CN107808324B - Online commodity credit value calculation method, network transaction platform and computer - Google Patents

Online commodity credit value calculation method, network transaction platform and computer Download PDF

Info

Publication number
CN107808324B
CN107808324B CN201710996982.8A CN201710996982A CN107808324B CN 107808324 B CN107808324 B CN 107808324B CN 201710996982 A CN201710996982 A CN 201710996982A CN 107808324 B CN107808324 B CN 107808324B
Authority
CN
China
Prior art keywords
evaluation
representing
value
user
malicious
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201710996982.8A
Other languages
Chinese (zh)
Other versions
CN107808324A (en
Inventor
裴庆祺
张潘頔
马立川
李子
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.)
Xidian University
Original Assignee
Xidian 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 Xidian University filed Critical Xidian University
Priority to CN201710996982.8A priority Critical patent/CN107808324B/en
Publication of CN107808324A publication Critical patent/CN107808324A/en
Application granted granted Critical
Publication of CN107808324B publication Critical patent/CN107808324B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0609—Buyer or seller confidence or verification

Landscapes

  • Business, Economics & Management (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Strategic Management (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention belongs to the technical field of electronic commerce, and discloses an online commodity credit value calculation method, a network transaction platform and a computer, which comprise the following steps: malicious evaluation detection, user interest evaluation and reputation value calculation. The invention utilizes the two-phase detection theory to detect the slight change of the evaluation model, can process the condition when the malicious user carries out malicious evaluation on all commodities, and can also process the condition when the malicious user only carries out the malicious evaluation on the target commodities. According to the method and the system, the similarity between the target item and the user evaluation item is compared to determine the weights of different user evaluations, so that the degree of interest of the user in the target commodity and the reliability of the user can be more accurately reflected.

Description

Online commodity credit value calculation method, network transaction platform and computer
Technical Field
The invention belongs to the technical field of electronic commerce, and particularly relates to a credit value calculation method of an online commodity, a network transaction platform and a computer.
Background
With the vigorous development of electronic commerce (eBay, Amazon, Taobao, Youtube, Netflix and other network transaction platforms), more than half of the population using the Internet in most regions of the world has experience of online shopping according to the report of the electronic commerce foundation. Electronic commerce has great potential to promote economic development and bring convenience to daily life of people. As more and more goods and services are provided by the network trading platform, the selection range of consumers is also larger and larger. But at the same time, interest-oriented sellers are emerging to offer counterfeit goods. Driven by economic interest, a large number of sellers would like to give their own product an unusually high rating, but have a low rating of their competitor's goods; abnormal evaluations are referred to as malicious evaluations, and those who give malicious evaluations are referred to as attackers. When attackers carry out malicious evaluation on various commodities, the server side can easily determine the credibility of the attackers according to the evaluation history and the existing work of the attackers, and the malicious evaluation of the attackers can be effectively detected. Once an attacker provides only a malicious evaluation of a target commodity (or a small group of target commodities) and gives only a normal evaluation of other commodities, an additional solution is needed. When the malicious evaluation is intensive, it means that a large number of evaluations of the reputation value of the target commodity can be given quickly in a short time. Since the evaluation data is open to all people, an attacker can monitor the evaluation data and give a malicious evaluation. Making it possible for an attacker to manipulate the distribution of malicious evaluations to follow any pattern, with the worst case being that the malicious evaluations are evenly distributed among all the evaluations. How to determine the weights of different user ratings; the weights reflect the reliability of the user. Different users may have different preferences, however, the interests of a user are often concentrated on some topics or categories, rather than being spread across all items; the weight of the rating should reflect the degree of interest of the user in the target good. A patent application "a dynamic Web service trust evaluation method based on user honesty" proposed by southeast university (application No. 2012102796.4 application publication No. CN 102880637B) discloses a dynamic Web service trust evaluation method based on user honesty, which comprises the following specific steps: firstly, automatically mining user preference according to the requirement of a user on service; the method comprises the steps of dividing similar user groups according to user preferences, carrying out evaluation consistency clustering in the preference similar user groups, distinguishing honest users and malicious users, calculating the user honest degree of the preference similar user groups, and dynamically adjusting the weight of subjective evaluation in comprehensive trust calculation according to the change of the user honest degree, so that the influence of the malicious evaluation on Web service trust evaluation is reduced, the accuracy of the trust evaluation is improved, and the user is better guided to select trusted services. The patent application has the following disadvantages: how to deal with the situation that a malicious user only evaluates a target commodity but not all commodities is not considered, once the malicious user only provides malicious evaluation for the target commodity (or a small group of target commodities) and only gives normal evaluation to other commodities, the model cannot sense slight change of the evaluation model and cannot completely evaluate the reputation value of the commodity, so that the user is possibly misled to make wrong decisions. The patent application "a reputation calculation method in a reputation system" (application number 200710122393.3 application publication number CN 10199683B) proposed by the acoustic research institute of the chinese academy of sciences discloses a reputation calculation method in a reputation system, which comprises the following specific steps: the node i sends a request to surrounding nodes of a node j needing to calculate the reputation, and inquires about the evaluation of the surrounding nodes on the node j; returning evaluation to surrounding nodes; according to the deviation degree of each node and the reliability of the deviation degree, the node i corrects the evaluation given by the evaluation node; and the node i fuses the actual experience of the node i and the service node j with the evaluation of the modified evaluation node, and calculates the final credit. The patent application has the following disadvantages: the method model does not consider how to determine the evaluation weights of different nodes (namely different users), and the evaluation weights can reflect the interest degree of the users in the target commodities and the reliability of the users. The evaluation of the user to the service has a great relationship with the preference of the user, and if the evaluation weight of different users cannot be determined, the evaluation or objective ability credibility of a third party cannot cover the users with different preferences, so that the phenomenon of misjudgment is generated.
In summary, the problems of the prior art are as follows: the existing reputation value technical method does not consider how to process the situation that a malicious user only makes malicious evaluation on target commodities, but not on all commodities; how to determine the weights of different node evaluations is not considered, and the evaluation weights cannot reflect the interest degree of the user in the target commodity and the reliability of the user.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a credit value calculation method of an online commodity, a network transaction platform and a computer.
The invention is realized in this way, and a credit value calculation method of online commodities comprises the steps of carrying out malicious evaluation detection and user interest evaluation by a server, and finally carrying out credit value calculation of the commodities by integrating malicious evaluation detection results and user interest. The credit value calculation method of the online commodity comprises the following steps:
step one, malicious evaluation detection:
(1) the server calculates the target commodity i after receiving the user requesttWhen the credit value is requested, the information about the target commodity i is collectedtThe evaluation values of (1) are arranged in time sequence, and every n evaluation values form a group;
(2) the evaluation value of the k-th group was represented as
Figure BDA0001439681910000031
Figure BDA0001439681910000032
Wherein the content of the first and second substances,
Figure BDA0001439681910000033
represents the jth evaluation value in the kth group, and n represents the number of groups of evaluation values;
(3) will give an evaluation value
Figure BDA0001439681910000034
Is represented as
Figure BDA0001439681910000035
Figure BDA0001439681910000036
Wherein the content of the first and second substances,
Figure BDA0001439681910000037
show giving evaluation value
Figure BDA0001439681910000038
The set of users of (a) is,
Figure BDA0001439681910000039
show giving evaluation value
Figure BDA00014396819100000310
The user of (1);
(4) constructing intermediate vectors
Figure BDA00014396819100000311
Figure BDA00014396819100000312
Wherein the content of the first and second substances,
Figure BDA00014396819100000313
the intermediate vector is represented by a vector representing the intermediate vector,
Figure BDA00014396819100000314
denotes the jth evaluation value, μ, in the kth groupk-1Denotes an average value of evaluations in the (k-1) th group after deletion of a malicious evaluation, and when k is 1, μ0Represents the average value of the evaluation values in the first packet, N+Is a positive integer sign;
(5) detection by the following formula
Figure BDA00014396819100000315
Whether there is a malicious evaluation:
Figure BDA00014396819100000316
where n represents the number of sets of evaluation values,
Figure BDA00014396819100000317
representing intermediate vectors
Figure BDA00014396819100000318
Mean value of (1), s2The variance of the samples is represented by the average,
Figure BDA00014396819100000319
representing a hypothesis testing method, if the inequality holds, then
Figure BDA00014396819100000320
If there is a malicious evaluation, the process is executed (6), otherwise, the process is executed
Figure BDA0001439681910000041
If no malicious evaluation exists, executing the step (7);
(6) calculated using the following equation
Figure BDA0001439681910000042
Mean value of
Figure BDA0001439681910000043
Figure BDA0001439681910000044
Wherein the content of the first and second substances,
Figure BDA0001439681910000045
to represent
Figure BDA0001439681910000046
The average value of (a) of (b),
Figure BDA0001439681910000047
to represent
Figure BDA0001439681910000048
Evaluation mean of (1), μk-1Mean values representing the evaluations in group (k-1) after the malicious evaluation was deleted;
(7) calculating the log-likelihood ratio C using the following equationj：
Figure BDA0001439681910000049
Wherein, CjA log likelihood ratio, C, representing the jth evaluation value0＝0，
Figure BDA00014396819100000410
To represent
Figure BDA00014396819100000411
Mean value of (a)2To represent
Figure BDA00014396819100000412
The variance of the medium normal distribution is,
Figure BDA00014396819100000413
representing the jth intermediate vector in the kth group,
Figure BDA00014396819100000414
representing intermediate vectors
Figure BDA00014396819100000415
The mean value of (a); when C is presentjWhen the content of the carbon dioxide is more than or equal to 0.8,
Figure BDA00014396819100000416
is determined to be an abnormal value,
Figure BDA00014396819100000417
is determined to be a malicious evaluation,
Figure BDA00014396819100000418
is judged as a malicious user and is selected from
Figure BDA00014396819100000419
Deleting to obtain a normal user set corresponding to the k-th evaluation value group
Figure BDA00014396819100000420
Executing the step two;
(8) if (5) detecting
Figure BDA00014396819100000421
If there is no malicious evaluation, then
Figure BDA00014396819100000422
Step two, evaluating the user interest degree:
(1) similarity is calculated using the following equation
Figure BDA00014396819100000423
Figure BDA00014396819100000424
Wherein the content of the first and second substances,
Figure BDA00014396819100000425
representing the degree of similarity between the target item and items other than the target item, itIndicating a target commodity, i indicating a commodity other than the target commodity,
Figure BDA00014396819100000426
representing a set of labels, theta represents
Figure BDA00014396819100000427
Where Σ is a sum sign, rel represents the applied intensity, range 0 to 1, 0 represents no intensity at all, 1 represents the maximum intensity;
(2) calculating the interest d of the user u by using the following formulau：
Figure BDA0001439681910000051
Wherein d isuRepresenting the user's interest level, ncRepresenting a target item itThe number of the (c) component (a),
Figure BDA0001439681910000052
indicating the number of items that user u has evaluated, sigma indicates a sum symbol,
Figure BDA0001439681910000053
representing the degree of similarity between the target item and items other than the target item,
Figure BDA0001439681910000054
representation collection
Figure BDA0001439681910000055
The number of elements in (1);
step three, calculating a reputation value:
(1) target product i is calculated using the following formulatReputation value of (c):
Figure BDA0001439681910000056
where rep represents the reputation value of the good, itRepresenting the target commodity, k representing the k-th group evaluation value, δ representing the weight of the evaluation in the most recent evaluation group, rel representing the application intensity,
Figure BDA0001439681910000057
set of users u representing deleted malicious users, duWhich represents the level of interest of the user,
Figure BDA0001439681910000058
indicating that user u is about target item itA satisfaction degree of 1 to 5, 1 being very unsatisfactory, 5 being very satisfactory;
(2) target commodity itIs returned to the requested user u.
Further, the hypothesis testing method
Figure BDA0001439681910000059
Refers to a hypothesis testing method for testing whether the overall variances of two normal random variables are equal,
Figure BDA00014396819100000510
representing a first degree of freedom of 1, a second degree of freedom of n-1, with a degree of confidence of alpha
Figure BDA00014396819100000511
Checking the value, and obtaining the result by looking up an F-distribution table.
Further, s is2And
Figure BDA00014396819100000512
the specific calculation method is as follows:
Figure BDA00014396819100000513
wherein s is2Representing the sample variance, n representing the number of sets of evaluation values,
Figure BDA00014396819100000514
representing the jth intermediate vector in the kth group,
Figure BDA00014396819100000515
representing intermediate vectors
Figure BDA00014396819100000516
Is measured.
Further, said CjThe log-likelihood ratio of (d) means:
Figure BDA0001439681910000061
wherein the content of the first and second substances,
Figure BDA0001439681910000062
and
Figure BDA0001439681910000063
respectively represent theta0And theta1Probability density function of yjRepresenting a series of samples, N, generated by a random process of a parameter θ+Are positive integer symbols.
Further, the σ2The specific calculation method is as follows:
Figure BDA0001439681910000064
wherein σ2To represent
Figure BDA0001439681910000065
Where the variance of the normal distribution, n denotes the number of groups of evaluation values, j denotes the number of evaluation values, Σ is the sum sign,
Figure BDA0001439681910000066
representing the jth intermediate vector in the kth group,
Figure BDA0001439681910000067
representing intermediate vectors
Figure BDA0001439681910000068
Is measured.
Another object of the present invention is to provide a network trading platform using the reputation value calculation method for online goods.
Another object of the present invention is to provide a computer using the reputation value calculation method for an online good.
Another object of the present invention is to provide a server using the reputation value calculation method for an online good.
Another object of the present invention is to provide a computer program using the reputation value calculation method for an online good.
The invention provides a reputation calculation framework to obtain a reliable reputation value of any online commodity to help consumers make correct purchasing decisions. The framework comprises three modules, namely a malicious rating detection module, an interestingness calculation module and a reputation calculation module. Collected by the data collector, the malicious rating detector may detect malicious ratings by a two-phase detection method. Experimental data show that, in the prior art, when malicious evaluations are not filtered, the average absolute error between a calculation result and an ideal value linearly increases along with the increase of the proportion of the malicious evaluations in all evaluations. The invention can maintain the average absolute error of the calculation result near 0, which shows that the calculated reputation value is very close to an ideal value and can effectively process uniformly distributed malicious evaluation. According to the method and the system, the similarity between the target item and the user evaluation item is compared to determine the weights of different user evaluations, so that the degree of interest of the user in the target commodity and the reliability of the user can be more accurately reflected.
Drawings
Fig. 1 is a flowchart of a reputation value calculation method for an online commodity according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
E-commerce has the experience of online shopping in most of the global areas, and more than half of people using the Internet have the experience of online shopping; electronic commerce has great potential to promote economic development and bring convenience to daily life of people.
The following detailed description of the principles of the invention is provided in connection with the accompanying drawings.
As shown in fig. 1, the method for calculating the reputation value of an online commodity according to an embodiment of the present invention includes the following steps:
s101: malicious evaluation detection;
s102: evaluating the interest degree of the user;
s103: and calculating a reputation value.
The application of the principles of the present invention will now be described in further detail with reference to specific embodiments.
The method for calculating the credit value of the online commodity provided by the embodiment of the invention comprises the following steps:
step one, malicious evaluation detection;
(1) the server calculates the target commodity i after receiving the user requesttWhen the credit value is requested, the information about the target commodity i is collectedtThe evaluation values of (a) are arranged in chronological order, one group for every n evaluation values. The server expresses the evaluation value of the kth group as
Figure BDA0001439681910000071
The specific expression method is as follows:
Figure BDA0001439681910000072
wherein the content of the first and second substances,
Figure BDA0001439681910000073
represents the jth evaluation value in the kth group, and n represents the number of groups of evaluation values;
(2) the server will give an evaluation value
Figure BDA0001439681910000081
Is represented as
Figure BDA0001439681910000082
Figure BDA0001439681910000083
Wherein the content of the first and second substances,
Figure BDA0001439681910000084
show giving evaluation value
Figure BDA0001439681910000085
The set of users of (a) is,
Figure BDA0001439681910000086
show giving evaluation value
Figure BDA0001439681910000087
The user of (1);
(3) constructing intermediate vectors
Figure BDA0001439681910000088
Will be provided with
Figure BDA0001439681910000089
Is converted into
Figure BDA00014396819100000810
Figure BDA00014396819100000811
Wherein the content of the first and second substances,
Figure BDA00014396819100000812
the intermediate vector is represented by a vector representing the intermediate vector,
Figure BDA00014396819100000813
denotes the jth evaluation value, μ, in the kth groupk-1Denotes an average value of evaluations in the (k-1) th group after deletion of a malicious evaluation, and when k is 1, μ0Represents the average value of the evaluation values in the first packet, N+Is a positive integer sign;
(4) server utilizes the following formula to detect
Figure BDA00014396819100000814
Whether there is a malicious evaluation:
Figure BDA00014396819100000815
where n represents the number of sets of evaluation values,
Figure BDA00014396819100000816
representing intermediate vectors
Figure BDA00014396819100000817
The average value of (a) of (b),
Figure BDA00014396819100000818
s2the variance of the samples is represented by the average,
Figure BDA00014396819100000819
representing hypothesis testing methods
Figure BDA00014396819100000820
Refers to a hypothesis testing method for testing whether the overall variances of two normal random variables are equal,
Figure BDA00014396819100000821
representing a first degree of freedom of 1, a second degree of freedom of n-1, with a degree of confidence of alpha
Figure BDA00014396819100000822
Checking the value, and obtaining the result by looking up an F-distribution table. If the inequality is true, then
Figure BDA00014396819100000823
If there is a malicious evaluation, the process is executed (6), otherwise, the process is executed
Figure BDA00014396819100000824
If no malicious evaluation exists, executing (7);
(5) the server calculates using the following equation
Figure BDA00014396819100000825
Mean value of
Figure BDA00014396819100000826
Figure BDA00014396819100000827
Wherein the content of the first and second substances,
Figure BDA00014396819100000828
to represent
Figure BDA00014396819100000829
The average value of (a) of (b),
Figure BDA00014396819100000830
to represent
Figure BDA00014396819100000831
Evaluation mean of (1), μk-1Mean values representing the evaluations in group (k-1) after the malicious evaluation was deleted;
(6) the server calculates the log-likelihood ratio C using the following equationj：
Figure BDA0001439681910000091
Wherein, C0＝0，
Figure BDA0001439681910000092
To represent
Figure BDA0001439681910000093
Mean value of (a)2To represent
Figure BDA0001439681910000094
The variance of the medium normal distribution is,
Figure BDA0001439681910000095
Figure BDA0001439681910000096
representing the jth intermediate vector in the kth group,
Figure BDA0001439681910000097
representing intermediate vectors
Figure BDA0001439681910000098
The mean value of (a); when C is presentjWhen the content of the carbon dioxide is more than or equal to 0.8,
Figure BDA0001439681910000099
is determined to be an abnormal value,
Figure BDA00014396819100000910
is determined to be a malicious evaluation,
Figure BDA00014396819100000911
is judged as a malicious user and is selected from
Figure BDA00014396819100000912
Deleting to obtain a normal user set corresponding to the k-th evaluation value group
Figure BDA00014396819100000913
Executing the step two;
in the above formula CjA log likelihood ratio, C, representing the jth evaluation valuejSpecific calculation method of log likelihood ratioThe method comprises the following steps:
Figure BDA00014396819100000914
wherein the content of the first and second substances,
Figure BDA00014396819100000915
and
Figure BDA00014396819100000916
respectively represent theta0And theta1Probability density function of yjRepresenting a series of samples, N, generated by a random process of a parameter θ+Is a positive integer sign;
(7) if (4) detecting
Figure BDA00014396819100000917
If there is no malicious evaluation, then
Figure BDA00014396819100000918
Step two: and evaluating the interestingness of the user.
(1) The server calculates the similarity using the following equation
Figure BDA00014396819100000919
Figure BDA00014396819100000920
Wherein the content of the first and second substances,
Figure BDA00014396819100000921
representing the degree of similarity between the target item and items other than the target item, itIndicating a target commodity, i indicating a commodity other than the target commodity,
Figure BDA00014396819100000922
representing a set of labels, theta represents
Figure BDA00014396819100000923
With Σ being a sum sign, rel representing the applied intensity, ranging from 0 to 1, 0 representing no intensity at all, 1 representing maximum intensity;
(2) the server calculates the interest d of the user u by using the following formulau：
Figure BDA0001439681910000101
Wherein d isuRepresenting the user's interest level, ncRepresenting a target item itThe number of the (c) component (a),
Figure BDA0001439681910000102
indicating the number of items that user u has evaluated, sigma indicates a sum symbol,
Figure BDA0001439681910000103
representing the degree of similarity between the target item and items other than the target item,
Figure BDA0001439681910000104
representation collection
Figure BDA0001439681910000105
Number of elements in (1).
Step three: and (3) calculating a reputation value:
(1) the server calculates the target commodity i using the following equationtReputation value of (c):
Figure BDA0001439681910000106
where rep represents the reputation value of the good, itRepresenting the target commodity, k representing the k-th group evaluation value, δ representing the weight of the evaluation in the most recent evaluation group, rel representing the application intensity,
Figure BDA0001439681910000107
to representSet of users u from which malicious users have been deleted, duWhich represents the level of interest of the user,
Figure BDA0001439681910000108
indicating that user u is about target item itA satisfaction degree of 1 to 5, 1 being very unsatisfactory, 5 being very satisfactory;
(2) the server sends the target commodity itIs returned to the requested user u.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.

Claims (8)

1. A credit value calculation method of an online commodity is characterized by comprising the following steps:
step one, malicious evaluation detection:
(1) the server calculates the target commodity i after receiving the user requesttWhen the credit value is requested, the information about the target commodity i is collectedtThe evaluation values of (1) are arranged in time sequence, and every n evaluation values form a group;
(2) the evaluation value of the k-th group was represented as
Figure FDA0003021419100000011
Figure FDA0003021419100000012
Wherein the content of the first and second substances,
Figure FDA0003021419100000013
represents the jth evaluation value in the kth group, and n represents the number of groups of evaluation values;
(3) will give an evaluation value
Figure FDA0003021419100000014
Is represented as
Figure FDA0003021419100000015
Figure FDA0003021419100000016
Wherein the content of the first and second substances,
Figure FDA0003021419100000017
show giving evaluation value
Figure FDA0003021419100000018
The set of users of (a) is,
Figure FDA0003021419100000019
show giving evaluation value
Figure FDA00030214191000000110
The user of (1);
(4) constructing intermediate vectors
Figure FDA00030214191000000111
Figure FDA00030214191000000112
Wherein the content of the first and second substances,
Figure FDA00030214191000000113
the intermediate vector is represented by a vector representing the intermediate vector,
Figure FDA00030214191000000114
denotes the jth evaluation value, μ, in the kth groupk-1Denotes an average value of evaluations in the (k-1) th group after deletion of a malicious evaluation, and when k is 1, μ0To representMean value of evaluation values in the first group, N+Is a positive integer sign;
(5) detection by the following formula
Figure FDA00030214191000000115
Whether there is a malicious evaluation:
Figure FDA00030214191000000116
where n represents the number of sets of evaluation values,
Figure FDA00030214191000000117
representing intermediate vectors
Figure FDA00030214191000000118
Mean value of (1), s2The variance of the samples is represented by the average,
Figure FDA00030214191000000119
representing a hypothesis testing method, if the inequality holds, then
Figure FDA00030214191000000120
If there is a malicious evaluation, the process is executed (6), otherwise, the process is executed
Figure FDA00030214191000000121
If no malicious evaluation exists, executing the step (8);
(6) calculated using the following equation
Figure FDA00030214191000000122
Mean value of
Figure FDA00030214191000000123
Figure FDA00030214191000000124
Wherein the content of the first and second substances,
Figure FDA0003021419100000021
to represent
Figure FDA0003021419100000022
The average value of (a) of (b),
Figure FDA0003021419100000023
to represent
Figure FDA0003021419100000024
Evaluation mean of (1), μk-1Mean values representing the evaluations in group (k-1) after the malicious evaluation was deleted;
(7) calculating the log-likelihood ratio C using the following equationj：
Figure FDA0003021419100000025
Wherein, CjA log likelihood ratio, C, representing the jth evaluation value0＝0，
Figure FDA0003021419100000026
To represent
Figure FDA0003021419100000027
Mean value of (a)2To represent
Figure FDA0003021419100000028
The variance of the medium normal distribution is,
Figure FDA0003021419100000029
representing the jth intermediate vector in the kth group,
Figure FDA00030214191000000210
representing intermediate vectors
Figure FDA00030214191000000211
The mean value of (a); when C is presentjWhen the content of the carbon dioxide is more than or equal to 0.8,
Figure FDA00030214191000000212
is determined to be an abnormal value,
Figure FDA00030214191000000213
is determined to be a malicious evaluation,
Figure FDA00030214191000000214
is judged as a malicious user and is selected from
Figure FDA00030214191000000215
Deleting to obtain a normal user set corresponding to the k-th evaluation value group
Figure FDA00030214191000000216
Executing the step two;
(8) if (5) detecting
Figure FDA00030214191000000217
If there is no malicious evaluation, then
Figure FDA00030214191000000218
Step two, evaluating the user interest degree:
(1) similarity is calculated using the following equation
Figure FDA00030214191000000219
Figure FDA00030214191000000220
Wherein the content of the first and second substances,
Figure FDA00030214191000000221
representing the degree of similarity between the target item and items other than the target item, itIndicating a target commodity, i indicating a commodity other than the target commodity,
Figure FDA00030214191000000222
representing a set of labels, theta represents
Figure FDA00030214191000000223
Where Σ is a sum sign, rel represents the applied intensity, range 0 to 1, 0 represents no intensity at all, 1 represents the maximum intensity;
(2) calculating the interest d of the user u by using the following formulau：
Figure FDA00030214191000000224
Wherein d isuRepresenting the user's interest level, ncRepresenting a target item itThe number of the (c) component (a),
Figure FDA00030214191000000225
indicating the number of items that user u has evaluated, sigma indicates a sum symbol,
Figure FDA0003021419100000031
representing the degree of similarity between the target item and items other than the target item,
Figure FDA0003021419100000032
representation collection
Figure FDA0003021419100000033
The number of elements in (1);
step three, calculating a reputation value:
(1) target product i is calculated using the following formulatReputation value of (c):
Figure FDA0003021419100000034
where rep represents the reputation value of the good, itRepresenting the target commodity, k representing the k-th group evaluation value, δ representing the weight of the evaluation in the most recent evaluation group, rel representing the application intensity,
Figure FDA0003021419100000035
set of users u representing deleted malicious users, duIndicates the degree of interest of the user, ru,itIndicating that user u is about target item itA satisfaction degree of 1 to 5, 1 being very unsatisfactory, 5 being very satisfactory;
(2) target commodity itIs returned to the requested user u.
2. The reputation value calculation method of an online good according to claim 1, wherein the hypothesis testing method
Figure FDA0003021419100000036
Refers to a hypothesis testing method for testing whether the overall variances of two normal random variables are equal,
Figure FDA0003021419100000037
representing a first degree of freedom of 1, a second degree of freedom of n-1, with a degree of confidence of alpha
Figure FDA0003021419100000038
Checking the value, and obtaining the result by looking up an F-distribution table.
3. The reputation value calculation method of an online good according to claim 1, wherein s is2And
Figure FDA00030214191000000316
is particularly shownThe calculation method comprises the following steps:
Figure FDA0003021419100000039
wherein s is2Representing the sample variance, n representing the number of sets of evaluation values,
Figure FDA00030214191000000310
representing the jth intermediate vector in the kth group,
Figure FDA00030214191000000311
representing intermediate vectors
Figure FDA00030214191000000312
Is measured.
4. The method for calculating a reputation value of an online good according to claim 1, wherein C isjThe log-likelihood ratio of (d) means:
Figure FDA00030214191000000313
wherein the content of the first and second substances,
Figure FDA00030214191000000314
and
Figure FDA00030214191000000315
respectively represent theta0And theta1Probability density function of yjRepresenting a series of samples, N, generated by a random process of a parameter θ+Are positive integer symbols.
5. The method of calculating a reputation value of an online good according to claim 1, wherein σ is the amount of time that the product has been placed on the online good2The specific calculation method is as follows:
Figure FDA0003021419100000041
wherein σ2To represent
Figure FDA0003021419100000042
Where the variance of the normal distribution, n denotes the number of groups of evaluation values, j denotes the number of evaluation values, Σ is the sum sign,
Figure FDA0003021419100000043
representing the jth intermediate vector in the kth group,
Figure FDA0003021419100000044
representing intermediate vectors
Figure FDA0003021419100000045
Is measured.
6. A network transaction system using the method for calculating the credit value of an online commodity according to any one of claims 1 to 5.
7. A computer using the method for calculating the reputation value of an online good according to any one of claims 1 to 5.
8. A server using the method for calculating the reputation value of an online good according to any one of claims 1 to 5.
CN201710996982.8A 2017-10-20 2017-10-20 Online commodity credit value calculation method, network transaction platform and computer Active CN107808324B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710996982.8A CN107808324B (en) 2017-10-20 2017-10-20 Online commodity credit value calculation method, network transaction platform and computer

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710996982.8A CN107808324B (en) 2017-10-20 2017-10-20 Online commodity credit value calculation method, network transaction platform and computer

Publications (2)

Publication Number Publication Date
CN107808324A CN107808324A (en) 2018-03-16
CN107808324B true CN107808324B (en) 2021-05-28

Family

ID=61592358

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710996982.8A Active CN107808324B (en) 2017-10-20 2017-10-20 Online commodity credit value calculation method, network transaction platform and computer

Country Status (1)

Country Link
CN (1) CN107808324B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108665339B (en) * 2018-03-27 2021-11-23 北京航空航天大学 E-commerce product reliability index based on subjective emotion measure and implementation method thereof
CN111915210A (en) * 2020-08-11 2020-11-10 曲阜师范大学 Trust management method in distributed Internet of things based on block chain

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103957062A (en) * 2014-03-31 2014-07-30 西安电子科技大学 Method for evaluating reputation of cognitive users in distributed cognitive radio network
CN104350490A (en) * 2012-04-11 2015-02-11 新加坡国立大学 Methods, apparatuses and computer-readable mediums for organizing data relating to a product
CN106296282A (en) * 2016-08-08 2017-01-04 南京大学 A kind of net purchase Product evaluation method marked based on user comment and history
CN106570525A (en) * 2016-10-26 2017-04-19 昆明理工大学 Method for evaluating online commodity assessment quality based on Bayesian network
CN106779858A (en) * 2016-12-26 2017-05-31 西安理工大学 A kind of product analysis method based on multidimensional perception information semantic level association

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160034968A1 (en) * 2014-07-31 2016-02-04 Huawei Technologies Co., Ltd. Method and device for determining target user, and network server

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104350490A (en) * 2012-04-11 2015-02-11 新加坡国立大学 Methods, apparatuses and computer-readable mediums for organizing data relating to a product
CN103957062A (en) * 2014-03-31 2014-07-30 西安电子科技大学 Method for evaluating reputation of cognitive users in distributed cognitive radio network
CN106296282A (en) * 2016-08-08 2017-01-04 南京大学 A kind of net purchase Product evaluation method marked based on user comment and history
CN106570525A (en) * 2016-10-26 2017-04-19 昆明理工大学 Method for evaluating online commodity assessment quality based on Bayesian network
CN106779858A (en) * 2016-12-26 2017-05-31 西安理工大学 A kind of product analysis method based on multidimensional perception information semantic level association

Also Published As

Publication number Publication date
CN107808324A (en) 2018-03-16

Similar Documents

Publication Publication Date Title
Kumar et al. Rev2: Fraudulent user prediction in rating platforms
Su et al. TAP: A personalized trust-aware QoS prediction approach for web service recommendation
CN109711955B (en) Poor evaluation early warning method and system based on current order and blacklist base establishment method
US20140122294A1 (en) Determining a characteristic group
Bidgoly et al. Benchmarking reputation systems: A quantitative verification approach
CN112231570B (en) Recommendation system support attack detection method, device, equipment and storage medium
Shi et al. A social sensing model for event detection and user influence discovering in social media data streams
Kaur et al. A comparative analysis of structural graph metrics to identify anomalies in online social networks
CN108335131B (en) Method and device for estimating age bracket of user and electronic equipment
Zhang et al. A trust model stemmed from the diffusion theory for opinion evaluation
Liu et al. An integrated clustering‐based approach to filtering unfair multi‐nominal testimonies
Allahbakhsh et al. Representation and querying of unfair evaluations in social rating systems
Serra et al. Deeptrust: An automatic framework to detect trustworthy users in opinion-based systems
KR20100002915A (en) Behavior based method for filtering out unfair rating in trust model
CN107808324B (en) Online commodity credit value calculation method, network transaction platform and computer
Wang et al. An unsupervised strategy for defending against multifarious reputation attacks
Raghebi et al. A new trust evaluation method based on reliability of customer feedback for cloud computing
CN109313541A (en) For showing and the user interface of comparison attacks telemetering resource
Al-Tamimi et al. Employing cluster-based class decomposition approach to detect phishing websites using machine learning classifiers
Davoudi Effects of user interactions on online social recommender systems
Mohammadinejad et al. Employing personality feature to rank the influential users in signed networks
Huang et al. A hybrid decision approach to detect profile injection attacks in collaborative recommender systems
Guesmi et al. FFED: a novel strategy based on fast entropy to detect attacks against trust computing in cloud
Farooq et al. A multi source product reputation model
CN108985895B (en) Method for obtaining merchant credit value in electronic commerce

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