CN104579850A - Quality of service (QoS) prediction method for Web service under mobile Internet environment - Google Patents
Quality of service (QoS) prediction method for Web service under mobile Internet environment Download PDFInfo
- Publication number
- CN104579850A CN104579850A CN201510039675.1A CN201510039675A CN104579850A CN 104579850 A CN104579850 A CN 104579850A CN 201510039675 A CN201510039675 A CN 201510039675A CN 104579850 A CN104579850 A CN 104579850A
- Authority
- CN
- China
- Prior art keywords
- qos
- user
- web service
- data
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Provided is a quality of service (QoS) prediction method for a Web service under a mobile Internet environment. Firstly, QoS original data of a certain Web service repeatedly accessed by a certain user under the mobile Internet environment, and carrying out preprocessing on the QoS original data by using logarithm processing mode is collected; secondly, similar user groups of a target user is found through an improved and user-based collaborative filtering method, and then a normal value section is confirmed according to features of QoS attributes; thirdly, an authentic QoS prediction value can be obtained through the calculation. The quality of service (QoS) prediction method is characterized in that the preprocessing on the original data is carried out through logarithm calculation; the normal value section is selected to avoid an abnormal value generated by QoS volatility; a user personality characteristic shown by the volatility is comprehensively considered; the reliability of QoS data is guaranteed when calculating the similarity; the accurate QoS prediction value is obtained by the calculation through similar users. The quality of service (QoS) prediction method has the advantages that the calculation accurateness is high, the practicality is high, and the value for the application and popularization is good.
Description
Technical field
The present invention relates to service quality QoS (Quality ofService) Forecasting Methodology of Web service under a kind of mobile internet environment; Belong to the technical field of computer application.
Background technology
Due to developing rapidly of mobile network, various Web service of magnanimity just starts constantly in succession to occur successively, and therefore, when selecting Web service, user needs in the face of the identical but candidate's Web service that performance difference is different of increasing function.And it is on probation that user can not carry out one by one to each Web service, to select best Web service.This just needs the performance of Web service on each service quality QoS attribute considering various candidate.The QoS specific performance parameter of Web service comprises: one group of QoS attribute of availability, response time, bandwidth sum throughput etc., each QoS attribute is wherein for characterizing Web service quality information in one aspect.But the history of any user is accessed the various QoS data produced and might not be contained all candidate's Web services, so predict in the face of the QoS data lacked just needs to adopt a kind of method to make QoS, and select corresponding Web service according to predicting the outcome.Visible, predict that QoS performance parameter is carrying out prerequisite and the key of follow-up Web service recommendation exactly, so be the emphasis of scholar's research of being correlated with for the research of this problem always both at home and abroad.
External prior art situation is: Balke and Wagner of Hanoverian, Germany university proposes Web service QoS query scheme user's using forestland, demand and user preference combined.Balke also further provides the collaborative discovery mechanism of the Web service QoS based on user preference.
The people such as the Maamar of Zha Yede university of the United Arab Emirates propose about the mutual model of Web Service, and except user preference, the feature of this work highlights the mutual resource of Web Service.The focusing on of the method provides a mechanism, changes the resource of the preference of service consumer, history, sight and service provider in the form.
The people such as the Kokash of Univ Leiden Netherlands propose a kind of Web service discover method based on user's previous experiences.Interactive form between Web service and consumer is dissolved as implicit expression culture by this method, recommends to provide good Web service with this Web service consumer to new registration.
Domestic situation is: the Web service Forecasting Methodology based on Euclidean distance that the people such as the Shao Lingshuan of Peking University propose, and every QoS data is converted into distance vector, and then calculates the similarity degree between user, thus realizes the prediction of disappearance QoS data.
The WSRec method that the people such as the Zheng Zibin of Hong Kong Chinese University propose, the optimization carrying out the degree of depth on the basis of collaborative filtering, by to the weights of every QoS data and the analysis of parameter, and the calculating to user's history QoS data, eliminate the error of the Similarity Measure that user's individual factors causes, thus realize the accurate location of similar users, and finally realize QoS prediction accurately.
In sum, mostly there is following shortcoming in current existing QoS Forecasting Methodology:
Solving of the data processing of the various QoS Forecasting Methodologies of currently available technology and user's similarity is all that this network environment basis that substantially can realize stable transfer based on such as conventional internet is carried out, and the overwhelming majority is for certain QoS initial data once gathered independent of the QoS data analyzed also just Stochastic choice.Compare traditional the Internet, mobile Internet just just enters in recent years and widely uses, its time occurred is shorter, and in transmission, there is its inborn feature and defect: during as carried out information transmission by wireless signal, can postpone continually, the situation such as packet loss, the frequent generation of these transmission situations, causes the uncertainty of QoS data to change, and as large in response time change, throughput diminishes etc.Even in the good situation of network state, by various objective factors (load etc. as the position of user, surrounding environment, the radio transmission apparatus) impact in wireless transmission process, this situation also can often occur.Thus, cause the numerical value of the qos parameter of numerous Web service to manifest the change of fluctuation, namely there is exceptional value and normal value.Therefore, if the Forecasting Methodology of these prior aries above-mentioned be applied directly in the QoS prediction of the Web service under mobile internet environment, just there will be larger predicated error.
The reason of QoS predicated error is caused to mainly contain following 2 points:
(1) due to the impact of fluctuation, the QoS data of the data centralization used during prediction differs and reflects this node QoS situation under normal circumstances surely truly, and namely QoS data is unreliable.Therefore this insecure QoS data, may cause the error of Similarity Measure, and what cause mistake is used as non-similar users as similar users process.
(2) QoS data that same because fluctuation causes is unreliable, even if having selected correct similar users, in the determination of QoS data predicted value, also can be subject to the impact of fluctuation.
Therefore, how to improve the above-mentioned two problems causing the QoS numerical prediction error of Web service, just become the new problem that scientific and technical personnel in the industry pay close attention to, and a large amount of Exploration & stu dy has been carried out to it.
Summary of the invention
In view of this, the object of this invention is to provide a kind of service quality QoS (Quality of Service) Forecasting Methodology of Web service under mobile internet environment, the inventive method can under the mobile internet environment that the numerical fluctuations of qos parameter is stronger Accurate Prediction QoS numerical value, the method first utilizes simple mode to carry out preliminary treatment to the QoS initial data gathered, to reduce the fluctuation of the QoS data obtained under mobile internet environment well, and then by calculating the similar users group finding out targeted customer, determine that its QoS normal value is interval, to evade and to delete the exceptional value in QoS data, thus obtain QoS predicted value comparatively accurately, as the key reference factor of choose reasonable Web service, for applying of Web service contributes.
In order to reach foregoing invention object, the invention provides a kind of Forecasting Methodology of service quality QoS (Quality of Service) of Web service under mobile internet environment, it is characterized in that: first under collection mobile internet environment, certain user repeatedly accesses the QoS initial data of certain Web service, utilizes the mode of logarithm process to carry out preliminary treatment to this QoS initial data; Then the collaborative filtering method based on user by improving, finds out the similar users group of targeted customer, and then interval according to the feature determination normal value of QoS attribute, finally calculates a believable QoS predicted value; Described method comprises following operative step:
Step 1, preliminary treatment is performed from the QoS initial data in real world: because the data of mobile Internet have multiple uncertain noises factor in wireless transmission process to gathering, the fluctuation causing QoS initial data during access Web service to present in various degree changes, and especially the numerical fluctuations of response time is more obvious; Therefore first obtain the repeatedly QoS initial data of every user to each Web service that it was accessed, then preliminary treatment is carried out to these QoS initial data: the QoS data difference utilizing Logarithmic calculation to reduce fluctuation to cause, obtains the QoS data that subsequent calculations is used;
Step 2, calculate the similarity between user: the Pearson came Pearson correlation coefficient PCC (Pearson Correlation Coefficient) repeatedly between QoS data adopting the collaborative filtering method based on user to calculate two users to obtain in its each Web service of jointly accessing, again the Pearson correlation coefficient that these two users calculate in its all common Web service of accessing is converted to a similarity numerical value, for representing the user's similarity between these two users; Then, aforesaid way is adopted to try to achieve all users similarity each other;
Step 3, determine that the normal value of similar users group and QoS data thereof is interval, evade exceptional value: first according to the result of calculation determination similar users group of step 2, the constant interval of QoS data normal value is selected again according to QoS attribute feature, for shielding exceptional value, thus determine that the normal value of the QoS data of each similar users in similar users group is interval;
Step 4, solves QoS predicted value: be weighted process to the multiple similar users determined and QoS data thereof, and by after normal value interval shielding exceptional value, obtains final QoS predicted value.
The crucial innovative technology of the QoS Forecasting Methodology of Web service under mobile internet environment of the present invention is: have employed two-step and process exceptional value: the first step, with Logarithmic calculation, preliminary treatment is carried out to QoS initial data, then similarity between user is calculated with the PCC after improvement, so both reduce exceptional value to impact during Similarity Measure, consider again the user personality feature that fluctuation embodies.Second step, carrying out the mode of data screening, deleting the exceptional value that QoS fluctuation produces by choosing normal value interval, thus thoroughly eliminates exceptional value to the impact of predictor calculation, realizes QoS prediction accurately.
In addition, the innovative technology of the inventive method also comprises following 3 points:
(A) by the mode of simple Logarithmic calculation, preliminary treatment is carried out to the QoS initial data gathered, realize effective, and method is simple, be applicable to the requirement that large data calculate.
(B) by calculating its PCC coefficient correlation to the QoS data of two same Web services of user's repeated accesses, and the coefficient correlation that the Web service of all common access of these two users obtains is weighted process, thus draw to the similarity of two users.The individualized feature that fluctuation QoS data when PCC computational methods after this improvement consider the Web service of user's repeated accesses embodies, improves the accuracy of user's Similarity Measure.
(C) according to the own characteristic of QoS data attribute, determine that normal value is interval, evade the impact of exceptional value further, this mode is simply effective, can farthest delete all exceptional values.
The advantage of the inventive method is: can carry out preliminary treatment by simple mode to the fluctuation QoS initial data in mobile Internet, obtain reliable QoS data and carry out Similarity Measure, thus obtain reliable similar users group.Then interval according to the own characteristic determination normal value of QoS attribute, evade further and delete the impact of QoS data fluctuation, to obtain reliable QoS predicted value.
The applicable performance of the inventive method is strong, can be used in the QoS numerical prediction of all kinds Web service.The most important thing is, the processing mode of the inventive method is simple to operate, easily realize, computation complexity is low, is applicable to very much large data and calculates, effectively can ensure efficiency when large-scale consumer data calculate.
Therefore, the present invention has higher calculating accuracy and stronger practicality, has good application value.
Accompanying drawing explanation
Fig. 1 is the operating procedure flow chart of the QoS Forecasting Methodology of Web service under mobile internet environment of the present invention.
Embodiment
For making the object, technical solutions and advantages of the present invention clearly, below in conjunction with accompanying drawing, the present invention is described in further detail.
The QoS Forecasting Methodology of Web service under mobile internet environment of the present invention, is first to gather under mobile internet environment the QoS initial data that certain user repeatedly accesses certain Web service, carries out preliminary treatment by the mode of logarithm process to this QoS initial data; Then the method for the collaborative filtering based on user by improving, finds out the most similar users group of targeted customer, and then interval according to the attribute feature determination normal value of QoS data, finally calculates a believable QoS predicted value.
See Fig. 1, introduce the concrete operation step of the inventive method:
Step 1, performs preliminary treatment to gathering from the QoS initial data in real world: due in the network data wireless transmission process in mobile Internet, have multiple uncertain noises factor; Especially when network busy, packet loss, time-out constantly occur and retransmits, the fluctuation causing QoS initial data during access Web service to present in various degree changes, and especially the numerical fluctuations of response time is more obvious.Such as, normal value and exceptional value will be there is in the QoS initial data set of collection.Although the number of exceptional value is fewer, exceptional value compares the numerical value of normal value can much larger (as the response time).Carry out simple calculations iff on QoS initial data basis, the deviation of acquired results must be very large.Therefore, first every user is obtained to after the repeatedly QoS initial data of each Web service that it was accessed, again preliminary treatment is carried out to these QoS initial data: the QoS data difference utilizing Logarithmic calculation to reduce fluctuation to cause, obtains the QoS data that subsequent calculations is used.This step 1 comprises following content of operation:
(11) the QoS raw data set of i-th user to common t the repeated accesses of a jth Web service is set and is combined into Q
original i,j={ q
1 i,j, q
2 i,j..., q
r i,j..., q
t i,j, the Web service sequence number that wherein, natural number i, j and r are user's sequence number respectively, this user accessed and the sequence number of repeated accesses number of times thereof, the maximum of r is t; q
r i,jbe that i-th user performs the QoS initial data of the r time access to a jth Web service.
(12) the QoS initial data set Q of i-th user to common t the repeated accesses of a jth Web service is set
original i,jin each QoS initial data to carry out pretreated QoS data set be Q
i,j={ p
1 i,j, p
2 i,j..., p
r i,j..., p
t i,j; Wherein, p
r i,jbe the QoS initial data q of i-th user to a jth Web service the r time access
r i,jcarry out pretreated QoS data.
(13) to the QoS initial data q of i-th user to common t the repeated accesses of a jth Web service
r i,jaccording to natural logrithm computing formula p
r i,j=ln (q
r i,j) carry out the preliminary treatment of the logarithm operation taking natural logrithm the end of as.
Step 2, calculate the similarity between user: the collaborative filtering method based on user that employing improves calculates the Pearson came Pearson correlation coefficient PCC repeatedly between QoS data that two users obtain in its each Web service of jointly accessing, again the Pearson correlation coefficient that these two users calculate in its all common Web service of accessing is converted to a similarity numerical value, for representing the user's similarity between these two users; Then, aforesaid way is adopted to try to achieve all users similarity each other.
The collaborative filtering method based on user of the improvement that the present invention uses is the historical behavior data according to user, find that the similar users of destination object is individual based on the neighborhood between user, thus the prediction realized the behavioral data of destination object, reach the object that individual character is recommended.
In step 2, because calculating simple and precision is high, the similarity between user adopts the method calculating Pearson correlation coefficient PCC; But because of the feature of QoS data and data set thereof, need the account form improving PCC: the personal characteristics both having considered the user of QoS data when user accesses Web service, the personal characteristics of QoS data fluctuation when taking the Web service of user's repeated accesses again into account.
Between the calculating user in this step 2, the operation of similarity comprises following content:
(21), when calculating the similarity of two user a and b, the set US of the Web service that these two users accessed jointly is first found out
a∩ US
b={ s
1, s
2, s
3..., s
j..., s
n, in formula, s
jfor the jth Web service that these two user a and b accessed jointly, the sequence number that natural number j and n is respectively the Web service that these two user a and b accessed jointly and the sum of Web service of jointly accessing, namely the maximum of j is n.
(22), when calculating the degree of correlation of these two user a and b in each Web service that it was accessed jointly, the Pearson correlation coefficient of these two users in each common Web service of accessing is obtained; Then arithmetic average mode is adopted to be converted to a similarity numerical value, the user's similarity as the non-weighting process of these two users these obtained coefficient correlations:
in formula, sim
a,bfor user's similarity numerical value of the non-weighting process of these two user a and b,
with
be respectively the desired value of the QoS data obtained when user a and user b repeats t access jth Web service separately,
With
Be respectively the variance of the QoS data obtained when user a and user b repeats t access jth Web service separately,
all t QoS data p
1 a,j+ p
2 a,j+ p
3 a,j+ ... + p
t a,jcumulative sum.
(23) PCC of QoS data because obtaining when the computational process of above-mentioned steps (22) only considered the Web service that these two user a and b access jointly, and solve its mean value, do not have to get rid of when the Web service quantity n jointly accessed is less, on the impact that result of calculation causes; Therefore need the result of calculation sim to step (22)
a,bby following formula
be weighted process; In formula, sim '
a,bfor to result of calculation sim
a,bbe weighted the numerical value after process, for representing user's similarity of final these two user a and b; | US
a∩ US
b| be the number of the Web service that these two users accessed jointly, | US
a| with | US
b| be respectively the number of the Web service that user a and user b accessed separately respectively.
(24) return to execution step (21) ~ (23), solve the similarity obtained between all users.
Step 3, determine that the normal value of similar users group and QoS data thereof is interval, evade exceptional value: first according to the result of calculation determination similar users group of step 2, the constant interval of QoS data normal value is selected again according to QoS attribute feature, for shielding exceptional value, thus determine that the normal value of the QoS data of each similar users in similar users group is interval.This is because the QoS attribute of Web service has two types: forward attribute and negative sense attribute, wherein when network environment is deteriorated and is unfavorable for information transmission, the QoS numerical value (throughput as system) of forward attribute can diminish, and the QoS numerical value (as response time and packet loss etc.) of negative sense attribute can become large; Therefore the constant interval of QoS data normal value to be selected according to QoS attribute feature, for shielding exceptional value, to determine that the normal value of the QoS data of each similar users in similar users group is interval.
This step 3 comprises following content of operation:
(31) time to predict the disappearance QoS data of certain user a to certain Web service w, first obtain the similarity data between user a and all users according to step 2, then all users having history to access to Web service w are carried out descending according to the size of itself and user a similarity numerical value.
(32) front K user in the descending user obtained according to step (31) forms the similar users group of user a: S
sam (a)={ u
1, u
2, u
3..., u
c..., u
k, in formula, natural number c is the similar users sequence number in user a similar users group, and its maximum is K, u
csimilar users group S
sam (a)in c similar users; The size of number of users K depends on the data set of calculating time institute foundation.Wherein, the size of the quantity K of user a similar users group is carry out test of many times on used QoS data collection after, K value when prediction error value is minimum.
(33) the QoS data set Q of K similar users in user a similar users group is set
uc, jset be
Wherein,
for c similar users in the similar users group of user a repeats the QoS data set of a t access jth Web service.
(34) each similar users u in similar users group is solved
cthe QoS data set obtained during the Web service of repeated accesses jth
desired value
wherein,
for
in r QoS data element; Natural number r is the access times sequence number of certain Web service of user's repeated accesses, and its maximum is t.
(35) because QoS data attribute comprises forward attribute or negative sense attribute, this characteristic also can find from the QoS initial data gathered: when network condition is deteriorated and affects information transmission, response time numerical value trends towards fluctuation and becomes large, and throughput numerical value trends towards fluctuation and diminishes; Therefore the normal value of forward attribute QoS data interval be (
p
max), the normal value interval of negative sense attribute QoS data is (p
min,
); In formula,
for c similar users u in similar users group
cthe QoS data set obtained during the Web service of repeated accesses jth
desired value, p
minand p
maxbe respectively corresponding QoS data set
in minimum value and maximum.
(36) screen according to described QoS data normal value interval, obtain the normal value set of QoS data:
And
Or (
, p
max); Wherein,
represent
for
subset, namely
in data be all from QoS data set
middle screening, and, if set
in QoS data
for forward attribute, then its numerical value is positioned at interval (p
min,
) in; If set
in QoS data
for negative sense attribute, then its numerical value be positioned at interval (
p
max) in, natural number e is the sequence number of QoS data, and its maximum is v.
Step 4, solves QoS predicted value: be weighted process to the multiple similar users determined and QoS data thereof, and by after normal value interval shielding exceptional value, obtains final QoS predicted value.
This step 4 comprises following content of operation:
(41) successively to set
In each parameter according to after step (36) anomaly sieving Value Data, obtain screen after QoS data set
(42) according to formula
calculate, obtain common K QoS predicted value F (u
c, j); Wherein, u
c∈ Q
sim (a)={ u
1, u
2..., u
c..., u
k,
v is
in the total number of QoS data;
(43) to this K predicted value result F (u
c, after j) being weighted process according to the following equation, obtain disappearance QoS data predicted value Forecast when this user a accesses jth Web service:
Wherein,
for similar users u in user a and its similar users group
cuser between similarity, natural number subscript c is the similar users sequence number in similar users group, and its maximum is K.
Repeatedly Case Experiments On A has been carried out on the basis of the QoS data collection that the present invention has collected in mobile Internet, and the result of test is successful, achieves goal of the invention.
Claims (8)
1. the Forecasting Methodology of the service quality QoS (Quality of Service) of a Web service under mobile internet environment, it is characterized in that: first under collection mobile internet environment, certain user repeatedly accesses the QoS initial data of certain Web service, utilizes the mode of logarithm process to carry out preliminary treatment to this QoS initial data; Then the collaborative filtering method based on user by improving, finds out the similar users group of targeted customer, and then interval according to the feature determination normal value of QoS attribute, finally calculates a believable accurate QoS predicted value; Described method comprises following operative step:
Step 1, preliminary treatment is performed from the QoS initial data in real world: because the data of mobile Internet have multiple uncertain noises factor in wireless transmission process to gathering, the fluctuation causing QoS initial data during access Web service to present in various degree changes, and especially the numerical fluctuations of response time is more obvious; Therefore first obtain the repeatedly QoS initial data of every user to each Web service that it was accessed, then preliminary treatment is carried out to these QoS initial data: the QoS data difference utilizing Logarithmic calculation to reduce fluctuation to cause, obtains the QoS data that subsequent calculations is used;
Step 2, calculate the similarity between user: the collaborative filtering method based on user that employing improves calculates the Pearson came Pearson correlation coefficient PCC (Pearson Correlation Coefficient) repeatedly between QoS data that two users obtain in its each Web service of jointly accessing, again the Pearson correlation coefficient that these two users calculate in its all common Web service of accessing is converted to a similarity numerical value, for representing the user's similarity between these two users; Then, aforesaid way is adopted to try to achieve all users similarity each other;
Step 3, determine that the normal value of similar users group and QoS data thereof is interval, evade exceptional value: first according to the result of calculation determination similar users group of step 2, the constant interval of QoS data normal value is selected again according to QoS attribute feature, for shielding exceptional value, thus determine that the normal value of the QoS data of each similar users in similar users group is interval;
Step 4, solves QoS predicted value: be weighted process to the multiple similar users determined and QoS data thereof, and by after normal value interval shielding exceptional value, obtains final QoS predicted value.
2. method according to claim 1, is characterized in that: described step 1 comprises following content of operation:
(11) arrange the QoS raw data set of i-th user to common t the repeated accesses of a jth Web service to be combined into
wherein, the Web service sequence number that natural number i, j and r are user's sequence number respectively, this user accessed and the sequence number of repeated accesses number of times thereof, the maximum of r is t; q
r i,jbe that i-th user performs the QoS initial data of the r time access to a jth Web service;
(12) the QoS initial data set Q of i-th user to common t the repeated accesses of a jth Web service is set
original i,jin each QoS initial data to carry out pretreated QoS data set be Q
i,j={ p
1 i,j, p
2 i,j..., p
r i,j..., p
t i,j; Wherein, p
r i,jbe the QoS initial data q of i-th user to a jth Web service the r time access
r i,jcarry out pretreated QoS data;
(13) to the QoS initial data q of i-th user to common t the repeated accesses of a jth Web service
r i,jaccording to natural logrithm computing formula p
r i,j=ln (q
r i,j) carry out the preliminary treatment of the logarithm operation taking natural logrithm the end of as.
3. method according to claim 1, it is characterized in that: the collaborative filtering method based on user of described improvement is the historical behavior data according to user, find that the similar users of destination object is individual based on the neighborhood between user, thus the prediction realized the behavioral data of destination object, reach the object that individual character is recommended.
4. method according to claim 3, is characterized in that: in described step 2, and because calculating simple and precision is high, the similarity between user adopts the method calculating Pearson correlation coefficient PCC; But because of the feature of QoS data and data set thereof, need the account form improving PCC: the personal characteristics both having considered the user of QoS data when user accesses Web service, the personal characteristics of QoS data fluctuation when taking the Web service of user's repeated accesses again into account.
5. method according to claim 4, is characterized in that: between the calculating user in described step 2, the operation of similarity comprises following content:
(21), when calculating the similarity of two user a and b, the set US of the Web service that these two users accessed jointly is first found out
a∩ US
b={ s
1, s
2, s
3..., s
j..., s
n, in formula, s
jfor the jth Web service that these two user a and b accessed jointly, the sequence number that natural number j and n is respectively the Web service that these two user a and b accessed jointly and the sum of Web service of jointly accessing, namely the maximum of j is n;
(22), when calculating the degree of correlation of these two user a and b in each Web service that it was accessed jointly, the Pearson correlation coefficient of these two users in each common Web service of accessing is obtained; Then arithmetic average mode is adopted to be converted to a similarity numerical value, the user's similarity as the non-weighting process of these two users these obtained coefficient correlations:
in formula, sim
a,bfor user's similarity numerical value of the non-weighting process of these two user a and b,
with
be respectively the desired value of the QoS data obtained when user a and user b repeats t access jth Web service separately,
With
Be respectively the variance of the QoS data obtained when user a and user b repeats t access jth Web service separately,
all t QoS data p
1 a,j+ p
2 a,j+ p
3 a,j+ ... + p
t a,jcumulative sum;
(23) PCC of QoS data because obtaining when the computational process of above-mentioned steps (22) only considered the Web service that these two user a and b access jointly, and solve its mean value, do not have to get rid of when the Web service quantity n jointly accessed is less, on the impact that result of calculation causes; Therefore the result of calculation sim of reply step (22)
a,bby following formula
be weighted process; In formula, sim '
a,bfor to result of calculation sim
a,bbe weighted the numerical value after process, for representing user's similarity of final these two user a and b; | US
a∩ US
b| be the number of the Web service that these two users accessed jointly, | US
a| with | US
b| be respectively the number of the Web service that user a and user b accessed separately respectively;
(24) return to execution step (21) ~ (23), solve the similarity obtained between all users.
6. method according to claim 4, is characterized in that: described step 3 comprises following content of operation:
(31) time to predict the disappearance QoS data of certain user a to certain Web service w, first obtain the similarity data between user a and all users according to step 2, then all users having history to access to Web service w are carried out descending according to the size of itself and user a similarity numerical value;
(32) front K user in the descending user obtained according to step (31) forms the similar users group of user a: S
sam (a)={ u
1, u
2, u
3..., u
c..., u
k, in formula, natural number c is the similar users sequence number in user a similar users group, and its maximum is K, u
csimilar users group S
sam (a)in c similar users; The size of number of users K depends on the data set of calculating time institute foundation;
(33) the QoS data set of K similar users in user a similar users group is set
set be
wherein,
for c similar users in the similar users group of user a repeats the QoS data set of a t access jth Web service;
(34) each similar users u in similar users group is solved
cthe QoS data set obtained during the Web service of repeated accesses jth
desired value
wherein,
for
in r QoS data element; Natural number r is the access times sequence number of certain Web service of user's repeated accesses, and its maximum is t;
(35) because QoS data attribute comprises forward attribute or negative sense attribute, wherein, the response time is negative sense attribute, and throughput is forward attribute; Therefore can find the QoS initial data gathered: when network condition is deteriorated and affects information transmission, response time numerical value trends towards becoming large, and throughput numerical value trends towards diminishing; Therefore the normal value interval of forward attribute QoS data is
the normal value interval of negative sense attribute QoS data is
in formula,
for c similar users u in similar users group
cthe QoS data set obtained during the Web service of repeated accesses jth
desired value, p
minand p
maxbe respectively corresponding QoS data set
in minimum value and maximum;
(36) screen according to described QoS data normal value interval, obtain the normal value set of QoS data:
And
wherein,
represent
for
subset, namely
in data be all from QoS data set
middle screening, and, if set
in QoS data
for forward attribute, then its numerical value is positioned at interval
in; If set
in QoS data
for negative sense attribute, then its numerical value is positioned at interval
in, natural number e is the sequence number of QoS data, and its maximum is v.
7. method according to claim 6, it is characterized in that: in described step (31), the size of the quantity K of user a similar users group is carry out test of many times selection on the data set of institute's foundation after, numerical value when demonstration validation prediction error value is minimum.
8. method according to claim 1, is characterized in that: described step 4 comprises following content of operation:
(41) successively to set
in each parameter according to after step (36) anomaly sieving Value Data, obtain screen after QoS data set
(42) according to formula
calculate, obtain common K QoS predicted value F (u
c, j); Wherein, u
c∈ Q
sim (a)={ u
1, u
2..., u
c..., u
k,
v is
in the total number of QoS data;
(43) to this K predicted value result F (u
c, after j) being weighted process according to the following equation, obtain disappearance QoS data predicted value Forecast when this user a accesses jth Web service:
Wherein,
for similar users u in user a and its similar users group
cuser between similarity, natural number subscript c is the similar users sequence number in similar users group, and its maximum is K.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510039675.1A CN104579850A (en) | 2015-01-27 | 2015-01-27 | Quality of service (QoS) prediction method for Web service under mobile Internet environment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510039675.1A CN104579850A (en) | 2015-01-27 | 2015-01-27 | Quality of service (QoS) prediction method for Web service under mobile Internet environment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN104579850A true CN104579850A (en) | 2015-04-29 |
Family
ID=53095114
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510039675.1A Pending CN104579850A (en) | 2015-01-27 | 2015-01-27 | Quality of service (QoS) prediction method for Web service under mobile Internet environment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104579850A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105071961A (en) * | 2015-07-30 | 2015-11-18 | 中国人民解放军信息工程大学 | Method and device for predicting service quality of Web service |
CN109040214A (en) * | 2018-07-25 | 2018-12-18 | 北京邮电大学 | A kind of service arrangement method that reliability enhances under cloud environment |
CN110060086A (en) * | 2019-03-01 | 2019-07-26 | 汕头大学 | A kind of on-line prediction method based on User reliability in Web cloud service |
CN113364621A (en) * | 2021-06-04 | 2021-09-07 | 浙江大学 | Service quality prediction method under service network environment |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070259673A1 (en) * | 2006-05-04 | 2007-11-08 | Telefonaktiebolaget Lm Ericsson (Publ) | Inactivity monitoring for different traffic or service classifications |
CN103209102A (en) * | 2013-03-11 | 2013-07-17 | 北京邮电大学 | Web quality of service distributed measurement system and method |
-
2015
- 2015-01-27 CN CN201510039675.1A patent/CN104579850A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070259673A1 (en) * | 2006-05-04 | 2007-11-08 | Telefonaktiebolaget Lm Ericsson (Publ) | Inactivity monitoring for different traffic or service classifications |
CN103209102A (en) * | 2013-03-11 | 2013-07-17 | 北京邮电大学 | Web quality of service distributed measurement system and method |
Non-Patent Citations (1)
Title |
---|
LUBAO WANG, QIBO SUN, SHANGGUANG WANG等: ""Web Service QoS Prediction Approach in Mobile Internet Environments"", <IEEE INTERNATIONAL CONFERENCE ON DATA MINING WORKSHOP , 2014 > * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105071961A (en) * | 2015-07-30 | 2015-11-18 | 中国人民解放军信息工程大学 | Method and device for predicting service quality of Web service |
CN105071961B (en) * | 2015-07-30 | 2018-12-14 | 中国人民解放军信息工程大学 | A kind of the service quality prediction technique and device of Web service |
CN109040214A (en) * | 2018-07-25 | 2018-12-18 | 北京邮电大学 | A kind of service arrangement method that reliability enhances under cloud environment |
CN109040214B (en) * | 2018-07-25 | 2020-07-17 | 北京邮电大学 | Service deployment method for enhancing reliability in cloud environment |
CN110060086A (en) * | 2019-03-01 | 2019-07-26 | 汕头大学 | A kind of on-line prediction method based on User reliability in Web cloud service |
CN113364621A (en) * | 2021-06-04 | 2021-09-07 | 浙江大学 | Service quality prediction method under service network environment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102262661B (en) | Web page access forecasting method based on k-order hybrid Markov model | |
US20130246383A1 (en) | Cursor Activity Evaluation For Search Result Enhancement | |
Wu et al. | Fine-grained user profiling for personalized task matching in mobile crowdsensing | |
US20080270549A1 (en) | Extracting link spam using random walks and spam seeds | |
Zhang et al. | LA-LMRBF: Online and long-term web service QoS forecasting | |
CN109918602B (en) | Web data preloading method and system | |
CN109492076A (en) | A kind of network-based community's question and answer website answer credible evaluation method | |
CN104579850A (en) | Quality of service (QoS) prediction method for Web service under mobile Internet environment | |
CN112907128A (en) | Data analysis method, device, equipment and medium based on AB test result | |
CN110458601B (en) | Method and device for processing resource data, computer equipment and storage medium | |
CN105760649A (en) | Big-data-oriented creditability measuring method | |
CN101404591B (en) | Self-adapting dynamic trust weight estimation method | |
CN110454910B (en) | Method and equipment for defrosting of air conditioner | |
CN106547901A (en) | It is a kind of to forward behavior prediction method based on energy-optimised microblog users | |
Jiang et al. | Multifractal temporally weighted detrended cross-correlation analysis of multivariate time series | |
Madhawa et al. | A multi-armed bandit approach for exploring partially observed networks | |
Zhu et al. | BGCL: Bi-subgraph network based on graph contrastive learning for cold-start QoS prediction | |
CN104820719A (en) | Web service creditworthiness measuring method based on context data of user | |
CN102521362B (en) | Web service recommendation method and device | |
CN113010782B (en) | Demand acquisition method, device, electronic equipment and computer readable medium | |
Mochinski et al. | Developing an Intelligent Decision Support System for large-scale smart grid communication network planning | |
CN111598390B (en) | Method, device, equipment and readable storage medium for evaluating high availability of server | |
CN107577698A (en) | A kind of mobile subscriber's preference Forecasting Methodology based on influence power between user | |
CN115345635A (en) | Processing method and device for recommended content, computer equipment and storage medium | |
Jiang et al. | Quality-aided Annotation Service Selection in MLaaS Market |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150429 |
|
WD01 | Invention patent application deemed withdrawn after publication |