CN108830460B - Method for relieving data sparsity of recommendation system based on step-by-step dynamic filling - Google Patents
Method for relieving data sparsity of recommendation system based on step-by-step dynamic filling Download PDFInfo
- Publication number
- CN108830460B CN108830460B CN201810500434.6A CN201810500434A CN108830460B CN 108830460 B CN108830460 B CN 108830460B CN 201810500434 A CN201810500434 A CN 201810500434A CN 108830460 B CN108830460 B CN 108830460B
- Authority
- CN
- China
- Prior art keywords
- user
- merchant
- merchants
- scoring
- users
- 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
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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
Abstract
The invention discloses a method for relieving data sparsity of a recommendation system based on step-by-step dynamic filling. The method comprises the steps of firstly preprocessing user behavior data, and establishing a scoring matrix of a user-food merchant; secondly, selecting a target user to calculate user similarity, and selecting the user with the similarity larger than a threshold value alpha as a preselected similar neighbor user of the target user; then, calculating a common score difference mean value of the users, selecting a preselected neighbor user with the common score difference mean value smaller than a threshold value beta as a final similar neighbor user, and performing first-step filling on a score matrix by using the neighbor user; and finally, selecting similar merchants by using a similarity threshold method and a common score difference mean value for the remaining unfilled data in the scoring matrix, and filling the scoring matrix in the second step by using the similar merchants. The method and the device have the advantages that the problem of sparsity of the scoring matrix in the food merchant recommendation system is relieved to a certain extent, and the accuracy of the recommendation system is improved.
Description
Technical Field
The invention belongs to the technical field of food recommendation, and particularly relates to a method for relieving data sparsity of a recommendation system based on step-by-step dynamic filling.
Background
With the rapid development of internet technology, the explosive growth of data scale brings about a serious problem of information overload. How to quickly and effectively acquire valuable information from complex data becomes a key problem of the development of current big data. To solve this problem, personalized recommendation algorithms are widely studied, the most common of which is collaborative filtering algorithms. In the field of food merchant recommendation, due to the fact that the number of food merchants on a website is large, but only few users are willing to score the food merchants, a scoring data matrix of the user and the food merchants is extremely sparse, and a merchant recommendation list obtained based on a collaborative filtering algorithm is not accurate enough. Therefore, how to accurately pre-fill a sparse user-food merchant scoring matrix is the first problem to be solved by the food recommendation algorithm.
Disclosure of Invention
The present invention is directed to solving the above problems of the prior art. The method for improving the recommendation accuracy and relieving the recommendation system data sparsity based on step-by-step dynamic filling is provided. The technical scheme of the invention is as follows:
a method for relieving data sparsity of a recommendation system based on step-by-step dynamic filling comprises the following steps:
and 7, selecting the most similar merchant by adopting a similarity threshold method and a common score difference mean value for the remaining unfilled data in the scoring matrix, and filling the scoring matrix of the user-food merchant in the second step by utilizing the similar merchant set.
Further, the user behavior data preprocessing process in step 1 is as follows:
step 1.1: only selecting relevant information of food merchants and users thereof for obtaining user behavior data from the commenting website;
step 1.2: establishing a user-food merchant scoring matrix, (1) screening out information related to food merchants and food users from the total data; (2) this information is used to build a user-gourmet merchant rating matrix.
Further, the specific process of step 2 is as follows:
step 2.1: constructing a historical scoring record set of the users for each user; constructing a historical scoring record set of the merchants for each merchant;
step 2.2: and constructing a user set, counting the number of scoring merchants of each user, and sequencing the users in the user set according to the number of scoring merchants of the users from large to small.
Further, the process of constructing the pre-selected similar neighbor user set of the target user in step 4 is as follows:
step 4.1: according to the sequence of the users in the user set, one user is taken as a target user;
step 4.2: calculating the similarity between the other users and the target user by utilizing a user-food merchant scoring matrix and through a Pearson correlation coefficient formula; step 4.3: and selecting users with similarity greater than alpha to the target user to construct a preselected similar neighbor user set p _ N (u) of the target user.
Further, the Pearson correlation coefficient calculation formula is as follows:
wherein, simu,vRepresenting the degree of similarity of user u and user v, Iu,vFor a common scoring set of merchants, R, for user u and user vui、RviAre respectively usersu, user v's rating of merchant i,
the average scores of the user u and the user v are respectively.
Further, the selecting process of the final similar neighbor set in step 5 is as follows:
step 5.1: calculating the historical common score difference mean value of the target user and each selected and pre-similar neighbor user, wherein the calculation formula is as follows:
wherein avg (u, v) is the difference mean value, I 'of historical common scores of the user u and the user v'u,vCo-scoring a set of merchants, R, for the history of target user u and user vui、RviRespectively scoring the merchant i for the user u and the user v;
step 5.2: and selecting the preselected neighbor users with the common score difference mean value smaller than beta to construct a final similar neighbor user set N (u).
Further, the process of filling score data in the sparse user-food merchant score matrix by using the similar neighbor user set n (u) in step 6 is as follows:
step 6.1: filling unscored items of the target user by using a similar neighbor user set;
step 6.2: and (4) repeatedly selecting the next user in the user set as the target user, and filling the scoring matrix, namely repeatedly executing the step 4 to the step 6, wherein the scoring matrix of the user, namely the food merchant, for filling and calculating the similarity of the user each time is the matrix filled by the last user. And finishing the first step of filling until all the users in the user set are taken out.
Further, in step 7, the unfilled data left in the scoring matrix is filled in the scoring matrix by using a similar merchant set in a similar way, and the filling process is as follows:
step 7.1: constructing a merchant set I for the user-food merchant scoring matrix obtained after the first step of filling, counting the scoring times of each merchant, and sequencing the merchant set from large to small according to the scoring times of the merchants;
step 7.2: setting a threshold value gamma of the similarity degree of the merchants and a threshold value eta of a difference mean value of historical common scores of the merchants;
step 7.3: according to the sequence of merchants in the merchant set, one target merchant is selected, and the similarity between the other merchants and the target merchant is calculated by utilizing the Pearson correlation coefficient; constructing a preselected neighbor merchant set p _ N (i) by taking merchants with the similarity larger than gamma as preselected neighbor merchants of a target merchant;
step 7.4: calculating the common score difference mean value of the histories of the merchants and the target merchants in the preselected neighbor set, which are scored by the same user, wherein if the histories of the two merchants are not scored by the same user, the common score difference mean value of the merchants is + ∞; taking the merchants with the common score difference mean value smaller than eta as neighbor merchants of the target merchants, and constructing a neighbor merchant set N (i);
step 7.5: and (4) filling unscored data of the target merchant in the scoring matrix by using the neighbor merchant set, and repeatedly selecting the next merchant in the merchant set, namely repeatedly executing the steps 7.3-7.5, wherein the scoring matrix of the user-food merchant for similarity calculation and filling each time is the matrix after the last merchant is filled. And filling the user-food merchant scoring matrix until all merchants in the merchant set are obtained.
The invention has the following advantages and beneficial effects:
compared with the existing sparse matrix filling method, the sparse matrix filling method has the advantages and beneficial effects that: compared with the selection of similar neighbor sets in the traditional collaborative filtering algorithm, the method adds the historical common score difference mean value to screen the neighbors, removes the neighbors with larger score difference with the target item, enables the selection of the similar neighbor sets to be more accurate, effectively avoids the situation that a merchant which one user dislikes is recommended as a merchant which another user likes, and enables the recommendation to be more accurate. The invention adopts a step-by-step dynamic filling method, firstly performs the first filling from the perspective of a user, and then performs the second filling from the perspective of a merchant, so that the filling of a sparse user-food merchant scoring matrix is more complete. And simultaneously, each step of filling adopts a dynamic filling mode, the target users or the target merchants are sequentially selected according to the number of the scores for filling, and the matrix filled each time is the matrix filled by the last target. The dynamic filling increases the common scoring number of the user or the merchant, the similarity calculation is more accurate, so the filling of a sparse scoring matrix of the user, namely the food merchant is more accurate, the finally given recommendation list is more in line with the mind of the user, and the viscosity of the user to a recommendation system is improved.
Drawings
FIG. 1 is a table of the preferred embodiment of the present invention providing a catear merchant information;
FIG. 2 is a user information table
FIG. 3 is a user-gourmet merchant scoring matrix;
fig. 4 is an algorithm flow chart.
Detailed Description
The technical solutions in the embodiments of the present invention will be described in detail and clearly with reference to the accompanying drawings. The described embodiments are only some of the embodiments of the present invention.
The technical scheme for solving the technical problems is as follows:
example 1
On the basis of a traditional collaborative filtering recommendation algorithm based on users and projects, the embodiment of the invention takes the consideration of the historical common score difference mean value of users and merchants into consideration for the selection of similar neighbor sets, and finally fills the unscored data in the scoring matrix of the user-food merchant step by step and dynamically, wherein the method comprises the following steps:
101: preprocessing user behavior data, and establishing a scoring matrix of a user-food merchant;
102: building a historical scoring record set for each user and each merchant; constructing a user set, and sequencing the users in the user set from large to small according to the number of the user scoring merchants;
103: setting a user similarity threshold value alpha and a user history common score difference mean value threshold value beta;
104: according to the sequence of the users in the user set, a target user is selected, the similarity between the users and the target user is calculated, and a preselected similar neighbor user set of the target user is constructed;
105: calculating the historical common score difference mean value of the preselected similar neighbor user and the target user to construct a target
A final similar neighbor user set of users;
106: filling the scoring matrix of the user-food merchant by using the user similar neighbor set, and repeatedly selecting
Repeating steps 104-106 until the next user in the user set
All users are taken until the first step of filling is finished;
107: after the first step of filling is completed, the unfilled items left in the matrix are scored for the user-food merchant,
based on the traditional commodity-based collaborative filtering, a method similar to the steps is adopted for each time
The unscored items of each merchant are populated.
The embodiment of the invention firstly combines the steps 101-107 to fill the scoring matrix of the user-food merchant, thereby better relieving the data sparsity problem of the scoring matrix and improving the recommendation quality based on the traditional collaborative filtering algorithm.
Example 2
The protocol of example 1 is further described below in conjunction with specific calculation formulas and experimental data, as described in detail below:
201: for the user behavior data obtained from yelp of the American commenting website, only the relevant information of the food merchants and the users of the food merchants is selected, as shown in fig. 1 and 2;
a user-gourmet merchant rating matrix is established as shown in fig. 3.
202: constructing a historical scoring record set I (u) of the user for each user, and constructing a historical scoring record set U (i) of each merchant;
constructing a user set U, counting the number of scoring merchants of each user, and sequencing the users in the user set from large to small according to the number of scoring merchants of the users;
203: setting a user similarity threshold value alpha and a user common score difference mean value threshold value beta;
204: selecting a target project, calculating the score similarity of users, and constructing a preselected similar neighbor set of the target user;
(1) according to the sequence of the users in the user set, one user is taken as a target user;
(2) calculating the rest users and the target user u in the user set by adopting a Pearson correlation coefficient formula
The calculation formula of the score similarity is as follows:
wherein, simu,vRepresenting the degree of similarity of user u and user v, Iu,vFor a common scoring set of merchants, R, for user u and user vui、RviRespectively scoring the merchants i for the user u and the user v,
the average scores of the user u and the user v are respectively.
(3) Selecting users with similarity greater than a threshold value alpha with a target user as preselected similar neighbor users of the target user, and constructing a preselected similar neighbor user set p _ N (u) of the target user;
205: constructing a final similar neighbor user set of the target user:
(1) calculating the historical common score difference mean value of the users in the preselected similar neighbor set and the target user u, wherein the calculation formula is as follows:
wherein, I'u,vFor target users u and vHistorical set of co-scoring merchants, Rui、RviAnd respectively scoring the merchants i for the user u and the user v.
(2) And taking the users with the historical common score difference mean value smaller than a threshold beta in the preselected neighbor set and the target user as the final neighbor users of the target user, and constructing a final similar neighbor user set N (u) of the target user.
206, filling a scoring matrix of the user and the food trade company in the first step;
(1) calculating the scoring value of the target user u to the unscored merchant i by using the similar neighbor user set, and using the scoring value
Filling corresponding items in a scoring matrix of the user-food merchant, wherein the calculation formula is as follows:
(2) and after the filling of the score values of the unscored merchants of the user U is completed, repeatedly selecting the next user from the user set U as a target user, namely, repeatedly executing the steps 204-206, and realizing the dynamic filling of the user-food merchant score matrix according to the user set. And when all the users in the user score set U are taken, finishing the first step filling of the user-food merchant score matrix.
207: and performing second-step filling on the unfilled items left in the scoring matrix of the user-food merchant.
(1) Constructing a merchant set I for the user-food merchant scoring matrix obtained after the first step of filling, counting the scoring times of each merchant, and sequencing the merchant set I from large to small according to the scoring times of the merchants;
(2) setting a threshold value gamma of the similarity degree of the merchants and a threshold value eta of a difference mean value of historical common scores of the merchants;
(3) selecting a target merchant according to the order of the merchants in the merchant set, and calculating the similarity between the merchant and the target merchant i by using the Pearson correlation coefficient, wherein the calculation formula is as follows:
wherein U isi,jFor a set of users scoring both merchant i and merchant j, RuiRating of Merchant i for user u, RujFor the user u's rating of merchant j,
is the average score for the merchant i,
is the average score for merchant j.
And selecting merchants with similarity higher than gamma to the target merchant i as neighbor merchants of the target merchant i, and constructing a pre-selected neighbor merchant set p _ N (i).
(4) Constructing a final neighbor merchant set of the target merchant;
calculating the historical common score difference mean value of the merchants in the preselected neighbor merchant set and the target merchant, wherein the calculation formula is as follows:
wherein U'i,jRepresenting a user set which scores the merchant i and the merchant j simultaneously in the historical scoring records of the merchants; l U'i,jI represents the number of users scoring both the merchant i and the merchant j in the historical scoring record of the merchant; ruiScoring merchant i for user u; rujThe user u is scored for merchant j.
And taking the merchant with the common score difference mean value smaller than eta as the neighbor merchant of the target merchant, and constructing a final similar neighbor merchant set N (i) of the target merchant.
(5) Filling the scoring matrix of the user-food merchant in the second step;
calculating the scoring value of the user u on the unvalued item of the target merchant i by using the similar neighbor merchant set, and filling the corresponding item in the scoring matrix of the user-food merchant by using the value, wherein the calculation formula is as follows:
and (4) after filling the unevaluated items of the current target merchant I, repeatedly selecting the next merchant from the merchant set as the target merchant, namely repeatedly executing the steps (3), (4) and (5), and realizing dynamic filling of the scoring matrix of the user-food merchant according to the merchant set I. And when all the merchants in the merchant set I are fetched, the user-food merchant scoring matrix is filled. The flow chart is shown in fig. 4.
The above examples are to be construed as merely illustrative and not limitative of the remainder of the disclosure. After reading the description of the invention, the skilled person can make various changes or modifications to the invention, and these equivalent changes and modifications also fall into the scope of the invention defined by the claims.
Claims (6)
1. A method for relieving data sparsity of a recommendation system based on step-by-step dynamic filling is characterized by comprising the following steps:
step 1, preprocessing user behavior data, selecting information related to a food merchant and a food user, and obtaining a user-food merchant scoring matrix;
step 2, establishing a historical scoring record set for each user and each food merchant by using the established user-food merchant scoring matrix, establishing a user set at the same time, and sequencing the users in the user set from large to small according to the scoring merchant number of the users;
step 3, setting a user similarity threshold alpha and a user history common score difference mean value threshold beta;
step 4, selecting a target user according to the sequence of the users in the user set; calculating the similarity between the other users and the target user according to the user-food merchant scoring matrix; selecting users with similarity greater than alpha to the target user to construct a preselected similar neighbor user set of the target user;
step 5, calculating the historical common score difference mean value of the target user and each preselected similar neighbor user, wherein if two user histories do not have a common score merchant, the common score difference mean value of the two users is + ∞; selecting preselected neighbor users with the common score difference mean value smaller than beta to construct a final similar neighbor user set;
step 6, carrying out first-step filling on a scoring matrix of a user-food merchant by using a similar neighbor user set of a target user;
step 7, selecting the most similar merchants by adopting a similarity threshold method and a common score difference mean value for the remaining unfilled data in the scoring matrix, and performing second-step filling on the scoring matrix of the user-food merchant by using a similar merchant set;
the process of filling score data in the sparse user-food merchant score matrix by using the similar neighbor user set N (u) in the step 6 is as follows:
step 6.1: filling unscored items of the target user by using a similar neighbor user set;
step 6.2: repeatedly selecting the next user in the user set as a target user, and filling the scoring matrix, namely repeatedly executing the steps 4-6, wherein the scoring matrix of the user, namely the food merchant, for filling and calculating the similarity of the user each time is the matrix filled by the last user; the first step of filling is completed until all users in the user set are taken out;
and 7, filling the score matrix of the user-food merchant by using a similar method and a similar merchant set for the left unfilled data in the score matrix, wherein the filling process is as follows:
step 7.1: constructing a merchant set I for the user-food merchant scoring matrix obtained after the first step of filling, counting the scoring times of each merchant, and sequencing the merchant set from large to small according to the scoring times of the merchants;
step 7.2: setting a threshold value gamma of the similarity degree of the merchants and a threshold value eta of a difference mean value of historical common scores of the merchants;
step 7.3: according to the sequence of merchants in the merchant set, one target merchant is selected, and the similarity between the other merchants and the target merchant is calculated by utilizing the Pearson correlation coefficient; constructing a preselected neighbor merchant set p _ N (i) by taking merchants with the similarity larger than gamma as preselected neighbor merchants of a target merchant;
step 7.4: calculating the common score difference mean value of the histories of the merchants and the target merchants in the preselected neighbor set, which are scored by the same user, wherein if the histories of the two merchants are not scored by the same user, the common score difference mean value of the merchants is + ∞; taking the merchants with the common score difference mean value smaller than eta as neighbor merchants of the target merchants, and constructing a neighbor merchant set N (i);
step 7.5: filling unscored data of the target merchant in the scoring matrix by using a neighbor merchant set, and repeatedly selecting the next merchant in the merchant set, namely repeatedly executing the steps 7.3-7.5, wherein the scoring matrix of the user-food merchant for similarity calculation and filling each time is the matrix after the previous merchant is filled; and filling the user-food merchant scoring matrix until all merchants in the merchant set are obtained.
2. The method for mitigating recommendation system data sparsity based on step-by-step dynamic population as claimed in claim 1, wherein the user behavior data preprocessing procedure in step 1 is as follows:
step 1.1: only selecting relevant information of food merchants and users thereof for obtaining user behavior data from the commenting website;
step 1.2: establishing a user-food merchant scoring matrix, (1) screening out information related to food merchants and food users from the total data; (2) this information is used to build a user-gourmet merchant rating matrix.
3. The method for mitigating recommendation system data sparsity based on step-by-step dynamic population as claimed in claim 1, wherein the specific process of step 2 is as follows:
step 2.1: constructing a historical scoring record set of the users for each user; constructing a historical scoring record set of the merchants for each merchant;
step 2.2: and constructing a user set, counting the number of scoring merchants of each user, and sequencing the users in the user set according to the number of scoring merchants of the users from large to small.
4. The method for mitigating recommendation system data sparsity based on step-by-step dynamic population as claimed in claim 1, wherein the process of constructing the pre-selected similar neighbor user set of the target user in step 4 is as follows:
step 4.1: according to the sequence of the users in the user set, one user is taken as a target user;
step 4.2: calculating the similarity between the other users and the target user by utilizing a user-food merchant scoring matrix and through a Pearson correlation coefficient formula; step 4.3: and selecting users with similarity greater than alpha to the target user to construct a preselected similar neighbor user set p _ N (u) of the target user.
5. The method for mitigating recommendation system data sparsity based on step-wise dynamic population according to claim 4, wherein said Pearson correlation coefficient calculation formula is as follows:
wherein, simu,vRepresenting the degree of similarity of user u and user v, Iu,vFor a common scoring set of merchants, R, for user u and user vui、RviRespectively scoring the merchants i for the user u and the user v,
the average scores of the user u and the user v are respectively.
6. The method for mitigating recommendation system data sparsity based on step-by-step dynamic population as claimed in claim 1, wherein the selection process of the final similar neighbor set in step 5 is as follows:
step 5.1: calculating the historical common score difference mean value of the target user and each selected and pre-similar neighbor user, wherein the calculation formula is as follows:
wherein avg (u, v) is the difference mean value, I 'of historical common scores of the user u and the user v'u,vCo-scoring a set of merchants, R, for the history of target user u and user vui、RviRespectively scoring the merchant i for the user u and the user v;
step 5.2: and selecting the preselected neighbor users with the common score difference mean value smaller than beta to construct a final similar neighbor user set N (u).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810500434.6A CN108830460B (en) | 2018-05-23 | 2018-05-23 | Method for relieving data sparsity of recommendation system based on step-by-step dynamic filling |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810500434.6A CN108830460B (en) | 2018-05-23 | 2018-05-23 | Method for relieving data sparsity of recommendation system based on step-by-step dynamic filling |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108830460A CN108830460A (en) | 2018-11-16 |
| CN108830460B true CN108830460B (en) | 2021-12-17 |
Family
ID=64148370
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810500434.6A Active CN108830460B (en) | 2018-05-23 | 2018-05-23 | Method for relieving data sparsity of recommendation system based on step-by-step dynamic filling |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108830460B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109684549A (en) * | 2018-12-24 | 2019-04-26 | 拉扎斯网络科技（上海）有限公司 | Target data prediction technique, device, electronic equipment and computer storage medium |
| CN116596640A (en) * | 2023-07-19 | 2023-08-15 | 国网山东省电力公司营销服务中心(计量中心) | Recommendation method, system, equipment and storage medium for electric power retail electric charge package |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101329683A (en) * | 2008-07-25 | 2008-12-24 | 华为技术有限公司 | Recommendation system and method |
| CN102495864A (en) * | 2011-11-25 | 2012-06-13 | 清华大学 | Collaborative filtering recommending method and system based on grading |
| CN103971161A (en) * | 2014-05-09 | 2014-08-06 | 哈尔滨工程大学 | Hybrid recommendation method based on Cauchy distribution quantum-behaved particle swarm optimization |
| CN105354330A (en) * | 2015-11-27 | 2016-02-24 | 南京邮电大学 | Sparse data preprocessing based collaborative filtering recommendation method |
| CN105976229A (en) * | 2016-05-13 | 2016-09-28 | 云南大学 | Collaborative filtering algorithm based on user and project mixing |
| CN107391670A (en) * | 2017-07-21 | 2017-11-24 | 云南电网有限责任公司教育培训评价中心 | A kind of mixing recommendation method for merging collaborative filtering and user property filtering |
| CN107833070A (en) * | 2017-10-24 | 2018-03-23 | 重庆邮电大学 | Alleviate the intersection dynamic fill method of Sparse sex chromosome mosaicism in a kind of commending system |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8484226B2 (en) * | 2011-07-09 | 2013-07-09 | Yahoo! Inc. | Media recommendations for a social-software website |
| CN106484876A (en) * | 2016-10-13 | 2017-03-08 | 中山大学 | A kind of based on typical degree and the collaborative filtering recommending method of trust network |
-
2018
- 2018-05-23 CN CN201810500434.6A patent/CN108830460B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101329683A (en) * | 2008-07-25 | 2008-12-24 | 华为技术有限公司 | Recommendation system and method |
| CN102495864A (en) * | 2011-11-25 | 2012-06-13 | 清华大学 | Collaborative filtering recommending method and system based on grading |
| CN103971161A (en) * | 2014-05-09 | 2014-08-06 | 哈尔滨工程大学 | Hybrid recommendation method based on Cauchy distribution quantum-behaved particle swarm optimization |
| CN105354330A (en) * | 2015-11-27 | 2016-02-24 | 南京邮电大学 | Sparse data preprocessing based collaborative filtering recommendation method |
| CN105976229A (en) * | 2016-05-13 | 2016-09-28 | 云南大学 | Collaborative filtering algorithm based on user and project mixing |
| CN107391670A (en) * | 2017-07-21 | 2017-11-24 | 云南电网有限责任公司教育培训评价中心 | A kind of mixing recommendation method for merging collaborative filtering and user property filtering |
| CN107833070A (en) * | 2017-10-24 | 2018-03-23 | 重庆邮电大学 | Alleviate the intersection dynamic fill method of Sparse sex chromosome mosaicism in a kind of commending system |
Non-Patent Citations (1)
| Title |
|---|
| 代金龙." 协同过滤算法中数据稀疏性问题研究".《中国优秀硕士论文全文数据库 信息科技辑》.2014,I138-1110. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108830460A (en) | 2018-11-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3819821B1 (en) | User feature generating method, device, and apparatus, and computer-readable storage medium | |
| CN108510402A (en) | Insurance kind information recommendation method, device, computer equipment and storage medium | |
| US20230043174A1 (en) | Method for pushing anchor information, computer device, and storage medium | |
| JP4906846B2 (en) | Scoring user compatibility in social networks | |
| US20180308152A1 (en) | Data Processing Method and Apparatus | |
| CN111651704B (en) | Content recommendation method, computing device and storage medium | |
| CN108460082B (en) | Recommendation method and device and electronic equipment | |
| CN108876537B (en) | Mixed recommendation method for online marketplace system | |
| CN104462383A (en) | Movie recommendation method based on feedback of users' various behaviors | |
| CN110532351B (en) | Recommendation word display method, device and equipment and computer readable storage medium | |
| CN108509407A (en) | Text semantic similarity calculating method, device and user terminal | |
| CN104615741B (en) | Cold-start project recommendation method and device based on cloud computing | |
| CN110516057B (en) | Petition question answering method and device | |
| CN110796313A (en) | Session recommendation method based on weighted graph volume and item attraction model | |
| CN108830460B (en) | Method for relieving data sparsity of recommendation system based on step-by-step dynamic filling | |
| CN111324807A (en) | Collaborative filtering recommendation method based on trust degree | |
| CN110263973B (en) | Method and device for predicting user behavior | |
| CN116362823A (en) | Recommendation model training method, recommendation method and recommendation device for behavior sparse scene | |
| CN108875071B (en) | Learning resource recommendation method based on multi-view interest | |
| CN109993026A (en) | The training method and device of relatives' identification network model | |
| CN112612948B (en) | Deep reinforcement learning-based recommendation system construction method | |
| CN116662673B (en) | User preference data analysis method based on data monitoring | |
| CN106651564B (en) | Target client acquisition method and system | |
| CN109086386A (en) | Data processing method, device, computer equipment and storage medium | |
| CN116935057A (en) | Target evaluation method, electronic device, and computer-readable storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |