CN110020918A - A kind of recommendation information generation method and system - Google Patents
A kind of recommendation information generation method and system Download PDFInfo
- Publication number
- CN110020918A CN110020918A CN201910197097.2A CN201910197097A CN110020918A CN 110020918 A CN110020918 A CN 110020918A CN 201910197097 A CN201910197097 A CN 201910197097A CN 110020918 A CN110020918 A CN 110020918A
- Authority
- CN
- China
- Prior art keywords
- label
- vector
- user
- transaction
- recommendation information
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0623—Item investigation
- G06Q30/0625—Directed, with specific intent or strategy
- G06Q30/0629—Directed, with specific intent or strategy for generating comparisons
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of recommendation information generation method and systems, method includes: acquisition original transaction data, original transaction data includes user information and transaction record, assigns rule process user information and transaction record according to preset label to obtain user vector and Trade Vector；User vector and Trade Vector are handled based on cross product mould to obtain similarity between user's commodity and assign corresponding scoring；The sequence for carrying out user and commodity association according to scoring exports recommendation information according to Top-K list to obtain Top-K list.System is for executing method.The present invention is by parsing original transaction data to obtain user information and transaction record, rule process is assigned according to label with the matrix of relationship, based on cross product mould processing array to obtain similarity and scoring, the generation that recommendation information is carried out according to scoring improves the treatment effect to mass data relative to traditional dot product or range range mode.
Description
Technical field
The present invention relates to network information processing technical field, especially a kind of recommendation information generation method and system.
Background technique
With the development of the network business, the article sold in online store is more and more, commodity it is many kinds of
Have become a kind of information overload, user needs quickly to select oneself interested from the commodity of magnanimity and meet the object of demand
Product, recommender system just seem very to the interest detection of user as a kind of information filtering system, the recommender system in online store
It is important.
In traditional algorithm, client interests correlation detection often is done with dot product or distance, generally requires to introduce
Artificial rule improves, and the big later robustness of data volume is poor.
Summary of the invention
The present invention is directed to solve at least some of the technical problems in related technologies.For this purpose, of the invention
One purpose is to provide a kind of recommendation information generation method and system.
The technical scheme adopted by the invention is that:
In a first aspect, the present invention provides a kind of recommendation information generation method, comprising steps of obtaining original transaction data, institute
Stating original transaction data includes user information and transaction record, according to preset label assign rule process described in user information and
Transaction record is to obtain user vector and Trade Vector；The user vector and the Trade Vector are handled based on cross product mould to obtain
It takes similarity between the commodity of family and assigns corresponding scoring；Sequence of the user with commodity association is carried out according to scoring to obtain
Top-K list exports recommendation information according to Top-K list.
Preferably, the user information according to preset label imparting rule process and transaction record specifically include:
According to user information and transaction record described in preset label imparting rule match, label and transaction record based on user information
Label production Methods matrix；Using user tag as starting point, transaction label is that the vector of terminal is user vector；With label of trading
For terminal, another transaction label is that the vector of terminal is Trade Vector.
Preferably, the cross product mould processing specifically includes: calculating the length square of minimum two vectors, calculates on vector
The quadratic sum of element value calculates the product A of quadratic sum, calculates square B of the sum of products of the element value of the point on vector, and score S
Square S2=A-B, wherein element value is the value of a little corresponding label.
Preferably, it further comprises the steps of: and updates label imparting rule, the label based on update assigns user described in rule process
Information and transaction record are to obtain user vector and Trade Vector.
Preferably, updating label to assign rule includes increasing or decreasing label, specifically, being turned to the maximum of cross product mould sum
Objective function increases or decreases label based on the decision of brainstorming algorithm.
Second aspect, the present invention provide a kind of recommendation information generation system, comprising: original data processing module, for obtaining
Original transaction data is taken, the original transaction data includes user information and transaction record, assigns rule according to preset label
The user information and transaction record are handled to obtain user vector and Trade Vector；Grading module, for based at cross product mould
The user vector and the Trade Vector are managed to obtain similarity between user's commodity and assign corresponding scoring；Recommending module,
Sequence for carrying out user and commodity association according to scoring exports recommendation according to Top-K list to obtain Top-K list
Breath.
Preferably, the user information according to preset label imparting rule process and transaction record specifically include:
According to user information and transaction record described in preset label imparting rule match, label and transaction record based on user information
Label production Methods matrix；Using user tag as starting point, transaction label is that the vector of terminal is user vector；With label of trading
For terminal, another transaction label is that the vector of terminal is Trade Vector.
Preferably, the cross product mould processing specifically includes: calculating the length square of minimum two vectors, calculates on vector
The quadratic sum of element value calculates the product A of quadratic sum, calculates square B of the sum of products of the element value of the point on vector, and score S
Square S2=A-B, wherein element value is the value of a little corresponding label.
Preferably, further include update module, assign rule for updating label, the label based on update assigns rule process
The user information and transaction record are to obtain user vector and Trade Vector
Preferably, updating label to assign rule includes increasing or decreasing label, specifically, being turned to the maximum of cross product mould sum
Objective function increases or decreases label based on the decision of brainstorming algorithm.
The beneficial effects of the present invention are:
The present invention, to obtain user information and transaction record, is assigned at rule by parsing original transaction data according to label
Reason is with the matrix of relationship, and based on cross product mould processing array to obtain similarity and scoring, the life of recommendation information is carried out according to scoring
At improving the treatment effect to mass data relative to traditional dot product or range range mode.
Detailed description of the invention
Fig. 1 is a kind of schematic diagram of recommendation information generation method of the invention；
Fig. 2 is the schematic diagram that a kind of recommendation information of the invention generates system.
Specific embodiment
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
Mutually combination.
Embodiment 1
Recommendation form this for " having seen seeing also for this commodity ", either Jingdone district or Taobao, or easily purchase net
It stands, such recommended products can be described as the standard configuration of recommender system." going back for this film is liked in similar comment class website
Like ", " having paid close attention to this people also to pay close attention to " of social network sites, these are all an ancient proposed algorithm, are called base
In the collaborative filtering of item entry, its shortcoming is that it is less to will appear demand number, it is this not have the case where still recommendation repeatedly
The recommended method for considering the shopping interest change of user, is not able to satisfy the demand of lead referral not only, is also easy to because showing
Stage useless article causes that client's is discontented.
The present invention proposes corresponding method on the basis of the attribute change and commodity for considering client update,
The present embodiment provides a kind of recommendation information generation methods as shown in Figure 1, comprising steps of
S1, original transaction data is obtained, the original transaction data includes user information and transaction record, according to preset
Label assigns user information described in rule process and transaction record to obtain user vector and Trade Vector；
S2, the user vector and the Trade Vector are handled based on cross product mould to obtain similarity between user's commodity and assign
Give corresponding scoring；
S3, the sequence of user and commodity association is carried out according to scoring to obtain Top-K list, it is defeated according to Top-K list
Recommendation information out.
Wherein, step S1 obtains original transaction data, specifically can be the data from online shopping mall, number of users and
Commodity amount is often huger, and the demand to computing resource becomes bottleneck, so this programme is age-based, education degree is lived
Ground, the labels such as income level (label of this part is suitable for user information, the information filled in when deriving from user's registration) come
Classification and cluster user, get up the expenditure horizontal relevance that the price of article is born with user, can at least be concerned about use in this way
The bulk migration of the interest of family group.The corresponding customer count's size of article can be by price, and at the age, income level educates journey
(labels) such as degree carries out classification adjustment, and the sparse consistency of similarity calculated between article can be needed by statistical accuracy
It arbitrarily to have adjusted；And transaction record is the article of user's actual purchase, it can be with household goods, tool, food, service life
Labels are waited to be marked；It is corresponding, same label is met in user, it can be using other labels as the trend changed (i.e.
Vector), for example, belonging to 30 years old~40 years old, then meet same (age) label, at this point, becoming using education degree as variation
A vector then can be generated in gesture, the vector using the age as starting point, to different educational degree (primary school, junior middle school, university etc., it is right
Answer a value, this value, that is, matrix wherein one-dimensional coordinate), and so on, can be formed about user transformation vector (i.e.
User vector)；And Trade Vector is specially the trend of the transformation of the label of personal tradable commodity, meanwhile, when individual becomes same
When the crowd of one user information, more general trend can be formed, for example, belonging to 30 years old~40 years old, is bought in commodity
There is commodity first, and the frequency of occurrence of commodity first increases with age, then illustrate with age, commodity first is again
Or similar commodity it is purchased a possibility that be increased by, corresponding trend is Trade Vector.
Using the label of user, trade (commodity) label as the one-dimensional of coordinate system, then can form the network of personal connections of multidimensional,
That is relational matrix, using user tag as starting point, transaction label is the vector of terminal is user vector (i.e. which type of user's buying
Which type of commodity)；Using label of trading as terminal, another transaction label is that the vector of terminal is Trade Vector (i.e. user
After having bought which type of commodity, which type of commodity next purchase commodity can be), by the two modes, can obtain
The relationship of user tag and Commercial goods labels, the relationship of Commercial goods labels and (next) Commercial goods labels.
Relational matrix is often a sparse matrix, and not having at intersection record (i.e. the point of matrix) is null value；The dimension of vector
The age that can be user is spent, level of education, the numbers such as income level (i.e. the value of label, such as the age was from 10 years old to 100 years old),
One user tag representation vector it is one-dimensional, total dimension of this vector is total user tag quantity；The each dimension of vector takes
Value is that cluster user records the education of this article, can be the Boolean of behavior itself, is also possible to consumer behavior quantization
Such as length of time, number are how many, can also be the evaluation score of consumption, and the point of different labels joined a little on i.e. vector.
Initial data (i.e. original transaction data) is often rating matrix of the user to commodity, this matrix is often huge
And it is sparse, if directly carrying out the calculating of proposed algorithm using initial data, generally require to consume huge memory source,
And effect is undesirable.Therefore we are usually by the rating matrix data characterization of higher-dimension at the user vector and commodity vector of low-dimensional,
In order to the calculating of next step, final recommendation is finally obtained.
Original transaction data processing: user record log is carried out to the cleaning of original transaction data, rating matrix is obtained, does not have
Having interactive partial data is 0；
The characterization of user vector and commodity vector: matrix decomposition, figure insertion, the machine learning algorithms such as term vector characterization are used
Rating matrix is decomposed into user vector matrix and commodity vector matrix (specifically can be with the example of reference implementation example 3)；
Existing method often calculates commodity and commodity, user and user, commodity and use using distance or cosine similarity
The similarity degree at family.Present invention cross product mould and user is calculated, the similitude between item, to pushing away between commodity and user
It recommends and makes score, compared to distance or cosine similarity, range information and misalignment angle information can be taken into account.
In step S2, cross product mould is specifically included: being calculated the length square of two vectors, is sought the quadratic sum A of element value again
Be multiplied subtract two vectors dot product square B, finally extraction of square root obtain diamond shaped surface product (the i.e. similar commodity of the surface area or
The correlation of behavior, according to preset coefficient, the available scoring S of combined surface area), dot product is exactly the element value of same position
Multiplication is summed again.Wherein, element value is that (different labels at this time may be used there is no corresponding mathematical value for the value of a little corresponding label
Think its setting, such as food can be set to 11, clothes can be set to 12, purpose facilitates the subsequent meter of various labels
Calculate), according to the Computing Principle of cross product mould, calculate two vectors or more vectors, can be formed a variety of users, commodity it
Between relationship, being conducive to accurately to obtain user may interested article.
In step S3, Top-K list is obtained, is capable of providing the processing mode of a standard, the processing formula to avoid confusion
To reduce the stability of the operation of method；This is very important for for the recommender system of mass data, and list is
A series of Multidimensional numerical, the dimension of array represent relevant entry number, for example according to step S2, calculated by (vector) two-by-two
The degree of correlation be recorded in two-dimensional array, be recorded in three-dimensional array by the degree of correlation that three or three (vectors) calculate, successively class
It pushes away.Then respectively the array sort from large to small by the sequence of degree of correlation size, to find out the high quotient of the degree of correlation
Product race (commodity rent the set for having the commodity of same label).
It is directed to the pursuit of the update and user of commodity for trend, needs constantly to update recommendation information, therefore need
Update the label of user and the label of commodity；But the increase of label or reduce can significant impact to cross product mould meter
It calculates, that is, influences whether the scoring of script, therefore for the purpose for reducing the burden calculated, need suitably tradeoff to update label and assign
Rule is given, specifically, turning to objective function with the maximum of cross product mould sum, mark is increased or decreased based on the decision of brainstorming algorithm
Label.
The circulation process of recommendation includes:
It collects user information to add new entry (i.e. label), calculates the degree of correlation ranking (scoring) of user's entry, more
The cluster list (Top-K list) of new dependent merchandise race, recommends relevant entry to associated user.
Embodiment 2
The present embodiment provides a kind of recommendation informations to generate system, comprising:
Original data processing module 1, for obtaining original transaction data, the original transaction data include user information and
Transaction record, according to user information described in preset label imparting rule process and transaction record to obtain user vector and transaction
Vector；
Grading module 2, for handling the user vector and the Trade Vector based on cross product mould to obtain user's commodity
Between similarity and assign corresponding scoring；
Recommending module 3, for carrying out the sequence of user and commodity association according to scoring to obtain Top-K list, according to
Top-K list exports recommendation information.
Original transaction data is very large data, if directly carrying out the calculating of proposed algorithm using initial data,
It generally requires to consume huge memory source, and effect is undesirable.Therefore by the rating matrix data characterization of higher-dimension at low-dimensional
User vector and commodity vector finally obtain final recommendation: original data processing in order to the calculating of next step: will recommend
The user record log of system carries out the cleaning of initial data, obtains rating matrix, and the partial data not interacted is 0；User
The characterization of vector sum commodity vector: using matrix decomposition, figure insertion, and the machine learning algorithms such as term vector characterization divide rating matrix
Solution is user vector matrix and commodity vector matrix；Original method often calculates commodity and quotient using distance or cosine similarity
The similarity degree of product, user and user, commodity and user.Our method is with cross product mould and to calculate user, between item
Similitude, score is made to the recommendation between commodity and user, compared to distance or cosine similarity, can take into account distance letter
Breath and misalignment angle information；To being ranked up, Top-K list is obtained to user's commodity according to score height；User is carried out
Recommend.
Embodiment 3
The example for the specific field that the present embodiment is used to illustrate that the present invention can be applicable in, such as:
Store such as (Taobao, Amazon) etc. user-customized recommendeds, specific steps include: online
Store log is read, obtains rating matrix after cleaning data；Commodity and user information characterization are become into low-dimensional vector；
Using product module and user is calculated, the similitude between item makes score to the recommendation between commodity and user, obtain Top-
K list.
Social networks such as (Facebook, Twitter) user-customized recommended specific steps include: reading social networks,
The network topological diagram between user and user is obtained after cleaning data；Using figure embedded mobile GIS such as DeepWalk by network topological diagram
Each of user all characterize become low-dimensional vector；Using cross product mould and calculate the similitude between user, to user and
Score is made in recommendation between user, obtains Top-K list.
K12 teaching platform user learns route personalized recommendation, and specific steps include:
User's learning network is read, obtains the network topological diagram between knowledge point after cleaning data；Utilize figure embedded mobile GIS
Become low-dimensional vector as DeepWalk characterizes each of network topological diagram knowledge point；Know using cross product mould and to calculate
Know the similitude between point, is recorded according to user's history, the recommendation between user and knowledge point is done in conjunction with collaborative filtering
Score out obtains Top-K list, obtains the next step study route of user.
UGC platform (YouTube, bean cotyledon) personalized recommendation, it is similar with online store.
It is to be illustrated to preferable implementation of the invention, but the invention is not limited to the implementation above
Example, those skilled in the art can also make various equivalent variations on the premise of without prejudice to spirit of the invention or replace
It changes, these equivalent deformations or replacement are all included in the scope defined by the claims of the present application.
Claims (10)
1. a kind of recommendation information generation method, which is characterized in that comprising steps of
Original transaction data is obtained, the original transaction data includes user information and transaction record, is assigned according to preset label
User information and transaction record described in rule process are given to obtain user vector and Trade Vector；
The user vector and the Trade Vector are handled based on cross product mould to obtain similarity between user's commodity and assign correspondence
Scoring；
The sequence for carrying out user and commodity association according to scoring exports recommendation according to Top-K list to obtain Top-K list
Breath.
2. a kind of recommendation information generation method according to claim 1, which is characterized in that described to be assigned according to preset label
User information and transaction record described in rule process is given to specifically include:
According to user information and transaction record described in preset label imparting rule match, label and transaction based on user information
The label production Methods matrix of record；
Using user tag as starting point, transaction label is that the vector of terminal is user vector；
Using label of trading as terminal, another transaction label is that the vector of terminal is Trade Vector.
3. a kind of recommendation information generation method according to claim 1, which is characterized in that the specific packet of cross product mould processing
It includes:
The length square of minimum two vectors is calculated, the quadratic sum of the element value on vector is calculated, calculates the product A of quadratic sum,
Calculate square B, square S2=A-B for the S that scores of the sum of products of the element value of the point on vector, wherein element value is that point corresponds to
Label value.
4. a kind of recommendation information generation method according to claim 1, which is characterized in that further comprise the steps of:
It updates label and assigns rule, the label based on update assigns user information described in rule process and transaction record to obtain use
Family vector sum Trade Vector.
5. a kind of recommendation information generation method according to claim 4, which is characterized in that updating label imparting rule includes
Label is increased or decreased, specifically,
Objective function is turned to the maximum of cross product mould sum, label is increased or decreased based on the decision of brainstorming algorithm.
6. a kind of recommendation information generates system characterized by comprising
Original data processing module, for obtaining original transaction data, the original transaction data includes user information and transaction
Record, according to preset label assign rule process described in user information and transaction record with obtain user vector and trade to
Amount；
Grading module obtains similar between user's commodity for handling the user vector and the Trade Vector based on cross product mould
It spends and assigns corresponding scoring；
Recommending module, for carrying out sequence of the user with commodity association to obtain Top-K list, according to Top-K according to scoring
List exports recommendation information.
7. a kind of recommendation information generates system according to claim 6, which is characterized in that described to be assigned according to preset label
User information and transaction record described in rule process specifically include:
According to user information and transaction record described in preset label imparting rule match, label and transaction based on user information
The label production Methods matrix of record；
Using user tag as starting point, transaction label is that the vector of terminal is user vector；
Using label of trading as terminal, another transaction label is that the vector of terminal is Trade Vector.
8. a kind of recommendation information generates system according to claim 6, which is characterized in that the specific packet of cross product mould processing
It includes:
The length square of minimum two vectors is calculated, the quadratic sum of the element value on vector is calculated, calculates the product A of quadratic sum,
Calculate square B, square S2=A-B for the S that scores of the sum of products of the element value of the point on vector, wherein element value is that point corresponds to
Label value.
9. a kind of recommendation information generates system according to claim 6, which is characterized in that further include update module, for more
New label assigns rule, and the label based on update assigns user information described in rule process and transaction record to obtain user vector
And Trade Vector.
10. a kind of recommendation information generates system according to claim 9, which is characterized in that updating label imparting rule includes
Increase or decrease label, specifically, turn to objective function with the maximum of cross product mould sum, determine to increase based on brainstorming algorithm or
Reduce label.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910197097.2A CN110020918B (en) | 2019-03-15 | 2019-03-15 | Recommendation information generation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910197097.2A CN110020918B (en) | 2019-03-15 | 2019-03-15 | Recommendation information generation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110020918A true CN110020918A (en) | 2019-07-16 |
| CN110020918B CN110020918B (en) | 2022-06-21 |
Family
ID=67189597
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910197097.2A Active CN110020918B (en) | 2019-03-15 | 2019-03-15 | Recommendation information generation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110020918B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113069773A (en) * | 2021-03-31 | 2021-07-06 | 南方科技大学 | Game friend recommendation method, system, server and storage medium based on contact |
| CN114429371A (en) * | 2022-04-06 | 2022-05-03 | 新石器慧通（北京）科技有限公司 | Unmanned vehicle-based commodity marketing method and device, electronic equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103425791A (en) * | 2013-08-28 | 2013-12-04 | 中国科学院深圳先进技术研究院 | Article recommending method and device based on ant colony collaborative filtering algorithm |
| CN103593792A (en) * | 2013-11-13 | 2014-02-19 | 复旦大学 | Individual recommendation method and system based on Chinese knowledge mapping |
| EP2706497A1 (en) * | 2012-09-11 | 2014-03-12 | Google Inc. | Method for recommending musical entities to a user |
| CN106056427A (en) * | 2016-05-25 | 2016-10-26 | 中南大学 | Spark-based big data hybrid model mobile recommending method |
| CN106127525A (en) * | 2016-06-27 | 2016-11-16 | 浙江大学 | A kind of TV shopping Method of Commodity Recommendation based on sorting algorithm |
| US20180247362A1 (en) * | 2017-02-24 | 2018-08-30 | Sap Se | Optimized recommendation engine |
-
2019
- 2019-03-15 CN CN201910197097.2A patent/CN110020918B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2706497A1 (en) * | 2012-09-11 | 2014-03-12 | Google Inc. | Method for recommending musical entities to a user |
| CN103425791A (en) * | 2013-08-28 | 2013-12-04 | 中国科学院深圳先进技术研究院 | Article recommending method and device based on ant colony collaborative filtering algorithm |
| CN103593792A (en) * | 2013-11-13 | 2014-02-19 | 复旦大学 | Individual recommendation method and system based on Chinese knowledge mapping |
| CN106056427A (en) * | 2016-05-25 | 2016-10-26 | 中南大学 | Spark-based big data hybrid model mobile recommending method |
| CN106127525A (en) * | 2016-06-27 | 2016-11-16 | 浙江大学 | A kind of TV shopping Method of Commodity Recommendation based on sorting algorithm |
| US20180247362A1 (en) * | 2017-02-24 | 2018-08-30 | Sap Se | Optimized recommendation engine |
Non-Patent Citations (2)
| Title |
|---|
| PENG-YU LU 等: ""Hybrid Recommendation Algorithm for E-Commerce Website"", 《 2015 8TH INTERNATIONAL SYMPOSIUM ON COMPUTATIONAL INTELLIGENCE AND DESIGN (ISCID)》 * |
| 孙建凯: ""面向排序的个性化推荐算法研究与实现"", 《中国优秀硕士学位论文全文数据库 · 信息科技辑》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113069773A (en) * | 2021-03-31 | 2021-07-06 | 南方科技大学 | Game friend recommendation method, system, server and storage medium based on contact |
| CN113069773B (en) * | 2021-03-31 | 2024-04-05 | 南方科技大学 | Close-contact-based game friend recommendation method, system, server and storage medium |
| CN114429371A (en) * | 2022-04-06 | 2022-05-03 | 新石器慧通（北京）科技有限公司 | Unmanned vehicle-based commodity marketing method and device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110020918B (en) | 2022-06-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gao et al. | Exploring temporal effects for location recommendation on location-based social networks | |
| Bauer et al. | Recommender systems based on quantitative implicit customer feedback | |
| CN110222272A (en) | A kind of potential customers excavate and recommended method | |
| CN104463637A (en) | Commodity recommendation method and device based on electronic business platform and server | |
| US20100268661A1 (en) | Recommendation Systems | |
| Ba et al. | Clustering collaborative filtering recommendation system based on SVD algorithm | |
| Hu et al. | An item orientated recommendation algorithm from the multi-view perspective | |
| CN105786838A (en) | Information matching processing method and apparatus | |
| JP6976207B2 (en) | Information processing equipment, information processing methods, and programs | |
| CN108805598A (en) | Similarity information determines method, server and computer readable storage medium | |
| CN111310038B (en) | Information recommendation method and device, electronic equipment and computer-readable storage medium | |
| JP2017054214A (en) | Determination device, learning device, information distribution device, determination method, and determination program | |
| US11587028B2 (en) | Computer system for optimizing garment inventory of retailer based on shapes of users | |
| Hemalatha | Market basket analysis–a data mining application in Indian retailing | |
| CN111104606B (en) | Weight-based conditional wandering chart recommendation method | |
| Fan et al. | A robust multi-criteria recommendation approach with preference-based similarity and support vector machine | |
| CN110020918A (en) | A kind of recommendation information generation method and system | |
| US20210279785A1 (en) | Method for providing initial size fit indicator | |
| CN111460300B (en) | Network content pushing method, device and storage medium | |
| Krisdhamara et al. | Improvement of collaborative filtering recommendation system to resolve sparsity problem using combination of clustering and opinion mining methods | |
| Pandey et al. | Recommending e-commerce products on cold start and long tail using transaction data | |
| Bai | A Hierarchical Model of E-Commerce Sellers Based on Data Mining. | |
| Sahu et al. | Personalized recommendation engine using HADOOP | |
| Haripriya et al. | Using Social Media to Promote E-Commerce Business. | |
| TWI635451B (en) | Similarity analysis method and system using virtual goods in recommendation system |
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 |