CN109829792A - A kind of personalized item recommendation method and system based on user location distribution - Google Patents
A kind of personalized item recommendation method and system based on user location distribution Download PDFInfo
- Publication number
- CN109829792A CN109829792A CN201910126341.6A CN201910126341A CN109829792A CN 109829792 A CN109829792 A CN 109829792A CN 201910126341 A CN201910126341 A CN 201910126341A CN 109829792 A CN109829792 A CN 109829792A
- Authority
- CN
- China
- Prior art keywords
- article
- user
- similarity
- scoring
- recommended
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 239000011159 matrix material Substances 0.000 claims abstract description 55
- 238000004364 calculation method Methods 0.000 claims abstract description 10
- 238000007781 pre-processing Methods 0.000 claims abstract description 4
- 238000011156 evaluation Methods 0.000 claims description 14
- 238000010606 normalization Methods 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 description 4
- 238000001914 filtration Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000007418 data mining Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The personalized item recommendation method based on user location distribution that the invention discloses a kind of, comprising: user's rating matrix of building user and article, and obtain the geographical location of user；Distance coefficient between all items is calculated according to the geographical location of user's rating matrix and user, generates distance coefficient matrix；The similarity of all items is calculated according to distance coefficient matrix, generates similarity matrix；Scoring according to the scored article of the similarity matrix of the article of generation, each user, to each user in predicting article to be recommended；The highest p article of score in predicting in article to be recommended that each user generates is recommended into corresponding each user.The present invention also provides a kind of recommender systems, comprising: distance coefficient computing module, article similarity calculation module, score in predicting module and recommending module between data preprocessing module, article.The recommended method and system may be implemented more accurate score in predicting and calculate, and then realize more convenient and efficient recommendation service.
Description
Technical field
It is especially a kind of geographical based on user the present invention relates to big data and data mining recommender system technical field is based on
The personalized item recommendation method and system of position.
Background technique
The fast development of current information technology and e-commerce not only brings to the production of the mankind and life revolutionary
Variation, and product class extremely abundant and consumer channel are brought for consumer.However it is limited in information retrieval capability
In the case of, user is difficult therefrom to screen and find oneself really interested information.In order to solve this information overload problem, electricity
Sub- business web site generallys use the mode of keyword search and classified navigation to improve the recall precision of user.Though both means
So reduce the screening range of user, but need user can the demand according to the knowledge and experience of itself to oneself have more
It accurately describes and portrays.However the renewal speed of modern new product and technology often surmounts the update of personal knowledge, makes user
It is difficult to describe most suitable product and demand with experience.Therefore, it can quickly help to use by personalized recommendation algorithm
Find oneself interested commodity in family.
As most successful recommended technology is applied in Technologies of Recommendation System in E-Commerce, traditional collaborative filtering based on article is pushed away
Algorithm is recommended there are some problems, relies solely on the diversity of values between commodity to measure similarity, causes to recommend precision lower.
The Chinese patent literature of Publication No. CN109063120A discloses a kind of collaborative filtering recommending side based on cluster
Method and device, comprising: obtain the label genome vector of the first article；Label genome vector based on the first article, by
One article is divided into the cluster class of the first quantity；For each target item: when the target item and the second article belong to same cluster class,
Relative coefficient is calculated at a distance from preset kind based on the target item between the second article；When the target item and second
When article belongs to different cluster classes, relative coefficient is calculated based on the Poisson related coefficient of the target item and the second article；By mesh
Mark user, which is weighted the relative coefficient of the preset scoring of the second article and the target item and the second article, to be asked
With obtain target user and score the prediction of the target item；The target item that prediction scoring is met to preset condition, is recommended
Target user.The Chinese patent literature of Publication No. CN104699687A discloses a kind of item recommendation method and server, institute
Stating item recommendation method includes: to obtain the history use information of article to be recommended, the history use information of the article to be recommended
History including article to be recommended uses temporal information using the history of location information and/or article to be recommended；It obtains to be recommended
The current information of the terminal of user, the current information of the terminal of the user to be recommended include the current of the terminal of user to be recommended
Location information and/or current time；According to the history use information of the article to be recommended and the terminal of the user to be recommended
Current information, determine article recommendation list；The article recommendation list is sent to the terminal of the user to be recommended.
But above patent document does not consider incidence relation existing for the similarity degree between article and the spatial position of user,
Cause to recommend precision lower, therefore, how to realize that more convenient and efficient recommendation service is that current e-commerce is still being explored
The problem of.
Summary of the invention
The purpose of the present invention is to provide a kind of personalized item recommendation method and system based on user location distribution, can
To realize that more accurate score in predicting calculates, and then realize more convenient and efficient recommendation service.
The invention provides the following technical scheme:
A kind of personalized item recommendation method based on user location distribution, comprising the following steps:
(1) active user is traversed, each user is obtained and the scoring for the article for having evaluation behavior is recorded, construct user
With user's rating matrix of article, and the geographical location of user is obtained；
(2) distance coefficient between all items is calculated according to the geographical location of user's rating matrix and user, generated any
The distance coefficient matrix of the distance between article coefficient composition two-by-two；
(3) similarity that all items are calculated according to distance coefficient matrix generates the similarity composition of any article two-by-two
Similarity matrix；
(4) article scored according to the similarity matrix of the article of generation, each user, waits for each user in predicting
Recommend the scoring of article, the article to be recommended is the article that each user did not had evaluation behavior also；
(5) the highest p article of score in predicting in article to be recommended that each user generates is recommended corresponding every
A user.
In the present invention, the article is the set of all items present in current recommender system, for each user
For, this article set includes that user had the article of evaluation behavior and do not had the article of evaluation behavior, did not had and comments
The article of valence behavior is the article to be recommended of this user, carries out recommending user to reselection after prediction scoring.
In step (1), the geographical location for obtaining user includes: by the postcode of user or place city or GPS
Location information obtains the geographical location of user, generates the latitude and longitude coordinates of user.
In step (2), the method for calculating distance coefficient between all items are as follows:
(2-1) obtains any two object for any two the article i and j in all items, according to user's rating matrix
The co-user of product i and j generate while there are user's set N1 of scoring behavior to two article i and j；
(2-2) generates all user's combination of two in user's set N1A unduplicated user pair set two-by-two
N2；
(2-3) calculates the distance d of every a pair of of user in user pair set N2 according to the geographical location of user, willA user
Pair distance sum to obtain distance and D, using D as the distance coefficient of article i and j.
In step (3), the method for the similarity for calculating all items are as follows:
For any two article i and j, calculating formula of similarity are as follows:
Wherein, N1 is that article i and article j, there are the user of scoring behavior set, R simultaneouslyu,i,Ru,jIt is user u to object
Product i, the scoring of article j,Indicate the average score of article i and article j, Mi,jIndicate the distance system of article i and article j
Number, α are the adjusting parameter of distance coefficient.
Customized parameter α ∈ (0,1) is adjusted according to the actual recommendation effect of recommender system.
Preferably, the similarity of article is normalized:
Wherein, simmax,simminFor the maximum value and minimum value in the similarity value of article, ei,jFor article i after normalization
With the similarity of article j, ei,j∈(0,1).Similarity after normalization narrows down to original very big numerical value between 0 and 1,
It is more convenient subsequent calculating.
In step (4), the method for the scoring to each user in predicting article to be recommended are as follows:
Wherein, Nu(i) scored and score similar article with article i for user u.
As can be seen from the above equation, user u scores by user N to the prediction of article iu(i) scoring of article carries out in
Weighted mean operation obtains.
In the present invention, by adjustment parameter α, prediction scoring under available difference α, by calculating under different α
The mean square error or/and mean absolute error prediction scoring and actually scored select mean square error or/and mean absolute error most
The small prediction scoring being worth under corresponding adjustment parameter α is that last prediction is scored.
The present invention also provides a kind of recommender systems using above-mentioned recommended method, comprising:
Data preprocessing module obtains each user to the article for having evaluation behavior for traversing all users
Scoring record, constructs user's rating matrix of user and article in current system；For acquire the postcode of user, place city or
GPS positioning information obtains the latitude and longitude coordinates of user；
Distance coefficient computing module between article, according to the latitude and longitude coordinates of user's rating matrix and user calculate article it
Between distance coefficient, generate the distance coefficient matrix of any composition of the distance between article coefficient two-by-two；
Article similarity calculation module calculates the similarity of all items according to distance coefficient matrix, generates arbitrarily two-by-two
The similarity matrix of the similarity composition of article；
Score in predicting module waits pushing away according to the scored article of similarity matrix, each user to each user in predicting
The scoring of article is recommended, the article to be recommended is the article that each user did not had evaluation behavior also；
Recommending module recommends the highest p article of score in predicting in article to be recommended that each user generates relatively
The each user answered.
The method that distance coefficient computing module calculates distance coefficient between all items between the article are as follows:
For any two the article i and j in all items, any two article i and j is obtained according to user's rating matrix
Co-user, generate simultaneously to two article i and j, there are user's set N1 of scoring behavior；
By all user's combination of two in user's set N1, generateA unduplicated user pair set N2 two-by-two；
The distance d that every a pair of of user in user pair set N2 is calculated according to the latitude and longitude coordinates of user, willA user couple
Distance sum to obtain distance and D, using D as the distance coefficient of article i and j.
The method that the article similarity calculation module calculates the similarity of all items are as follows:
For any two article i and j, calculating formula of similarity are as follows:
Wherein, N1 is that article i and article j, there are the user of scoring behavior set, R simultaneouslyu,i,Ru,jIt is user u to object
Product i, the scoring of article j,Indicate the average score of article i and article j, Mi,jIndicate the distance system of article i and article j
Number, α are the adjusting parameter of distance coefficient.
The similarity of article is normalized in the article similarity calculation module:
Wherein, simmax,simminFor the maximum value and minimum value in the similarity value of article, ei,jFor article i after normalization
With the similarity of article j, ei,j∈(0,1)。
Method of the score in predicting module to the scoring of each user in predicting article to be recommended are as follows:
Wherein, Nu(i) scored and score similar article with article i for user u.
In the present invention, since there are certain to be associated with the spatial position of user for the similarity degree between article, pass through
The connection between recommender system commodity, film and other items is investigated comprehensively and by the space for the user group for generating behavior to article
Geographical location calculates the distance between article coefficient to improve the collaborative filtering method based on article, utilizes being total between article
The similitude that article is preferably measured with the spatial behavior information and rule of user, to realize more accurate score in predicting
Calculating and recommendation service.
Detailed description of the invention
Fig. 1 is the structural schematic diagram of recommender system provided by the invention；
Fig. 2 is the flow diagram that distance coefficient between article is calculated in the present invention.
Specific embodiment
The personalized item recommendation method based on user location distribution that the embodiment of the invention provides a kind of realizes user couple
The prediction of article is scored.
In order to make the invention's purpose, features and advantages of the invention more obvious and easy to understand, below in conjunction with the present invention
Attached drawing in embodiment, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that disclosed below
Embodiment be only a part of the embodiments of the present invention, and not all embodiments.Based on the embodiments of the present invention, this field
Technical staff's every other embodiment obtained, shall fall within the protection scope of the present invention.It is described in detail below:
As shown in Figure 1, the present invention provides a kind of personalized article recommender system based on user location distribution, can be applied to
E-commerce field, recommender system provided by the invention include:
Data preprocessing module obtains each user to the article for having evaluation behavior for traversing all users
Scoring record, constructs user's rating matrix of user and article in current system；For acquire the postcode of user, place city or
GPS positioning information obtains the latitude and longitude coordinates of user；
Distance coefficient computing module between article, according to the latitude and longitude coordinates of user's rating matrix and user calculate article it
Between distance coefficient, generate the distance coefficient matrix of any composition of the distance between article coefficient two-by-two；
Article similarity calculation module calculates the similarity of all items according to distance coefficient matrix, generates arbitrarily two-by-two
The similarity matrix of the similarity composition of article；
Score in predicting module waits pushing away according to the scored article of similarity matrix, each user to each user in predicting
The scoring of article is recommended, the article to be recommended is the article that each user did not had evaluation behavior also；
Recommending module recommends the highest p article of score in predicting in article to be recommended that each user generates relatively
The each user answered.
Method that recommender system provided by the invention recommends article the following steps are included:
Step 1: traversal active user, obtains each user and records to the scoring for the article for having evaluation behavior, building
User's rating matrix of user and article, and obtain the geographical location of user.
In the present embodiment, it by a series of data in geographical locations that can extract user such as the postcode of user, obtains
The latitude and longitude coordinates of user.
Specifically, it such as acquires 4 users to record the scoring of 5 articles, the rating matrix of building is as shown in table 1.
1 user's rating matrix of table
Wherein, default scoring is article scoring to be predicted in table 1.
Specifically, the latitude and longitude coordinates of user are as shown in table 2.
The latitude and longitude coordinates of 2 user of table
| User | Latitude | Longitude |
| User 1 | 32.21 | -110.88 |
| User 2 | 37.41 | -122.05 |
| User 3 | 30.16 | -81.7 |
| User 4 | 41.57 | -83.65 |
Step 2: calculating distance coefficient between all items according to the geographical location of user's rating matrix and user, generate
The arbitrarily distance coefficient matrix that the distance between article coefficient forms two-by-two.
It is as shown in Figure 2:
For any two article i and j, the co-user of article i and j are obtained according to the rating matrix of user, that is, obtain
To the two articles, there are user's set N1 of scoring behavior simultaneously.
By all user's combination of two in user's set N, generateA unduplicated user pair set N2 two-by-two.
The distance d that every a pair of of user in user pair set N2 is calculated according to the latitude and longitude coordinates of user, by all users couple
Distance sum to obtain distance and D, set D to the distance coefficient of article i and j.
By the above-mentioned means, finally generating the distance between article and article coefficient matrix M.
In the present embodiment, it is as shown in table 3 to put to the proof M for the distance coefficient of generation.
3 distance coefficient matrix of table
Step 3: calculating the similarity of all items according to distance coefficient matrix, the similarity of any article two-by-two is generated
The similarity matrix of composition.
For any two article i and j, calculating formula of similarity are as follows:
Wherein N1 is that there are the user of scoring behavior set, R to article i and article ju,i,Ru,jIt is user u to article i, object
The scoring of product j,Indicate the average score of article i and article j, Mi,jIndicate that the distance coefficient of article i and article j, α are
The adjustable parameters of distance coefficient.
Then similarity is normalized:
simmax,simminIt is article to the maximum value and minimum value in similarity value, ei,jFor article i and object after normalization
The similarity of product j.Finally obtain the similarity matrix of article.
By taking adjustable parameters α=0.5 as an example, the similarity matrix between the article of calculating is as shown in table 4.
The similarity matrix of 4 article of table
Step 4: the article scored according to the similarity matrix of the article of generation, each user, pre- to each user
The scoring of article to be recommended is surveyed, the article to be recommended is the article that each user did not had evaluation behavior also.
After having obtained the similarity matrix of article, scoring of the user to article is calculated according to following predictor formula:
Nu(i) scored and score similar article with article i for user u.User u scores to the prediction of article i logical
It crosses to user Nu(i) scoring of article is weighted and averaged operation and obtains in.
Specifically, such as user 1, user 1 scores to article 1,2,4,5, but user 1 is not to article
3 score, so we carry out prediction calculating to this scoring.The similarity of article 3 and article 1,2,4,5 is respectively
0.994412119,0.41386635,0.226549112,
0.28630266.User is respectively 5,1,2,2 to the scoring of article 1,2,4,5, so calculating user 1 to article
3 predicted value is 1.6.Similarly obtain prediction scoring 2.35 of the user 3 to article 2, prediction scoring 1.34 of the user 4 to article 5.
By adjustment parameter α, the prediction under available difference α is scored, and is scored by the prediction calculated under different α and real
The mean square error or/and mean absolute error of border scoring, select mean square error or/and the corresponding tune of mean absolute error minimum value
Save the prediction scoring that the prediction under parameter alpha is scored the most last.
Step 5: the highest p article of score in predicting in article to be recommended that each user generates is recommended corresponding
Each user.
Preceding p article is recommended corresponding every according to the setting of current recommender system according to the height of prediction scoring
A user, first p according to system actual setting.
In order to test the performance of recommended method and system provided by the invention, the present invention uses RMSE (root-mean-square error),
MAE (mean absolute error) two indices test the precision performance of recommended method provided by the invention.Evidence show RMSE
Being promoted by a small margin for value will have a huge impact the top-K quality recommended.
Absolute error is the difference of predicted value (mean value of Individual forecast value or multiple predicted value) and practical true value.(MAE)
Square error refers to the average value of absolute error, can preferably reflect the actual conditions of predicted value error, and the value of MAE is smaller, says
Bright prediction model has better accuracy.(RMSE) mean absolute error increases the consumer articles scoring to forecasting inaccuracy
It punishes (square), as MAE, the value of RMSE is smaller, and the accuracy of prediction model is better.
The calculation formula of MAE:
The calculation formula of RMSE:
pi,qiRespectively practical scoring and prediction scoring, N are scoring quantity number.
In the present embodiment, if user 1 is 2 to the practical scoring of article 3, user 3 is 3 to the practical scoring of article 2, is used
Family 4 is 1 to the practical scoring of article 5.Then MAE value is 0.46, RMSE value 0.48.
By adjustment parameter α, the prediction under available difference α is scored, and then obtains the scoring of the prediction under different α and reality
The mean square error or/and mean absolute error of border scoring, select mean square error or/and the corresponding tune of mean absolute error minimum value
Saving the prediction scoring under parameter alpha is that last prediction is scored.
In conclusion the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although referring to upper
Stating embodiment, invention is explained in detail, those skilled in the art should understand that: it still can be to upper
Technical solution documented by each embodiment is stated to modify or equivalent replacement of some of the technical features；And these
It modifies or replaces, the spirit and scope for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution.
Claims (10)
1. a kind of personalized item recommendation method based on user location distribution, comprising the following steps:
(1) active user is traversed, each user is obtained and the scoring for the article for having evaluation behavior is recorded, construct user and object
User's rating matrix of product, and obtain the geographical location of user；
(2) distance coefficient between all items is calculated according to the geographical location of user's rating matrix and user, generated arbitrarily two-by-two
The distance coefficient matrix of the distance between article coefficient composition；
(3) similarity that all items are calculated according to distance coefficient matrix generates the phase of the similarity composition of any article two-by-two
Like degree matrix；
(4) article scored according to the similarity matrix of the article of generation, each user, it is to be recommended to each user in predicting
The scoring of article, the article to be recommended are the article that each user did not had evaluation behavior also；
(5) the highest p article of score in predicting in article to be recommended that each user generates is recommended into corresponding each use
Family.
2. the personalized item recommendation method according to claim 1 based on user location distribution, which is characterized in that in step
Suddenly in (2), the method for calculating distance coefficient between all items are as follows:
(2-1) for any two the article i and j in all items, according to user's rating matrix obtain any two article i and
The co-user of j generates while there are user's set N1 of scoring behavior to two article i and j；
(2-2) generates all user's combination of two in user's set N1A unduplicated user pair set N2 two-by-two；
(2-3) calculates the distance d of every a pair of of user in user pair set N2 according to the geographical location of user, willA user's couple
Distance summation obtains distance and D, using D as the distance coefficient of article i and j.
3. the personalized item recommendation method according to claim 1 based on user location distribution, which is characterized in that in step
Suddenly in (3), the method for the similarity for calculating all items are as follows:
For any two article i and j, calculating formula of similarity are as follows:
Wherein, N1 is that article i and article j, there are the user of scoring behavior set, R simultaneouslyu,i,Ru,jIt is user u to article i,
The scoring of article j,Indicate the average score of article i and article j, Mi,jIndicate the distance coefficient of article i and article j, α
For the adjusting parameter of distance coefficient.
4. the personalized item recommendation method according to claim 3 based on user location distribution, which is characterized in that object
The similarity of product is normalized:
Wherein, simmax,simminFor the maximum value and minimum value in the similarity value of article, ei,jFor article i and object after normalization
The similarity of product j, ei,j∈(0,1)。
5. the personalized item recommendation method according to claim 1 based on user location distribution, which is characterized in that in step
Suddenly in (4), the method for the scoring to each user in predicting article to be recommended are as follows:
Wherein, Nu(i) scored and score similar article with article i for user u.
6. a kind of personalized article recommender system based on user geographical location characterized by comprising
Data preprocessing module obtains scoring of each user to the article for having evaluation behavior for traversing all users
Record constructs user's rating matrix of user and article in current system；For acquiring postcode, place city or the GPS of user
Location information obtains the latitude and longitude coordinates of user；
Distance coefficient computing module between article calculates the spacing of article according to the latitude and longitude coordinates of user's rating matrix and user
From coefficient, the distance coefficient matrix of any composition of the distance between article coefficient two-by-two is generated；
Article similarity calculation module calculates the similarity of all items according to distance coefficient matrix, generates any article two-by-two
Similarity composition similarity matrix；
Score in predicting module, according to the scored article of similarity matrix, each user, to each user in predicting object to be recommended
The scoring of product, the article to be recommended are the article that each user did not had evaluation behavior also；
Recommending module is recommended the highest p article of score in predicting in article to be recommended that each user generates corresponding
Each user.
7. the personalized article recommender system according to claim 6 based on user geographical location, which is characterized in that described
The method that distance coefficient computing module calculates distance coefficient between all items between article are as follows:
For any two the article i and j in all items, being total to for any two article i and j is obtained according to user's rating matrix
Same user generates while there are user's set N1 of scoring behavior to two article i and j；
By all user's combination of two in user's set N1, generateA unduplicated user pair set N2 two-by-two；
The distance d that every a pair of of user in user pair set N2 is calculated according to the latitude and longitude coordinates of user, willA user couple away from
From summation obtain with a distance from and D, using D as the distance coefficient of article i and j.
8. the personalized article recommender system according to claim 6 based on user geographical location, which is characterized in that described
The method that article similarity calculation module calculates the similarity of all items are as follows:
For any two article i and j, calculating formula of similarity are as follows:
Wherein, N1 is that article i and article j, there are the user of scoring behavior set, R simultaneouslyu,i,Ru,jIt is user u to article i,
The scoring of article j,Indicate the average score of article i and article j, Mi,jIndicate the distance coefficient of article i and article j, α
For the adjusting parameter of distance coefficient.
9. the personalized article recommender system according to claim 8 based on user geographical location, which is characterized in that described
The similarity of article is normalized in article similarity calculation module:
Wherein, simmax,simminFor the maximum value and minimum value in the similarity value of article, ei,jFor article i and object after normalization
The similarity of product j, ei,j∈(0,1)。
10. the personalized article recommender system according to claim 6 based on user geographical location, which is characterized in that institute
Commentary divides the method for scoring of the prediction module to each user in predicting article to be recommended are as follows:
Wherein, Nu(i) scored and score similar article with article i for user u.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910126341.6A CN109829792A (en) | 2019-02-20 | 2019-02-20 | A kind of personalized item recommendation method and system based on user location distribution |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910126341.6A CN109829792A (en) | 2019-02-20 | 2019-02-20 | A kind of personalized item recommendation method and system based on user location distribution |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109829792A true CN109829792A (en) | 2019-05-31 |
Family
ID=66863886
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910126341.6A Pending CN109829792A (en) | 2019-02-20 | 2019-02-20 | A kind of personalized item recommendation method and system based on user location distribution |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109829792A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112381616A (en) * | 2020-11-28 | 2021-02-19 | 武汉虹信技术服务有限责任公司 | Item recommendation guiding method and device and computer equipment |
| CN112948670A (en) * | 2021-02-05 | 2021-06-11 | 洛阳理工学院 | Method for constructing platform transaction recommendation model based on user behavior preference |
-
2019
- 2019-02-20 CN CN201910126341.6A patent/CN109829792A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112381616A (en) * | 2020-11-28 | 2021-02-19 | 武汉虹信技术服务有限责任公司 | Item recommendation guiding method and device and computer equipment |
| CN112948670A (en) * | 2021-02-05 | 2021-06-11 | 洛阳理工学院 | Method for constructing platform transaction recommendation model based on user behavior preference |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11762818B2 (en) | Apparatus, systems, and methods for analyzing movements of target entities | |
| US10417653B2 (en) | Inferring consumer affinities based on shopping behaviors with unsupervised machine learning models | |
| CN104391849B (en) | Incorporate the collaborative filtering recommending method of time contextual information | |
| CN104850641B (en) | A kind of method and device of recommendation information | |
| US20190197570A1 (en) | Location-based analytic platform and methods | |
| CN104935963B (en) | A kind of video recommendation method based on timing driving | |
| JP6293642B2 (en) | General-purpose graphs, rules and spatial structure based on recommendation engine | |
| CN105791085B (en) | Friend recommendation method in the social networks of position based on position and time | |
| US9501524B2 (en) | Searching and determining active area | |
| US8401897B1 (en) | System and method of displaying ads based on location | |
| KR102340463B1 (en) | Sample weight setting method and device, electronic device | |
| CN106776930B (en) | A kind of location recommendation method incorporating time and geographical location information | |
| Huang et al. | Using trajectories for collaborative filtering-based POI recommendation | |
| JP7285521B2 (en) | System and method for predicting similar mobile devices | |
| KR20130019629A (en) | System, method and computer readable recording medium for providing an analysis service about location requirements and trade area according to types of real property | |
| CN103020049A (en) | Searching method and searching system | |
| US20120150753A1 (en) | Quantitative valuation of real estate based on qualitative assessment thereof | |
| Tang et al. | Dynamic personalized recommendation on sparse data | |
| CN104751353A (en) | Cluster and Slope One prediction based collaborative filtering method | |
| CN109829792A (en) | A kind of personalized item recommendation method and system based on user location distribution | |
| CN110322067A (en) | Location of mobile users prediction technique based on factor graph model | |
| JP2015132539A (en) | Sunshine quantity calculating apparatus, route proposing apparatus, and sunshine quantity calculating method | |
| Wakamiya et al. | Crowd-sourced cartography: Measuring socio-cognitive distance for urban areas based on crowd's movement | |
| Chang et al. | Hotel recommendation based on surrounding environments | |
| Cao et al. | An Improved Slope One Recommendation Algorithm based on User Preferences |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190531 |