CN110851716B - Personalized group recommendation method based on maximum harmony - Google Patents

Personalized group recommendation method based on maximum harmony Download PDF

Info

Publication number
CN110851716B
CN110851716B CN201911094944.9A CN201911094944A CN110851716B CN 110851716 B CN110851716 B CN 110851716B CN 201911094944 A CN201911094944 A CN 201911094944A CN 110851716 B CN110851716 B CN 110851716B
Authority
CN
China
Prior art keywords
user
users
group
degree
inconsistent
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
CN201911094944.9A
Other languages
Chinese (zh)
Other versions
CN110851716A (en
Inventor
吴坚
曹溟铄
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.)
Shanghai Maritime University
Original Assignee
Shanghai Maritime 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 Shanghai Maritime University filed Critical Shanghai Maritime University
Priority to CN201911094944.9A priority Critical patent/CN110851716B/en
Publication of CN110851716A publication Critical patent/CN110851716A/en
Application granted granted Critical
Publication of CN110851716B publication Critical patent/CN110851716B/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
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9536—Search customisation based on social or collaborative filtering

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The invention relates to a personalized group recommendation method based on maximum harmony, which comprises the following steps: evaluating the consistency degree of group users; identifying inconsistent users in the population; constructing an individual group recommendation model based on the maximum harmony degree, and solving individual group recommendation feedback parameters; calculating the personalized group recommendation feedback opinions; and determining the weight of the individual user according to the consistency degree, and aggregating the preferences of the individual user to determine a group recommendation scheme. The invention eliminates the inconsistency among users, increases the harmony of the system and improves the user-friendly degree of the group recommendation system; the differences of individual demands are eliminated through the personalized feedback parameters, and therefore a more accurate group recommendation system solution is achieved.

Description

Personalized group recommendation method based on maximum harmony
Technical Field
The invention belongs to the field of group recommendation algorithms. In particular to a personalized group recommendation method based on maximum harmony.
Background
With the development of economy and the improvement of living standard of people, the demands for activities such as cultural entertainment and the like are stronger. In the process of participating in cultural entertainment activities, how to better select an activity scheme to reasonably plan the activity process under the limitation of diversified requirements, expenses and time is a problem which is paid attention to and urgently needs to be solved by related marketing departments and internet platforms. With the development of the technology, the storage and processing functions of mass data of related enterprises and internet platforms are enhanced, and personalized solutions can be recommended according to historical scores and behavior data of a large number of users.
However, the personalized recommendation method does not meet the recommendation requirements in the actual social context. In an actual social scene, an activity is often participated by multiple people, such as dinning, traveling, watching movies and other activities. Therefore, the group recommendation technology is developed at the discretion, the group recommendation can effectively solve the problem of group social activities, and the group recommendation method has wide application prospects.
Currently, one of the research paradigms of the group recommendation algorithm is the understanding of the group decision behavior. In 2017, Akshita Agarwal and the like propose a method for improving the satisfaction degree of group recommendation based on Hungarian operator and minimum distress operator with priority; in 2018, TRAN DANG QUANG VINH et al proposed that a good group recommendation algorithm should be a group decision process and provide a flexible method to simulate a complex group decision process. Lara Quijano-S-nchez and the like provide a group recommendation method considering human cognitive information and social relationship information in a social network.
In the group recommendation system, one main task is to eliminate inconsistency among group members, and a feedback mechanism is a main method for eliminating the inconsistency in a traditional mode; however, the traditional feedback method mainly adopts random discrete feedback parameters, and the selection of the feedback parameters is not strictly verified, so that the opinion of inconsistent users is forced to change. Furthermore, for the case where there are multiple inconsistent users in the system, it is not reasonable to use the same feedback parameters.
Disclosure of Invention
In order to eliminate the inconsistency of users in the group recommendation system, the personalized group recommendation method based on the maximum harmony degree provides a harmony degree function, improves the harmony degree of the system while eliminating the inconsistency among the users, and improves the user-friendly degree of the group recommendation system; differences of individual requirements are eliminated through personalized feedback parameters, and therefore a more accurate solution of the group recommendation system is achieved.
The invention relates to a personalized group recommendation method based on maximum harmony, which comprises the following steps:
step 1, evaluating the consistency degree of group users, and measuring the consistency between individual users and group users from three levels of elements, schemes and matrixes;
step 2, identifying inconsistent users in the group from the three levels of individuals, schemes and elements;
step 3, constructing an individualized group recommendation model based on the maximum harmony and solving individualized group recommendation feedback parameters; the maximum harmony degree of group users is an objective function, the feedback degree is controlled by using personalized feedback parameters, and a user consistency boundary is taken as a constraint;
step 4, calculating the feedback opinions recommended by the personalized groups;
and 5, determining the weight of the individual user according to the consistency degree based on the BUM function, and aggregating the preference of the individual user to determine a group recommendation scheme.
The invention introduces the concept of a harmony function, provides a personalized group recommendation method based on maximum harmony, and aims at the user groups with inconsistency. Firstly, the method can provide visual user consistency degree, provide personalized feedback parameters according to the user consistency degree, reduce the degree that the user is forced to modify the preference degree, and provide better user experience; secondly, the method can minimize the adjustment degree of inconsistent user preferences, and improves the harmony of the group recommendation system.
For user consistency evaluation, the main content is to measure the consistency between individual users and group users from three levels of matrixes, schemes and elements; the consistency measurement of the user group is mainly applied to the identification process of inconsistent users in the group, the common identification degree of individual users is fully detected in multiple dimensions, and the problem that the invisible inconsistent users cannot be identified in a single measurement is solved.
For group inconsistent user identification, visualization processing of three levels of inconsistent sets of ECH, ACH and APS is involved; compared with the traditional scheme, the method can more intuitively sense the difference between the inconsistent individuals and the group, thereby better providing a feedback scheme and being beneficial to enabling the inconsistent users to receive feedback more efficiently.
Drawings
FIG. 1: a flow chart of a personalized group recommendation method based on maximum harmony;
FIG. 2: effect diagrams of the traditional method for eliminating inconsistency;
FIG. 3: effect graphs of the non-personalized recommendation method based on the maximum harmony degree;
FIG. 4: and personalizing an effect graph of the recommendation method based on the maximum harmony.
Detailed Description
The invention provides a personalized group recommendation method based on maximum harmony, and when the users in a group recommendation system are inconsistent, the inconsistency can be better eliminated by the method.
The method of the present invention, as shown in FIG. 1, comprises the following steps:
step 1, evaluating the consistency degree of group users;
step 2, identifying inconsistent users in the group;
step 3, constructing an individualized group recommendation model based on the maximum harmony and solving individualized group recommendation feedback parameters; the method comprises the following steps of controlling the feedback degree by utilizing an individualized feedback parameter with the maximum harmony of group users as an objective function, and using a user consistency boundary as a constraint;
step 4, calculating the personalized group recommendation feedback opinions;
and 5, determining the weight of the individual user according to the consistency degree based on the BUM function, and aggregating the preference of the individual user to determine a group recommendation scheme.
The following describes the personalized group recommendation method based on maximum harmony degree in detail.
Step 1, evaluating the consistency degree of group users;
and (3) evaluating the consistency of user element levels:
Figure BDA0002268038640000041
user scheme hierarchical consistency evaluation:
Figure BDA0002268038640000042
and (3) evaluating the consistency of the user matrix levels:
Figure BDA0002268038640000043
wherein m is the number of group users, p is the number of recommended schemes, and q is the number of group consideration criteria. Here, first, the present invention will be described with respect to k, s, h, i, j appearing more frequently:
k represents a certain user in the user group (k ═ 1.., m);
s represents users other than the user to be measured in the user group (s ═ 1.., m, and s ≠ k);
h represents an inconsistent user sub-population of the user population, (h is a proper subset of the set { 1., m });
i represents a certain alternative, (i ═ 1.., p);
j represents a certain decision criterion, (j ═ 1.., q);
the specific symbols referred to in equations (1), (2), and (3) will be explained again:
Figure BDA0002268038640000051
representing the grade of the measured user on the alternative scheme i under the decision criterion j;
Figure BDA0002268038640000052
representing the scores of other users except the measured user k on the alternative scheme i under the decision criterion j;
Figure BDA0002268038640000053
representing the distance between the measured user k and the other users s (s 1., m and s not equal to k),
Figure BDA0002268038640000054
then it is determined that,
Figure BDA0002268038640000055
Figure BDA0002268038640000056
are respectively
Figure BDA0002268038640000057
The elements (c):
wherein,
Figure BDA0002268038640000058
representing the trust degree of the user k to the scheme i under the evaluation criterion j, namely how feasible the user considers;
Figure BDA0002268038640000059
representing the distrust degree of the user k to the scheme i under the evaluation criterion j, namely how much the user considers to be infeasible;
Figure BDA00022680386400000510
representing the trust degree of the user s to the scheme i under the evaluation criterion j, namely how feasible the user considers;
Figure BDA00022680386400000511
indicating how much the user s does not trust the solution i under the evaluation criterion j, i.e. how much the user considers to be infeasible.
Step 2, identifying inconsistent users in the group;
inconsistent individual user identification:
ECH＝{h|ACDh＜β} (4)
inconsistent user profile identification:
Figure BDA00022680386400000512
inconsistent user element identification:
Figure BDA00022680386400000513
wherein β is a consistency threshold predefined by the group recommendation system;
ECH＝{h|ACDh< β is the set of all users with a degree of consensus less than a threshold.
With respect to ACDhWith ACDkDescription of the relationship between: ACDkThe representation is the degree of common knowledge of any one of all users, and h is the inconsistent user in the group, i.e., ACDhIndicating a degree of consensus for which any degree of consensus among all users is less than a threshold value beta.
Step 3, constructing an individualized group recommendation model based on the maximum harmony and solving individualized group recommendation feedback parameters;
the objective function being the maximization of the function of harmony, i.e.
Figure BDA0002268038640000061
Maximization;
the constraint condition is that all users (including consistent users and inconsistent users) meet the consistency requirement of the recommendation system, namely ACDhβ (h ═ 1.. multidot.n) and ACDk≥β(k＝n+1,...,m)。
That is to say that the temperature of the molten steel,
Figure BDA0002268038640000062
wherein, deltahPersonalized feedback parameters for inconsistent users;
n is the number of inconsistent users in the user group;
APS represents the set of all inconsistent elements;
Figure BDA0002268038640000063
Figure BDA0002268038640000064
Figure BDA0002268038640000065
representing the average distance sum between some inconsistent user h and the average opinion of other users than it.
Step 4, calculating the personalized group recommendation feedback opinions;
Figure BDA0002268038640000071
wherein,
Figure BDA0002268038640000072
representing the grading of the alternative scheme i by the inconsistent users under the decision criterion j;
Figure BDA0002268038640000073
representing the interactive scores of the inconsistent users on the alternative schemes i under the decision criterion j;
Figure BDA0002268038640000074
an evaluation element representing an inconsistent user;
Figure BDA0002268038640000075
an evaluation element representing any one of m users, so that the value of k is 1-m; and after the elements of the m users are summed, subtracting the evaluation elements of the inconsistent users, namely the element sum of all the users except the inconsistent user h. Equation (8) can be embodied in the interaction process, and the interaction objects of the inconsistent user h are all users except the inconsistent user h, including other inconsistent users.
Step 5, determining the weight of the individual user according to the consistency degree based on the BUM function, and utilizing a monotone increasing function
Figure BDA0002268038640000076
Individual weight
Figure BDA0002268038640000077
Wherein,
Figure BDA0002268038640000078
σ (k) is the rank value of ACD.
According to
Figure BDA0002268038640000079
Aggregating individual user preferences; finally, the clusters are aggregated according to the weightsAnd generating a scheme sequence by the scheme, and recommending the optimal scheme to the related group. Wherein R isiA scoring matrix representing user i (i ═ 1.., m), and CR represents a group user scoring matrix.
Specifically, the consistency degree of the group users is evaluated, and the consensus degree in the group decision research process is adopted for evaluation. The method mainly comprises three layers: (1) evaluating the consistency of user element levels; (2) evaluating the level consistency of the user scheme; (3) and (5) evaluating the consistency of the user matrix hierarchy.
And identifying the inconsistent users in the group comprises three levels corresponding to the consistency degree calculation and is a process inverse to the consistency degree calculation. The method mainly comprises the following three steps: (1) inconsistent individual user identification; (2) inconsistent user profile identification; (3) the user elements are not identified in agreement.
Wherein, the harmony measure function is expressed as:
Figure BDA0002268038640000081
wherein,
Figure BDA0002268038640000082
representing the original scoring matrix of the inconsistent users h;
Figure BDA0002268038640000083
representing the modified scoring matrix of the inconsistent user h; APShA set of inconsistent elements representing user h; # APShIndicating the number of inconsistent elements for inconsistent user h. Through the three-level identification method, the inconsistent elements (i, j) in the inconsistent user h scoring matrix can be identified and feedback adjustment is carried out.
The derivation can obtain:
Figure BDA0002268038640000084
the derivation process is as follows: order to
Figure BDA0002268038640000085
Figure BDA0002268038640000086
Figure BDA0002268038640000087
Figure BDA0002268038640000088
Figure BDA0002268038640000089
Order to
Figure BDA00022680386400000810
De equation
Figure BDA00022680386400000811
The performance of the method of the invention is analyzed in detail below by three sets of experiments:
first, a trust function is employed
Figure BDA0002268038640000091
The trust function is a preference expression mode commonly used in decision-making problems, and can effectively express fuzzy and uncertain information of a user. The trust score calculation method comprises the following steps:
Figure BDA0002268038640000092
wherein,
Figure BDA0002268038640000093
representing user pair ratingsThe degree of support of the object is,
Figure BDA0002268038640000094
indicating the degree of unsupportability of the user to the evaluation object. As user 1 for travel location scenario 1: hangzhou is based on the following main factors of 1: the score at distance was (0.4, 0.6), indicating that the user 1 considered the travel to the Hangzhou state with a support of 40% and a support of 60% at distance.
Assuming five users travel collectively during a short or long vacation, there are four alternatives for the travel location: a. the1Hangzhou, A2Suzhou, A3Chongming, A4The boat mountain; the main factors influencing the selection are: c. C1Distance, c2Activity item, c3Scene, c4Traffic convenience. Assuming that agreement is obtained by five users: the four factors are weighted by v ═ 0.3,0.4,0.1,0.2)T(ii) a Five users scored the four alternatives according to the above criteria as follows:
Figure BDA0002268038640000095
Figure BDA0002268038640000096
Figure BDA0002268038640000097
Figure BDA0002268038640000101
Figure BDA0002268038640000102
according to the user consistency degree calculation method in the step 1, the following steps can be obtained:
five user element level consistency:
Figure BDA0002268038640000103
Figure BDA0002268038640000104
Figure BDA0002268038640000105
five-user schema hierarchy consistency:
ACA1＝(0.778,0.859,0.813,0.822)ACA2＝(0.834,0.869,0.834,0.791)
ACA3＝(0.809,0.878,0.869,0.806)ACA4＝(0.803,0.866,0.853,0.653)
ACA5＝(0.713,0.816,0.806,0.816)
five-user matrix level consistency:
ACD1＝0.818 ACD2＝0.832 ACD3＝0.841 ACD4＝0.794 ACD5＝0.788
identifying inconsistent users in the population according to step 2 can:
APS＝{(4,4,1)；(4,4,2)；(5,1,1)；(5,1,2)；(5,1,3)}.
experiment 1: the traditional feedback mode is adopted to feed back the inconsistent individual preferences, the value of the feedback parameter is randomly selected to be 0.5, and the effect image is shown in figure 2.
Experiment 2: and (3) performing feedback by adopting a group recommendation method based on maximum harmony but not individuation, namely solving feedback parameters when the feedback parameters in the individualized model are equal, wherein an effect image is shown in figure 3.
Experiment 3: the personalized group recommendation method provided by the invention is adopted for feedback, the feedback parameters are solved according to the personalized group recommendation model based on the maximum harmony degree, and the effect image is shown in fig. 4.
The feedback parameters, harmony and consistency for the above three algorithms are shown in table 1:
δ4 δ4 ∑δi ACD4' ACD5' GHD
experiment 1 0.5 0.5 1 0.8329 0.8209 0.7958
Experiment 2 0.18 0.18 0.36 0.8081 0.8000 0.9253
Experiment 3 0.05 0.22 0.27 0.8000 0.8000 0.9537
The three groups of experiments show that the algorithm can well help the inconsistent users in the group recommendation system to meet the acceptable consistency requirement and realize the maximum harmony, so that the recommendation system is prevented from forcing the users to modify the preference, the individuation in the group recommendation system is realized, and the algorithm is more easily accepted by the users.
The feedback parameters obtained by solving in experiment 3 are further used for feeding back inconsistent users, so that:
(a) user 4 about A4At c1The following point suggests a correction of (0.69, 0.20);
(b) user 4 about A4At c2The following point suggests a correction of (0.23, 0.29);
(c) user 5 about A1At c1The following point suggests a correction of (0.69, 0.20);
(d) user 5 about A1At c2The following point suggests a correction of (0.46, 0.38);
(e) user 5 about A1At c3The following points suggest a correction of (0.17, 0.71).
Further, as described in step 5, the weights of the five users obtained according to the new consistency degree are:
ω＝(0.168,0.202,0.346,0.148,0.136)T；
aggregating individual user preferences:
Figure BDA0002268038640000121
solving each solution final score according to equation (9) is as follows:
TSA1＝0.3850TSA2＝0.4030TSA3＝0.3712TSA4＝0.6833
and finally, recommending a scheme 4 for five users of the collective tour group.
In summary, the invention discloses a personalized group recommendation method based on maximum harmony for inconsistent users. After the group user consistency is calculated, inconsistent users are identified, causes (sites or elements) causing the inconsistent users to be inconsistent are determined, then personalized feedback parameters are substituted, the harmony is maximum as an objective function, a nonlinear programming model is established by taking all users meeting a system consistency threshold as a constraint, feedback opinions to the inconsistent users are determined by solving the nonlinear programming model, so that the inconsistency elimination in a recommendation system is completed, finally, the group opinions are aggregated, the scores of all schemes are calculated and ranked, and the scheme with the highest score is recommended to the group users.
While the present invention has been described in detail with reference to the preferred embodiments, it should be understood that the above description should not be taken as limiting the invention. Various modifications and alterations to this invention will become apparent to those skilled in the art upon reading the foregoing description. Accordingly, the scope of the invention should be determined from the following claims.

Claims (7)

1. A personalized group recommendation method based on maximum harmony is characterized by comprising the following steps:
step 1, evaluating the consistency degree of group users, and measuring the consistency between individual users and group users from three levels of elements, schemes and matrixes;
step 2, identifying inconsistent users in the group from the three levels of individuals, schemes and elements;
step 3, constructing an individualized group recommendation model based on the maximum harmony and solving individualized group recommendation feedback parameters; the method comprises the following steps of controlling the feedback degree by utilizing an individualized feedback parameter with the maximum harmony of group users as an objective function, and using a user consistency boundary as a constraint;
step 4, calculating the personalized group recommendation feedback opinions;
step 5, determining the weight of the individual user according to the consistency degree based on the BUM function, and aggregating the preference of the individual user to determine a group recommendation scheme;
in the step 3, the personalized group recommendation model based on the maximum harmony degree is as follows:
Figure FDA0003532547700000011
wherein, deltahPersonalized feedback parameters for inconsistent users; m is the number of group users; h represents an inconsistent user sub-population of the user population, h being a proper subset of the set { 1., m }; n is the number of inconsistent users in the user group; β is a consistency threshold predefined by the group recommendation system; k represents one user in a user group, k 1.., m;
Figure FDA0003532547700000012
wherein p is the number of recommended schemes, and q is the number of group consideration criteria;
Figure FDA0003532547700000021
representing the scores of the alternative schemes i under the decision criterion j of the inconsistent users; i represents one of the alternatives, i 1.. ang., p; j represents one of the decision criteria, j 1.., q;
Figure FDA0003532547700000022
Figure FDA0003532547700000023
express an inconsistencyAverage distance sum between users and average viewpoints of other users than the users;
Figure FDA0003532547700000024
Figure FDA0003532547700000025
an evaluation element representing one of the inconsistent users;
Figure FDA0003532547700000026
an evaluation element representing any one of the m users;
#APSha number of inconsistent elements representing inconsistent users;
ACDkthe representation is the degree of common knowledge of any user among all users;
ACDhrepresenting the consensus degree of which any consensus degree in all users is smaller than a threshold value beta;
APS represents the set of all inconsistent elements.
2. The personalized population recommendation method based on maximum harmony of claim 1,
in step 1, when evaluating the consistency degree of the group users, the method further comprises:
and (3) evaluating the consistency of user element levels:
Figure FDA0003532547700000027
user scheme hierarchical consistency evaluation:
Figure FDA0003532547700000031
and (3) evaluating the consistency of the user matrix levels:
Figure FDA0003532547700000032
wherein,
s represents other users except the user to be measured in the user group; s ≠ k, 1., m;
Figure FDA0003532547700000033
representing the grade of the measured user on the alternative scheme i under the decision criterion j;
Figure FDA0003532547700000034
representing the scores of other users except the measured user k on the alternative scheme i under the decision criterion j;
Figure FDA0003532547700000035
representing the distance between the measured user k and the other users s;
Figure FDA0003532547700000036
then the process of the first step is carried out,
Figure FDA0003532547700000037
Figure FDA0003532547700000038
representing the trust degree of the user k to the scheme i under the evaluation criterion j;
Figure FDA0003532547700000039
representing the distrust degree of the user k to the scheme i under the evaluation criterion j;
Figure FDA00035325477000000310
representing the trust degree of the user s to the scheme i under the evaluation criterion j;
Figure FDA00035325477000000311
indicating the degree of distrust of the user s for the scheme i under the evaluation criterion j.
3. The personalized population recommendation method based on maximum harmony of claim 2,
in step 2, when inconsistent users in the group are identified, the method further comprises:
inconsistent individual user identification:
ECH＝{h|ACDh＜β} (4)
inconsistent user profile identification:
Figure FDA0003532547700000041
inconsistent user element identification:
Figure FDA0003532547700000042
wherein ECH ═ { h | ACDh< β is the set of all users with a degree of consensus less than a threshold.
4. The personalized population recommendation method based on maximum harmony of claim 3,
in step 4, calculating the personalized group recommendation feedback opinions:
Figure FDA0003532547700000043
wherein,
Figure FDA0003532547700000044
representing the scores of the inconsistent users after interaction with the alternative scheme i under the decision criterion j.
5. The personalized population recommendation method based on maximum harmony of claim 4,
in step 5, when determining the weight of the individual user according to the consistency degree based on the BUM function, utilizing a monotone increasing function
Figure FDA0003532547700000045
Then the individual weight
Figure FDA0003532547700000046
Wherein,
Figure FDA0003532547700000047
σ (k) is the rank value of ACD.
6. The personalized population recommendation method based on maximum harmony of claim 5,
according to
Figure FDA0003532547700000048
Aggregating individual user preferences;
Ria scoring matrix representing a user i, and CR represents a group user scoring matrix;
and aggregating the group schemes according to the weights and generating scheme ranking, and recommending the optimal scheme to the related groups.
7. The personalized group recommendation method based on maximum harmony as claimed in claim 6,
employing trust functions
Figure FDA0003532547700000051
To express user initial preferences; the trust score calculation method comprises the following steps:
Figure FDA0003532547700000052
after aggregating individual user preferences, solving the final score for each scheme by the above formula;
Figure FDA0003532547700000053
indicates the support degree of the user for the evaluation object,
Figure FDA0003532547700000054
indicating the degree of unsupportability of the user to the evaluation object.
CN201911094944.9A 2019-11-11 2019-11-11 Personalized group recommendation method based on maximum harmony Active CN110851716B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911094944.9A CN110851716B (en) 2019-11-11 2019-11-11 Personalized group recommendation method based on maximum harmony

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911094944.9A CN110851716B (en) 2019-11-11 2019-11-11 Personalized group recommendation method based on maximum harmony

Publications (2)

Publication Number Publication Date
CN110851716A CN110851716A (en) 2020-02-28
CN110851716B true CN110851716B (en) 2022-06-24

Family

ID=69601232

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911094944.9A Active CN110851716B (en) 2019-11-11 2019-11-11 Personalized group recommendation method based on maximum harmony

Country Status (1)

Country Link
CN (1) CN110851716B (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0699016A2 (en) * 1994-08-26 1996-02-28 Patent-Treuhand-Gesellschaft für elektrische Glühlampen mbH Circuit for operating low pressure discharge lamps
CN108647858A (en) * 2018-04-12 2018-10-12 华东师范大学 A kind of collaboration crowdsourcing method of quality control based on user's inconsistency information

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0699016A2 (en) * 1994-08-26 1996-02-28 Patent-Treuhand-Gesellschaft für elektrische Glühlampen mbH Circuit for operating low pressure discharge lamps
CN108647858A (en) * 2018-04-12 2018-10-12 华东师范大学 A kind of collaboration crowdsourcing method of quality control based on user's inconsistency information

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
A trust induced recommendation mechanism for reaching consensus in group decision making;Yujia Liu等;《ELSEVIER》;20171231;全文 *
基于特定群体兴趣的混合个性化推荐算法;毕行;《上海大学学报(自然科学版)》;20100630;第16卷(第3期);全文 *

Also Published As

Publication number Publication date
CN110851716A (en) 2020-02-28

Similar Documents

Publication Publication Date Title
CN110457589A (en) A kind of vehicle recommended method, device, equipment and storage medium
JP4956425B2 (en) Real-time selection of survey candidates
US20090287687A1 (en) System and method for recommending venues and events of interest to a user
US20130332264A1 (en) Method and system for determining touchpoint attribution
CN105809474B (en) Hierarchical commodity information filtering recommendation method
Bogaert et al. The added value of Facebook friends data in event attendance prediction
CN107423335B (en) Negative sample selection method for single-class collaborative filtering problem
CN108876069A (en) A kind of endowment service recommendation method
CN109636467A (en) A kind of comprehensive estimation method and system of the internet digital asset of brand
US11048768B1 (en) Social networking system with trading of electronic business cards
CN110334284B (en) Novel recommendation method integrating multi-factor decision making
US20190278819A1 (en) Systems and methods of providing recommendations of content items
US20200098031A1 (en) Product recommending apparatus and non-transitory computer readable medium
JP2018077615A (en) Advertising image generation device, advertising image generation method and program for advertising image generation device
CN109657962A (en) A kind of appraisal procedure and system of the volume assets of brand
CN102163304A (en) Method and system for collaborative networking with optimized inter-domain information quality assessment
US8843428B2 (en) Survey prioritization engine
CN109005431B (en) A kind of video evaluations recommender system
CN109190040B (en) Collaborative evolution-based personalized recommendation method and device
CN108846042B (en) Social network recommendation method combined with user feedback mechanism
Martínez-Torres et al. Identifying the features of reputable users in eWOM communities by using Particle Swarm Optimization
Jeong et al. User credit-based collaborative filtering
CN110851716B (en) Personalized group recommendation method based on maximum harmony
Wu et al. Modeling uncertainty driven curiosity for social recommendation
WO2017016403A1 (en) Method and apparatus for determining brand index information about service object

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