535305760000102
In formula (5), M is the total number of web pages searched from Google using the feature words x and y, and logf (x) and logf (y) are the number of clicks searched using the feature words x and y, respectively; f (x, y) represents the number of web pages using both the feature words x and y. In the formula (6)
Figure RE-RE-GDA0002535305760000103
And
Figure RE-RE-GDA0002535305760000104
respectively representing services si，sjThe feature word vectors of (a) are,
Figure RE-RE-GDA0002535305760000105
representing the cardinality of the vector. Wws(si，sj) Representation service siAnd service sjIn WSDL informationSemantic similarity of aspects.
Calculating service s by equation (7) using autonomous system information where the service is locatediAnd service sjSimilarity between them in terms of network location:
Figure RE-RE-GDA0002535305760000106
obtaining service s by comprehensively considering WSDL information and autonomous system information of serviceiAnd service sjFinal similarity between them
Figure RE-RE-GDA0002535305760000107
As shown in equation (8):
Figure RE-RE-GDA0002535305760000108
γ1Wws(si，sj)+γ2WAS(si，sj)＞η2# (8)
η therein2Is a given threshold value, when the final similarity value is greater than η2Is served by the time siAnd service sjConstruct a weight of
Figure RE-RE-GDA0002535305760000109
Edge of (1), other case service siAnd service sjThere is no edge connection between them; gamma ray1And gamma2And respectively representing the weight of the WSDL information and the autonomous system information of the service in the final similarity.
The similarity between the user and the service is obtained by using the QoS prediction matrix through the following transformation:
Eui，sj＝rti，j
rtmaxj＝max(rtij|i＝1，2，...，m)
rtminj＝min(rtij|i＝1，2，...，m)
rtmax＝(rtmax1，rtmax2，...,rtmaxn)
rtmin＝(rtmin1，rtmin2，...，rtminn)
Figure RE-RE-GDA0002535305760000111
the service call is first normalized and then the normalized RT is converted to a similarity using equation (9). Wherein rtmaxjAnd rtminjRespectively representing services sjAnd rtmax and rtmin represent the maximum and minimum values of all services, respectively, Im×1A one-dimensional column vector representing m users.
Respectively taking the similarity between users, the similarity between services and the similarity between users and services as the weight of three edges, taking the users and the services as nodes, and constructing a full graph model containing the users, the services and the three edges: g ═ V, E }, where V ═ U, S }, U ═ U }, and1，u2，...，umis and S ═ S1，s2，...，snRespectively representing that m users and n services are contained; the set of edges is represented as: e ═ Euu，Ess，EusIn which Euu，Ess，EusThe edges using the inter-user, inter-service, and user-service similarities as weights are respectively indicated.
Step 2: dividing the full graph model into a plurality of sub graph models;
since there are edges in the original data where the user is not very closely associated with the service, and these edges do not have a great influence on the predicted QoS recommendation and even have a certain negative effect, the embodiment of the present invention regards them as noise data. In order to weaken the influence of the noise data on QoS prediction and enhance strong connection between users and services, the embodiment of the invention provides a method for cutting a complete full graph into a plurality of sub-graphs and weakening the side information of the noise. Thereby cutting a large graph G into one
Figure RE-RE-GDA0002535305760000112
Wherein K represents the number of subgraphs, and the specific graph cutting mode is as follows:
inputting: a full graph G;
and (3) outputting:
Figure RE-RE-GDA0002535305760000113
1) converting G into a similarity matrix containing the similarity between the user and the service;
2) constructing an adjacent matrix according to the similarity matrix, and constructing a degree matrix at the same time;
3) constructing a Laplace matrix according to the adjacent matrix and the degree matrix;
4) normalizing the laplacian matrix;
5) calculating minimum K eigenvalues and eigenvectors of the Laplace matrix;
6) combining K eigenvectors into an eigenvector matrix, and standardizing the eigenvector matrix according to rows;
7) clustering the normalized feature matrix to obtain K Sub-graphs Sub1，Sub2，…，SubK。
The size of K determines the number of users and services within the sub-graph. The larger K is, the fewer services and users in the sub-graph are, and the smaller K is, the more services and users in the sub-graph are. And when the K value is 8, the prediction precision is highest. Through a graph cutting mode, the users and the services are divided into the same sub-graph, and the users or the services of the same sub-graph are directly recommended to the users or the services without any calling record, so that the computing steps of the similarity among the users and the similarity among the services are reduced, and the cold start problem is relieved to a certain extent.
And step 3: and (3) respectively carrying out optimized Probability Matrix Factorization (PMF) on the full graph model and the sub-graph model to obtain the QoS of global and local prediction.
The probability matrix decomposition algorithm model is shown as formula (10):
Figure RE-RE-GDA0002535305760000121
wherein R represents the final predicted QoS value, U, S represent the implicit factor vector of the decomposed user and service respectively, sigmaRThe variance is indicated. The optimized probability matrix decomposition algorithm is as follows:
the optimized probability matrix is decomposed as follows:
Figure RE-RE-GDA0002535305760000122
the main improvement of equation (11) over equation (10) is in μBBU and BS, which respectively represent average deviation of all QoS, QoS average deviation generated by user invoking service and QoS average deviation generated by service being invoked by user. The three deviation terms are defined as follows:
wherein:
Figure RE-RE-GDA0002535305760000123
Figure RE-RE-GDA0002535305760000124
Figure RE-RE-GDA0002535305760000125
where | R | represents the number of all Qos, R(i)Representing user uiSet of invoked services, R(j)Indicating that a service s has been invokedjIs selected.
The present invention uses an improved probability matrix method that takes into account the effect of bias terms on the matrix decomposition, where μBMean value, BU, representing QoSiRepresenting user uiInvoking the generated mean, BS, of all servicesjRepresentation service sjThe user generated mean was invoked by all. Applying the improved PMF algorithm to the whole graph to obtain the globally predicted QoS, and applying the improved PMF algorithm to the whole graphAnd obtaining the locally predicted QoS on the sub-graph model.
Inputting: g (full map) or
Figure RE-RE-GDA0002535305760000131
(drawing)
And (3) outputting: QoS (quality of service)
1) Calculating QoS average deviation, QoS average deviation of users and QoS average deviation of services;
2) converting the QoS in the full graph or the sub graph into a matrix form;
3) decomposing the matrix according to a PMF algorithm to respectively obtain a user hidden factor vector and a service hidden factor vector;
4) and multiplying the user hidden factor vector by the service hidden factor vector to obtain the predicted QoS, and simultaneously adding the predicted QoS and the three average value deviations in the step 1 to obtain the local and global QoS predicted by the improved PMF algorithm finally.
And 4, step 4: performing self-adaptive fusion processing on the globally and locally predicted QoS to obtain the final predicted QoS;
based on locally and globally predicted QoS, a Gaussian Mixture Model (GMM) is constructed, subject to a Gaussian Model, using the following equations (13) and (14):
Figure RE-RE-GDA0002535305760000132
Figure RE-RE-GDA0002535305760000133
where p (x | t) ═ N (x | u)t，∑t) A function of a gaussian model representing the t-th gaussian distribution, the parameters satisfying:
Figure RE-RE-GDA0002535305760000134
and is
Figure RE-RE-GDA0002535305760000135
Figure RE-RE-GDA0002535305760000136
And
Figure RE-RE-GDA0002535305760000137
respectively representing the proportion of the local predicted QoS and the global predicted QoS in the final predicted QoS;
Figure RE-RE-GDA0002535305760000138
and
Figure RE-RE-GDA0002535305760000139
mean and variance in gaussian models in local and global matrix decomposition are represented, respectively.
In order to fully mine the potential relation between the local information and the global information, the embodiment of the invention uses a method based on a Gaussian mixture model to adaptively fuse the QoS of local prediction and global prediction, and further obtains the QoS of final prediction.
Figure RE-RE-GDA00025353057600001310
Figure RE-RE-GDA00025353057600001311
Figure RE-RE-GDA00025353057600001312
Wherein
Figure RE-RE-GDA0002535305760000141
And
Figure RE-RE-GDA0002535305760000142
respectively represent locally predicted QoS and globally predicted QoS, and1and2local and global predicted QoS respectively represent the respective contributions of the local and global predicted QoS to the final prediction,
Figure RE-RE-GDA0002535305760000143
respectively, mean deviation of QoS in local matrix, user uiMean deviation of invoked local services and service sjAverage deviation of all local user calls;
Figure RE-RE-GDA0002535305760000144
respectively represents the average deviation of QoS in the global matrix, user uiMean deviation of all services invoked and service sjIs calculated from the average deviation of all local user invocations.
Example two
To verify the validity of the algorithm, this example uses the WSD read dataset published by chinese university in hong kong, which collects the actual QoS measurements obtained from 339 users on 5,825 Web services, and contains the following values as shown in table 1 below:
table 1 data set description
Figure RE-RE-GDA0002535305760000145
1. Processing data
1.1 calculating the similarity between users:
1.1.1, collecting longitude and latitude information of a user;
1.1.2 converting the longitude and latitude information of the users into similarity information among the users;
1.1.3 collecting the autonomous system information of the user and calculating the similarity of the autonomous system information;
1.1.4 integrating the longitude and latitude information and the autonomous system information to obtain the similarity information between the end users.
1.2 computing similarity between services
1.2.1 collecting WSDL document information;
1.2.2, carrying out structuralization processing on the WSDL information;
1.2.3 calculating the similarity of WSDL;
1.2.4 collecting the autonomous system information of the service and calculating the similarity of the autonomous system information;
1.2.5 integrating the WSDL information and the autonomous system information to obtain the final similarity between services;
1.3 calculating the similarity between the user and the service;
1.3.1 carrying out normalization operation on QoS;
1.3.2 converting the normalized QoS information into similarity between the user and the service;
2. constructing a full graph
And (3) constructing a full graph model by using the side information among the users, the services, the users and the services and the node information of the two types, which are calculated in the step (1).
3. Obtaining subgraph by cutting whole graph
As shown in fig. 4, the whole graph is segmented by using a spectral clustering idea to obtain sub-graphs;
4. respectively decomposing probability matrix on the whole graph and the subgraph
4.1 improving the probability matrix decomposition algorithm;
4.2 as shown in fig. 5, using 4.1 optimized probability matrix decomposition algorithm to make corresponding QoS prediction on the sub-graph and the whole graph;
5. self-adaptively fusing the sub-graph and the full graph to obtain QoS;
5.1, constructing a Gaussian mixture model according to the QoS result predicted by the sub-graph and the full graph;
5.2 determining the final predicted global and local weight proportion in the QoS according to the Gaussian mixture model;
6. performing QoS prediction;
QoS prediction is accomplished by steps 1-6 above.
Evaluation indexes are as follows: the evaluation indexes selected in the embodiment of the invention are MAE (Mean Square Error) and RMSE (Rootmean Square Error) which are used as the evaluation indexes of the final prediction result.
MAE is defined as follows:
Figure RE-RE-GDA0002535305760000151
the RMSE is defined as follows:
Figure RE-RE-GDA0002535305760000152
wherein gamma isi，jAnd gammai，jRespectively, the true and predicted values, and N represents the total number of predicted QoS. The lower both indices are values indicating more accurate results of the prediction.
To simulate real-world user invocation of services, the entire data set is divided into two parts. One part is used for training the model, and the other part is used for testing the prediction result. Specifically, we randomly deleted 95%, 90%, 85%, and 80% of the data in the call matrix as the test set, and the remaining 5%, 10%, 15%, and 20% of the data as the training set. The experimental results are shown in fig. 6 and 7. The comparison results for the parameters under different conditions are shown in fig. 8.
It can be seen from fig. 6 that as the training set density increases from 5% to 20%, both the MAE and RMSE for all methods become smaller, since a denser matrix will provide more information for the missing QoS prediction. However, the GMF always gets the smallest MAE and RMSE regardless of the training set density variation, which means that GMF can achieve higher prediction accuracy than all comparison methods (including memory-based and model-based methods). Furthermore, the GMF mean MAE and RMSE were improved by 10.13% and 9.70%, respectively, compared to the PMF model (which is the basis of our approach and is superior to most of the approaches in table 2). Even when compared to the HMF method, which first proposed a combination of local and global matrix decomposition and which was widely used in many studies, GMF was still 3.842% and 6.617% improvement in mean MAE and RMSE, respectively.
As can be seen from FIG. 7, the performance of GMF is superior to that of GMF-GMM and GMF-Graph, confirming that: (a) multi-source information is integrated into a Graph model, so that the QoS prediction precision is obviously improved; (b) and the local matrix decomposition and the global matrix decomposition of the GMM model are fused, so that the QoS prediction precision is improved. However, the GMF still yields the best performance for different training set densities. This can be attributed to graph-based integration and GMM-based fusion.
In summary, the embodiment of the present invention integrates multi-source information by using a graph structure based on the existing matrix decomposition QoS prediction method, and adaptively fuses local and global features to improve the prediction accuracy of the service. Compared with the prior art, the invention fully considers the multi-source information: the concept of a full graph is proposed to integrate various multi-source information. The full graph model not only reflects the relationship between users, but also extracts the implicit relationship between services, and the implicit relationship reflects the original relationship between users and services.
The embodiment of the invention solves the problem of data sparsity: with the rapid development of cloud computing, the number of cloud services increases rapidly, so that a QoS matrix for a user to call services is quite sparse; by the method, the missing QoS value can be accurately predicted, the sparse matrix is filled, the density degree of the matrix is improved, and the problem of sparse QoS in the field of cloud service recommendation is solved to a certain extent.
The embodiment of the invention relieves the cold start problem: for users without any call records and services that are not called by any user, when the users are firstly added into the cloud service platform, the suitable services are difficult to recommend to the new users, and meanwhile, the new services are difficult to recommend to the users. By integrating multi-source information, such as the geographic position information of the user, the network position of the user and the like, WSDL description information of the service and the like, the user and the service can be helped to find out their neighbors; and through the graph cutting mode, the users and the services which are potentially closely related are related in the sub-graph, the users and the services exist in the sub-graph, the services in the sub-graph are directly recommended to the users, and the cold start problem is relieved to a certain extent.
Those of ordinary skill in the art will understand that: the figures are merely schematic representations of one embodiment, and the blocks or flow diagrams in the figures are not necessarily required to practice the present invention.
From the above description of the embodiments, it is clear to those skilled in the art that the present invention can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which may be stored in a storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the embodiments or some parts of the embodiments.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for apparatus or system embodiments, since they are substantially similar to method embodiments, they are described in relative terms, as long as they are described in partial descriptions of method embodiments. The above-described embodiments of the apparatus and system are merely illustrative, and the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims (6)

1. A graph model-based QoS prediction method applied to cloud service recommendation is characterized by comprising the following steps:
constructing a full graph model containing multi-source information, wherein the full graph model comprises nodes representing users and services, and edges using the similarity among the users, the similarity among the services and the similarity among the users and the services as weights;
dividing the full graph model into a plurality of sub graph models;
respectively carrying out optimized probability matrix decomposition algorithm on the full graph model and the sub-graph model to obtain globally and locally predicted QoS;
and carrying out self-adaptive fusion processing on the globally and locally predicted QoS to obtain the final predicted QoS.
2. The method of claim 1, wherein constructing a full graph model containing multi-source information, the full graph model including nodes representing users and services, and edges using similarity between users, similarity between services, and similarity between users and services as weights comprises:
calculating the similarity W between the users on the geographical position according to the longitude and latitude information of the userslo(ui，uj)；
Calculating the similarity W between the user and the user in terms of network position by using the autonomous system information of the userAS(ui，uj)；
Obtaining user u by comprehensively considering longitude and latitude information of user and autonomous system informationiAnd user ujFinal similarity between them
Figure RE-FDA0002535305750000011
As shown in equation (1):
Figure RE-FDA0002535305750000012
λ1Wlo(ui，uj)+λ2WAS(ui，uj)＞η1#(1)
η therein1Is a given threshold value, when the final similarity value is greater than η1Then it will be at user uiAnd user ujConstruct a weight of
Figure RE-FDA0002535305750000013
Edge of (1), other cases user uiAnd user ujThere is no edge connection between them; lambda [ alpha ]1And λ2Respectively representing the weight of the geographical position information and the autonomous system information of the user in the final similarity;
calculating semantic similarity W between services according to WSDL information of the servicesws(si，sj)；
Computing similarity W between services and services in terms of network location using autonomous system information of the serviceAS(si，sj)；
Obtaining service s by comprehensively considering WSDL information and autonomous system information of serviceiAnd service siFinal similarity between them
Figure RE-FDA0002535305750000014
As shown in equation (2):
Figure RE-FDA0002535305750000015
γ1Wws(si，sj)+γ2WAs(si，sj)＞η2#(2)
η therein2Is a given threshold value, when the final similarity value is greater than η2Is served by the time siAnd service sjConstruct a weight of
Figure RE-FDA0002535305750000021
Edge of (1), other case service siAnd service sjThere is no edge connection between them; gamma ray1And gamma2Respectively representing the weight of the WSDL information and the autonomous system information of the service in the final similarity;
the similarity between the user and the service is obtained by using the QoS prediction matrix through the following transformation:
Eui，sj＝rti，j
rtmaxj＝max(rtij|i＝1，2，...，m)
rtminj＝min(rtij|i＝1，2，...，m)
rtmax＝(rtmax1，rtmax2，...，rtmaxn)
rtmin＝(rtmin1，rtmin2，...，rtminn)
Figure RE-FDA0002535305750000022
the service call is first normalized and then the normalized RT is converted to a similarity using equation (3), where rtmaxjAnd rtminjRespectively representing services sjAnd rtmax and rtmin represent the maximum and minimum values of all services, respectively, Im×1A one-dimensional column vector representing m users;
respectively taking the similarity between users, the similarity between services and the similarity between users and services as the weight of three edges, taking the users and the services as nodes, and constructing a full graph model containing the users, the services and the three edges: g ═ V, E }, where V ═ U, S }, U ═ U }, and1，u2，...，umis and S ═ S1，s2，...，snRespectively representing that m users and n services are contained; the set of edges is represented as: e ═ Euu，Ess，EusIn which Euu，Ess，EusThe edges using the inter-user, inter-service, and user-service similarities as weights are respectively indicated.
3. The method of claim 2, wherein constructing a full graph model containing multi-source information, the full graph model representing nodes of users and services, and edges using similarity between users, similarity between services, and similarity between users and services as weights comprises:
according to the longitude and latitude information of the userCalculating the similarity W between users in geographic positionlo(ui，uj)：
Figure RE-FDA0002535305750000023
Figure RE-FDA0002535305750000024
dis(ui，uj) Representing user uiAnd user ujThe Euclidean distance between the users is converted into the similarity W of the users on the geographical positions by using the formula (5)lo(ui，uj)，xiAnd yiRespectively represent users uiLongitude and latitude information of the geographical location, xjAnd yjRespectively represent users ujLongitude and latitude information of the geographic location;
calculating user u by formula (6) using the autonomous system information of the useriAnd user ujSimilarity W in network location betweenAS(ui，uj)：
Figure RE-FDA0002535305750000031
The method comprises the steps of performing text removal processing on WSDL information, removing structured format information, extracting feature words in the WSDL information, and converting special keywords in the WSDL information into semantic similarity between services by using a tf-idf algorithm in the field of natural language processing through the following two formulas:
Figure RE-FDA0002535305750000032
Figure RE-FDA0002535305750000033
in the formula (7), M is UtilityTotal number of web pages searched from Google for the token words x and y, logf (x) and logf (y) are the number of clicks searched using the token words x and y, respectively; f (x, y) represents the number of web pages using both the feature words x and y, in equation (8)
Figure RE-FDA0002535305750000034
And
Figure RE-FDA0002535305750000035
respectively representing services si，sjThe feature word vectors of (a) are,
Figure RE-FDA0002535305750000036
representing the cardinality of the vector. Wws(si，sj) Representation service siAnd service sjSemantic similarity in terms of WSDL information;
calculating service s by formula (9) using autonomous system information where service is locatediAnd service sjSimilarity in network location WAS(si，sj)：
Figure RE-FDA0002535305750000037
4. The method of claim 1, wherein the dividing the full graph model into a plurality of sub-graph models comprises:
converting the full graph model G into a similarity matrix containing the similarity between the user and the service;
constructing an adjacent matrix according to the similarity matrix, and constructing a degree matrix at the same time;
constructing a Laplace matrix according to the adjacent matrix and the degree matrix;
normalizing the laplacian matrix;
calculating minimum K eigenvalues and eigenvectors of the Laplace matrix;
combining K eigenvectors into an eigenvector matrix, and standardizing the eigenvector matrix according to rows;
clustering the normalized feature matrix to obtain a set of the segmented K sub-graphs
Figure RE-FDA0002535305750000041
Where K denotes the number of subgraphs, Sub1Denotes the 1 st Sub-figure, SubKRepresenting the K-th sub-graph.
5. The method according to any one of claims 1 to 4, wherein the probability matrix decomposition algorithm for optimizing the full graph model and the sub graph model respectively obtains the globally and locally predicted QoS, and comprises:
the optimized probability matrix decomposition algorithm is set as follows:
Figure RE-FDA0002535305750000042
r represents the final predicted QoS value,
Figure RE-FDA0002535305750000043
represents RijObedience mean 0 and variance σRNormal distribution of (2);
μBmean deviation of all QoS, BU mean deviation of QoS generated by a user invoking a service, BS mean deviation of QoS generated by a service invoked by a user, and three mean deviations are defined as follows:
Figure RE-FDA0002535305750000044
Figure RE-FDA0002535305750000045
Figure RE-FDA0002535305750000046
wherein R represents allNumber of Qos, R(i)Representing user uiSet of invoked services, R(j)Indicating that a service s has been invokedjA set of users of (1);
converting QoS in the full graph model into a matrix form, decomposing the matrix according to the optimized probability matrix decomposition algorithm to respectively obtain a user hidden factor vector and a service hidden factor vector, multiplying the user hidden factor vector and the service hidden factor vector to obtain predicted QoS, and multiplying the predicted QoS and the muBBU and BS are added to obtain globally predicted QoS;
converting QoS in the sub-graph model into a matrix form, decomposing the matrix according to the optimized probability matrix decomposition algorithm to respectively obtain a user hidden factor vector and a service hidden factor vector, multiplying the user hidden factor vector and the service hidden factor vector to obtain predicted QoS, and multiplying the predicted QoS and muBThe BU and the BS are added to obtain the locally predicted QoS.
6. The method of claim 5, wherein adaptively fusing the globally and locally predicted QoS to obtain a final predicted QoS, comprises:
based on locally and globally predicted QoS, a gaussian mixture model is constructed, obeying the gaussian model, using the following equations (12) and (13):
Figure RE-FDA0002535305750000051
Figure RE-FDA0002535305750000052
where p (x | t) ═ N (x | u)t，∑t) A function of a gaussian model representing the t-th gaussian distribution, the parameters satisfying:
Figure RE-FDA0002535305750000053
and is
Figure RE-FDA0002535305750000054
Figure RE-FDA0002535305750000055
And
Figure RE-FDA0002535305750000056
respectively representing the proportion of the local predicted QoS and the global predicted QoS in the final predicted QoS;
Figure RE-FDA0002535305750000057
and
Figure RE-FDA0002535305750000058
respectively representing the mean and variance in Gaussian models in local and global matrix decomposition;
and adaptively fusing local and global predicted QoS based on the Gaussian mixture model to obtain final predicted QoS:
Figure RE-FDA0002535305750000059
Figure RE-FDA00025353057500000510
Figure RE-FDA00025353057500000511
wherein
Figure RE-FDA00025353057500000512
And
Figure RE-FDA00025353057500000513
respectively represent locally predicted QoS and globally predicted QoS, and1and2local and global predicted QoS respectively represent the respective proportions of the local and global predicted QoS in the final predicted QoS,
Figure RE-FDA00025353057500000514
SubBU，SubBSrespectively represents the average deviation of QoS, user uiMean deviation of invoked local services and service sjAverage deviation of all local user calls; gμB，GBU，GBSRespectively expressed in the global matrix, average deviation of QoS, user uiMean deviation of all services invoked and service sjIs calculated from the average deviation of all local user invocations.
CN202010322193.8A 2020-04-22 2020-04-22 QoS prediction method applied to cloud service recommendation and based on graph model Active CN111553401B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010322193.8A CN111553401B (en) 2020-04-22 2020-04-22 QoS prediction method applied to cloud service recommendation and based on graph model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010322193.8A CN111553401B (en) 2020-04-22 2020-04-22 QoS prediction method applied to cloud service recommendation and based on graph model

Publications (2)

Publication Number Publication Date
CN111553401A true CN111553401A (en) 2020-08-18
CN111553401B CN111553401B (en) 2024-02-13

Family

ID=72002480

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010322193.8A Active CN111553401B (en) 2020-04-22 2020-04-22 QoS prediction method applied to cloud service recommendation and based on graph model

Country Status (1)

Country Link
CN (1) CN111553401B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113420421A (en) * 2021-05-28 2021-09-21 西安邮电大学 QoS prediction method based on time sequence regularization tensor decomposition in moving edge calculation
CN116132347A (en) * 2023-04-06 2023-05-16 湖南工商大学 Bi-LSTM-based service QoS prediction method in computing network convergence environment

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107135092A (en) * 2017-03-15 2017-09-05 浙江工业大学 A kind of Web service clustering method towards global social interaction server net
CN107291894A (en) * 2017-06-21 2017-10-24 武汉理工大学 A kind of probability matrix decomposition model for merging similitude and common scoring item quantity
US20180165554A1 (en) * 2016-12-09 2018-06-14 The Research Foundation For The State University Of New York Semisupervised autoencoder for sentiment analysis

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180165554A1 (en) * 2016-12-09 2018-06-14 The Research Foundation For The State University Of New York Semisupervised autoencoder for sentiment analysis
CN107135092A (en) * 2017-03-15 2017-09-05 浙江工业大学 A kind of Web service clustering method towards global social interaction server net
CN107291894A (en) * 2017-06-21 2017-10-24 武汉理工大学 A kind of probability matrix decomposition model for merging similitude and common scoring item quantity

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
HUA MA等: "Multi-valued collaborative QoS prediction for cloud service via time series analysis", 《FUTURE GENERATION COMPUTER SYSTEMS》 *
刘欣: "面向社会化媒体的内容推荐若干关键技术研究", 《中国博士学位论文全文数据库（电子期刊）信息科技辑》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113420421A (en) * 2021-05-28 2021-09-21 西安邮电大学 QoS prediction method based on time sequence regularization tensor decomposition in moving edge calculation
CN113420421B (en) * 2021-05-28 2023-07-25 西安邮电大学 QoS prediction method based on time sequence regularized tensor decomposition in mobile edge calculation
CN116132347A (en) * 2023-04-06 2023-05-16 湖南工商大学 Bi-LSTM-based service QoS prediction method in computing network convergence environment

Also Published As

Publication number Publication date
CN111553401B (en) 2024-02-13

Similar Documents

Publication Publication Date Title
CN109783582B (en) Knowledge base alignment method, device, computer equipment and storage medium
CN109948710B (en) Micro-service identification method based on API similarity
Kosowska-Stamirowska et al. Evolving structure of the maritime trade network: evidence from the Lloyd’s Shipping Index (1890–2000)
Chang et al. A graph-based QoS prediction approach for web service recommendation
CN107545038B (en) Text classification method and equipment
CN114998602B (en) Domain adaptive learning method and system based on low confidence sample contrast loss
CN111553401B (en) QoS prediction method applied to cloud service recommendation and based on graph model
CN112231592A (en) Network community discovery method, device, equipment and storage medium based on graph
Tang et al. A factorization machine-based QoS prediction approach for mobile service selection
CN111062431A (en) Image clustering method, image clustering device, electronic device, and storage medium
CN111198897A (en) Scientific research hotspot topic analysis method and device and electronic equipment
CN113591892A (en) Training data processing method and device
WO2023051085A1 (en) Object recognition method and apparatus, device, storage medium and program product
WO2023011062A1 (en) Information pushing method and apparatus, device, storage medium, and computer program product
Chen et al. A user dependent web service QoS collaborative prediction approach using neighborhood regularized matrix factorization
CN116226747A (en) Training method of data classification model, data classification method and electronic equipment
CN111209105A (en) Capacity expansion processing method, capacity expansion processing device, capacity expansion processing equipment and readable storage medium
CN112463964B (en) Text classification and model training method, device, equipment and storage medium
CN114861004A (en) Social event detection method, device and system
CN116861226A (en) Data processing method and related device
Miao et al. Informative core identification in complex networks
CN111046934B (en) SWIFT message soft clause recognition method and device
CN111930883A (en) Text clustering method and device, electronic equipment and computer storage medium
CN117807237B (en) Paper classification method, device, equipment and medium based on multivariate data fusion
CN116932832B (en) Data asset catalog generation method, device and computer readable storage medium

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