CN109474703A - Personalized product combines method for pushing, apparatus and system - Google Patents
Personalized product combines method for pushing, apparatus and system Download PDFInfo
- Publication number
- CN109474703A CN109474703A CN201811581395.3A CN201811581395A CN109474703A CN 109474703 A CN109474703 A CN 109474703A CN 201811581395 A CN201811581395 A CN 201811581395A CN 109474703 A CN109474703 A CN 109474703A
- Authority
- CN
- China
- Prior art keywords
- cluster
- product
- characteristic
- terminal
- informations
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/55—Push-based network services
-
- 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)
- Engineering & Computer Science (AREA)
- Finance (AREA)
- Marketing (AREA)
- Signal Processing (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application provides a kind of personalized product combination method for pushing, apparatus and system, wherein method includes: that multiple terminal characteristic informations and multiple Product Feature Informations are extracted from historical trading data, and building includes the characteristic information set of multiple terminal characteristic informations and multiple Product Feature Informations；Wherein, the characteristic dimension in terminal characteristic information is identical as the characteristic dimension in Product Feature Information；Cluster operation is executed to the characteristic information set and obtains multiple clusters；Wherein each cluster includes at least one Product Feature Information, and, one or more terminal characteristic informations；The corresponding terminal push personalized product combination of terminal characteristic information into each cluster, personalized product combination include the corresponding product of one or more Product Feature Information in the cluster.The application pushes personalized product combination based on transaction data, and the accuracy rate of push personalized product combination can be improved.
Description
Technical field
This application involves Internet technical fields more particularly to personalized product to combine method for pushing, apparatus and system.
Background technique
Recently as the development of financial technology in internet, the business combined to terminal push personalized product is derived
(personalized product combination includes one or more products), so that terminal shows personalized product combination, facilitates user to individual character
Change product mix, which executes, the operation such as checks, pays close attention to, subscribing to.
In order to realize the business combined to terminal push personalized product, the history row of multiple terminals would generally be acquired at present
(for example, browsing product behavioural characteristic, concern product behavioural characteristic and subscription product behavioural characteristic) is characterized to build preference mould
Type, later use preference pattern combine to estimate the personalized product of terminal, to combine to terminal push personalized product.
But the factors such as particularity and law compliance in view of class scene of trading belonging to product, it is difficult to be directed to individual character
Change product mix operation expanding social scene more abundant, leading to the building of preference pattern, there are scaling concerns
(Scalability) and Sparse Problems (Sparsity).
Therefore, the mode that personalized product combination is obtained based on preference pattern, cannot accurately determine the personalization of terminal
Product mix leads to that accurate personalized product combination cannot be pushed to terminal.
Summary of the invention
In consideration of it, the application, which provides personalized product, combines method for pushing, apparatus and system, it is accurate to push to terminal
Personalized product combination.
To achieve the goals above, the application provides following technical characteristics:
A kind of personalized product combination method for pushing, comprising:
Multiple terminal characteristic informations and multiple Product Feature Informations are extracted from historical trading data, building includes multiple ends
Hold the characteristic information set of characteristic information and multiple Product Feature Informations；Wherein, the characteristic dimension in the terminal characteristic information
It is identical as the characteristic dimension in Product Feature Information；
Cluster operation is executed to the characteristic information set and obtains multiple clusters；Wherein each cluster includes at least one product spy
Reference breath, and, one or more terminal characteristic informations；
The corresponding terminal push personalized product combination of terminal characteristic information, the personalized product combination into each cluster
Including the corresponding product of Product Feature Informations one or more in the cluster.
It is optionally, described that the multiple clusters of cluster operation acquisition are executed to the characteristic information set, comprising:
K Product Feature Information is determined from the characteristic information set, respectively as the cluster center of K cluster；
It divides remaining characteristic information in the characteristic information set, to the smallest cluster of cluster centre distance, obtains K
Cluster；
Redefine the cluster center of K cluster；Wherein, the cluster center of each cluster is a Product Feature Information in the cluster；
Judge whether the cluster center of K cluster changes；If then entering step: dividing its in the characteristic information set
Remaining characteristic information extremely and in the smallest cluster of cluster centre distance, obtains K cluster；
Determine that cluster operation terminates, and K cluster is determined as multiple clusters of cluster operation acquisition if not.
Optionally, the cluster center for redefining K cluster, comprising:
Following processes are executed for each cluster in K cluster:
The true cluster center for calculating cluster, calculate separately each Product Feature Information and the true cluster center in cluster away from
From, with the true the smallest Product Feature Information of cluster centre distance, the cluster center of cluster will be determined as in cluster.
Optionally, described that multiple terminal characteristic informations and multiple Product Feature Informations, structure are extracted from historical trading data
Build the characteristic information set comprising multiple terminal characteristic informations and multiple Product Feature Informations；Wherein, the terminal characteristic information
In characteristic dimension it is identical as the characteristic dimension in Product Feature Information, comprising:
From historical trading data, extract with multiple terminals correspondingly multiple terminal feature set and extract with it is more
A product multiple product feature set correspondingly；Wherein, the characteristic dimension in terminal feature set and product feature set
In characteristic dimension it is identical；
Determine the corresponding weight of each characteristic dimension；
To multiple terminal feature set and the multiple product feature set, vectorization operation and normalization operation are executed,
Obtain multiple terminal characteristic informations and multiple Product Feature Informations；
By multiple terminal characteristic informations and multiple Product Feature Informations, it is determined as characteristic information set.
Optionally, the characteristic dimension in terminal feature set and the characteristic dimension in product feature set, comprising:
Assets class, transaction class, profit and loss class, interest class and ability class.
A kind of personalized product combination driving means, comprising:
Construction unit, for extracting multiple terminal characteristic informations and multiple Product Feature Informations from historical trading data,
Building includes the characteristic information set of multiple terminal characteristic informations and multiple Product Feature Informations；Wherein, the terminal feature letter
Characteristic dimension in breath is identical as the characteristic dimension in Product Feature Information；
Cluster cell obtains multiple clusters for executing cluster operation to the characteristic information set；Wherein each cluster includes
At least one Product Feature Information, and, one or more terminal characteristic informations；
Push unit, it is described for the corresponding terminal push personalized product combination of terminal characteristic information into each cluster
Personalized product combination includes the corresponding product of one or more Product Feature Information in the cluster.
Optionally, the cluster cell includes:
First determines cluster center cell, for determining K Product Feature Information from the characteristic information set, makees respectively
For the cluster center of K cluster；
Division unit, for divide remaining characteristic information in the characteristic information set, to cluster centre distance it is the smallest
In cluster, K cluster is obtained；
Second determines cluster center cell, for redefining the cluster center of K cluster；Wherein, the cluster center of each cluster is to be somebody's turn to do
A Product Feature Information in cluster；
Judging unit, for judging whether the cluster center of K cluster changes；
Enter division unit if the cluster center of K cluster changes；
Determination unit, if the cluster center for K cluster does not change, it is determined that cluster operation terminates, and K cluster is determined
The multiple clusters obtained for cluster operation.A kind of personalized product combination supplying system, comprising:
Server, for extracting multiple terminal characteristic informations and multiple Product Feature Informations, structure from historical trading data
Build the characteristic information set comprising multiple terminal characteristic informations and multiple Product Feature Informations；Wherein, the terminal characteristic information
In characteristic dimension it is identical as the characteristic dimension in Product Feature Information；Cluster operation is executed to the characteristic information set to obtain
Multiple clusters；Wherein each cluster includes at least one Product Feature Information, and, one or more terminal characteristic informations；To each cluster
The corresponding terminal push personalized product combination of middle terminal characteristic information, the personalized product combination include in the cluster one or
The corresponding product of multiple Product Feature Informations；
Terminal, for receiving and showing that personalized product combines.
A kind of electronic equipment, comprising:
Processor；And
Memory, for storing the executable instruction of the processor；
Wherein, the processor, which is configured to execute the personalized product combination via the executable instruction is executed, pushes away
Delivery method.
A kind of storage medium, for the storage medium for storing software program, which can be used for realizing described
Property product mix method for pushing.
By the above technological means, may be implemented it is following the utility model has the advantages that
Compared to usage history behavioural characteristic (external feature) building preference pattern come determine personalized product combination and
Speech, it is a to determine in a manner of clustering that the application proposes the characteristic information collection determined based on historical trading data (internal characteristics) merging
Property product mix.
In order to realize the cluster operation of two different objects of terminal and product, the application make terminal feature set with
Product feature set characteristic dimension having the same, so as to which terminal and product are accordingly to be regarded as same type in cluster operation
Object, and then can be realized the cluster operation of terminal and product.
Since historical trading data (internal characteristics) is more able to reflect terminal than historical behavior feature (external feature) and produces
Constitutive relations between product, so can accurately determine that personalized product combines based on historical trading data, to improve to end
The accuracy rate of end push personalized product combination.
Detailed description of the invention
In order to illustrate the technical solutions in the embodiments of the present application or in the prior art more clearly, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of application for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is a kind of schematic diagram of personalized product combination supplying system disclosed in the embodiment of the present application；
Fig. 2 is a kind of flow chart of personalized product combination method for pushing disclosed in the embodiment of the present application；
Fig. 3 is a kind of flow chart of personalized product combination driving means disclosed in the embodiment of the present application.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of embodiments of the present application, instead of all the embodiments.It is based on
Embodiment in the application, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall in the protection scope of this application.
It is found by the applicant that: the prior art would generally acquire the historical behavior feature of multiple terminals (for example, historical viewings product
Object behavior feature, concern product object behavioural characteristic and subscribe to product object behavioural characteristic) build preference pattern.
Historical behavior feature is that the external level of terminal user shows the behavioural characteristic come, so historical behavior feature is main
Reflect the relationship between terminal and product from external level, so historical behavior feature is properly termed as external feature.
Applicant has found in the course of the research: there are a large amount of historical trading data between terminal and product object, going through
History transaction data can reflect that the relationship between terminal and product (especially can reflect out between terminal and product from inherent level
The features such as financial feature, such as average daily assets, share-holding number of days, short middle or long line preference), so, it will be mentioned from historical trading data
The feature taken is known as internal characteristics.
It is understood that due to internal characteristics than external feature can embody to a deeper level terminal and product object it
Between constitutive relations, so based on historical trading data can more acurrate determining personalized product combine, pushed away to improve to terminal
The accuracy rate for sending personalized product to combine.
For the ease of skilled in the art realises that application scenarios, the application provide a kind of push of personalized product combination
System.Referring to Fig. 1, the supplying system of personalized product combination includes:
Server 100, the multiple terminals 200 being connected with server 100.
Terminal may include the electronic equipments such as mobile phone, computer, ipad, no longer illustrate one by one herein；Server 100 can
To be realized using individual server or server cluster, for being implemented without limitation in practical application for server.
It is understood that server can store transaction during server 100 and terminal 200 execute product trading
Data.The transaction data of a period of time is known as historical trading data.Product trading, which is executed, about server 100 and stores history hands over
The process of easy data has been mature technology, and details are not described herein.
This application provides a kind of personalized products to combine method for pushing, is applied to personalized product shown in FIG. 1 and combines
Supplying system.
Referring to fig. 2, personalized product combination method for pushing may include steps of:
Step S201: server determines multiple characteristic dimensions.
The product feature of the terminal feature and characterization product characteristic that characterize user personality under normal conditions belongs to different characteristic
Type, so terminal feature and product feature can not carry out cluster operation.
The application can be realized terminal feature and product feature execution cluster operation in order to subsequent, and server can determine whether multiple
Characteristic dimension, terminal and product can share the multiple characteristic dimension.
That is, the application personalizes product under the premise of characteristic dimension of the terminal feature using characterization user personality,
So that product feature is also applied for the characteristic dimension of characterization user personality, so that characteristic dimension and product in terminal feature
Characteristic dimension is identical in feature.
Terminal feature and product feature can be accordingly to be regarded as same characteristic features type in this way, and then can be realized terminal feature letter
The cluster operation of breath set and Product Feature Information set.
For this purpose, multiple characteristic dimensions of terminal and product can be suitable for taking human as setting, and, the meter of multiple characteristic dimensions
Rule is calculated, and stores the computation rule of multiple characteristic dimensions and multiple characteristic dimensions in server.
Multiple characteristic dimensions of server storage are illustrated below, are showing for multiple characteristic dimensions referring to table 1
Example explanation:
Table 1
Multiple characteristic dimensions in server, can be dynamically determined according to practical application scene (can be with dynamic expansion spy
Sign dimension or dynamic delete characteristic dimension), certainly, the computation rule of characteristic dimension can also be determined according to practical application scene,
Specific computation rule about characteristic dimension is no longer described in detail.
Step S202: server is that each characteristic dimension assigns corresponding weight.
Server stores multiple characteristic dimensions, and multiple features are divided into multiple types.It is exemplified by Table 1, it can will be more
A characteristic dimension divides are as follows: assets class, transaction class, five seed type of profit and loss class, interest class and ability class.
It can determine that different type characteristic dimension to personalized product combined effect degree, is according to practical application scene
The big characteristic dimension of influence degree assigns greater weight, is that the small characteristic dimension of influence degree assigns smaller weight.Specific implementation
Process has been mature technology, and in this not go into detail.
The above-mentioned preparatory implementation procedure for the application, is described below the implementation procedure of the application.
Step S203: server extracts multiple terminal characteristic informations and multiple product features letter from historical trading data
Breath, building include the characteristic information set of multiple terminal characteristic informations and multiple Product Feature Informations；Wherein, terminal characteristic information
In characteristic dimension it is identical as the characteristic dimension in Product Feature Information.
According to one embodiment provided by the present application, step 203 may include step S2031~S2034:
Step S2031: server extracts and multiple terminals multiple terminal features correspondingly from historical trading data
It set and extracts and multiple products multiple product feature set correspondingly；Wherein, the feature dimensions in terminal feature set
It spends identical as the characteristic dimension in product feature set.
Multiple terminals and multiple products can be to need to be implemented in personalized product combined recommendation business in the present embodiment
Whole terminals and all over products；It is of course also possible to be part terminal and portioned product.Concrete application scene can be according to applied field
Depending on scape, it is not limited here.
Server is stored with the computation rule of historical trading data, multiple characteristic dimensions and each characteristic dimension.For more
For a terminal, it is corresponding each characteristic dimension can be extracted according to the computation rule of each characteristic dimension from historical trading data
Terminal feature, to obtain and multiple terminals multiple terminal feature set correspondingly.
Practical historical trading data is customer transaction data, so extracting the corresponding terminal of terminal from historical trading data
Characteristic set, being equivalent in fact and extracting characteristic dimension in the corresponding user characteristics set of user namely terminal feature set is table
Levy the characteristic dimension of user personality.
In order to realize the cluster operation of terminal feature set and product feature set, product is personalized, that is, is directed to
For multiple products: also according to the calculating of each characteristic dimension (characteristic dimension of characterization user personality) from historical trading data
Rule extracts the corresponding product feature of each characteristic dimension, to obtain and multiple products multiple product features correspondingly
Set.
Characteristic dimension in terminal feature set is identical as the characteristic dimension in product feature set, so terminal feature collection
It closes and product feature set can be considered same characteristic features type.
Step S2032: server obtains the corresponding weight of each characteristic dimension.
Data value between different characteristic dimension may differ by it is very big, so need by the data of each characteristic dimension into
Row normalized, normalized can normalize mode using min-max, and specific implementation process has been mature technology,
This is no longer described in detail.
Step S2033: server executes vectorization behaviour to multiple terminal feature set and the multiple product feature set
Work and normalization operation, obtain multiple terminal characteristic informations and multiple Product Feature Informations.
It, can be by terminal feature set and product feature set vectorization for the ease of subsequent processing.With terminal feature
For set, it is assumed that terminal feature set includes T seed type feature, and each type includes N number of characteristic dimension, then terminal feature collection
Closing can be indicated using T N-dimensional vector.
Certainly, characteristic dimension can be different in each type feature.It is exemplified by Table 1, terminal feature set includes 5 kinds
Type feature: assets class includes 8 kinds of characteristic dimensions, then can be indicated using 8 dimensional vectors；Class of trading includes 7 kinds of feature dimensions
Degree can be indicated using 7 dimensional vectors；Profit and loss class includes that 13 kinds of characteristic dimensions can then be indicated using 13 dimensional vectors；It is emerging
Interesting class includes that 4 kinds of characteristic dimensions can then be indicated using 4 dimensional vectors；Ability class includes 4 kinds of characteristic dimensions；It can then use
One 4 dimensional vector indicates.
In multiple terminal feature set after normalization and vectorization, multiple terminal feature set are generated, for the ease of
It is distinguished with multiple terminal feature set and is known as multiple terminal characteristic informations.
In multiple product feature set after normalization and vectorization, multiple product feature vector set are generated, in order to
Convenient for being distinguished with multiple product feature set, referred to as multiple Product Feature Informations.
Step S2034: building includes the characteristic information set of multiple terminal characteristic informations and multiple Product Feature Informations.
Meet step S203 and enter step S204: it is multiple that server executes cluster operation acquisition to the characteristic information set
Cluster；Wherein each cluster includes at least one Product Feature Information, and, one or more terminal characteristic informations.
As the term suggests cluster operation is to be brought together similar characteristic information, so being in one after cluster operation
Terminal characteristic information and Product Feature Information in a cluster are more similar.Namely the corresponding terminal of terminal characteristic information and production
The corresponding product of product characteristic information be it is more similar, that is, be suitble to the terminal into the cluster to recommend the product in the cluster.
According to one embodiment provided by the present application, step S204 includes step S2041~S2044:
Step S2041: server determines K Product Feature Information from the characteristic information set, respectively as K cluster
Cluster center.
Server determines K Product Feature Information after determining characteristic information set from characteristic information set at random, makees
For the cluster center of K cluster.Why need to select K Product Feature Information rather than terminal characteristic information, its object is to make
It obtains each cluster and includes at least a Product Feature Information, so that the corresponding Products Show of Product Feature Information to be given to the end in the cluster
Hold the corresponding terminal of characteristic information.
Step S2042: server divides remaining characteristic information in the characteristic information set, to minimum with cluster centre distance
Cluster in, obtain K cluster.
Server (includes terminal characteristic information for remaining characteristic information in characteristic information set in addition to K cluster center
With remaining Product Feature Information) in each characteristic information perform the following operations:
Calculate separately the Euclidean distance between characteristic information and K cluster center, the determining Euclidean distance with this feature information
This feature information is divided in cluster belonging to the cluster center by the smallest cluster center.
Each characteristic information is performed both by upper segment description operation to server in remaining characteristic information in characteristic information set
Afterwards, it just realizes once to the cluster operation of characteristic information set.That is, executing cluster operation to characteristic information set obtains K cluster.
Step S2043: server redefines the cluster center of K cluster；Wherein, the cluster center of each cluster is a production in the cluster
Product characteristic information.
Server executes following processes for each cluster in K cluster center:
S1: the true cluster center of server calculating cluster.
Assuming that containing M characteristic information (terminal characteristic information and Product Feature Information), each characteristic information in some cluster
Comprising a characteristic type, the type includes N-dimensional vector.
Then including being a characteristic information, this feature information includes a characteristic type, this feature type at true cluster center
Comprising N-dimensional vector, the data value of every dimensional vector is the average value of the dimensional vector data value in M characteristic information.
S2: server calculates separately in cluster each Product Feature Information at a distance from the true cluster center.
For each characteristic information in cluster, the Euclidean distance of each characteristic information Yu true cluster center is calculated.
S3: server, with the true the smallest Product Feature Information of cluster centre distance, will be determined as the cluster center of cluster in cluster.
Since true cluster center may not correspond to some Product Feature Information, in order to guarantee at least one production of each cluster
Product characteristic information, the present embodiment are biased to Product Feature Information when redefining cluster center.
Guarantee that cluster center is Product Feature Information, can guarantee at least one product feature in each cluster so every time
Information, its object is to so as to by the corresponding Products Show of Product Feature Information to the terminal characteristic information corresponding end in the cluster
End.
Meet step S2043 and enter step S2044: server judges whether the cluster center of K cluster changes；If then into
Enter step S2042, terminates if not.
Server judges original cluster center of K cluster and whether redefine the cluster center after K cluster completely the same, if complete
Complete consistent namely be no longer changed, then K cluster center, which no longer changes, also illustrates that K cluster is no longer changed, characteristic information collection
Clustered completion is closed, cluster operation can be terminated.
If not quite identical namely cluster center also changes, then it represents that characteristic information set does not cluster completion also, can be with
Enter step S2042, unlatching recycles cluster operation next time, until K cluster center no longer changes namely K cluster no longer changes.
It meets step S204 and enters step S205: the corresponding terminal push individual character of server terminal characteristic information into each cluster
Change product mix, personalized product combination includes the corresponding product of one or more Product Feature Information in the cluster.
It is understood that including at least one Product Feature Information and several terminals in each cluster after cluster operation
Characteristic information, since Product Feature Information is corresponding with product, terminal characteristic information is corresponding with terminal；So each cluster includes at least
One product and several terminals.
The distance between product recently, for the product in other clusters, is located in terminal and the cluster in each cluster
Product in a cluster is more adapted to terminal in the cluster.Therefore, in each cluster product be terminal in the cluster personalization
Product.
Server is performed both by following processes for each cluster in K cluster:
It is understood that the corresponding product of a Product Feature Information, it is possible to by one or more in the cluster
The corresponding product of Product Feature Information, composition personalized product combination.
Personalized product combination can be using tabular form expression, optionally, can be by products one or more in list
Random alignment so that each product be recommended frequency equilibrium, then also in balanced each cluster product by customer investment
Frequency is invested, the standardization of industry is enhanced.
Step S206: multiple terminals receive and show that personalized product combines.
After pushing corresponding personalized product combination to multiple terminals, multiple terminals can be shown with personalization server
Product mix the operation such as checks, pays close attention to, subscribes to so that user executes.
Through the foregoing embodiment it is known that the application has the following beneficial effects:
Compared to usage history behavioural characteristic namely external feature come determine personalized product combination for, the application propose
Determine that personalized product is combined based on transaction data namely internal characteristics.Since internal characteristics are more able to reflect than external feature
Push can be improved so pushing personalized product combination based on transaction data in constitutive relations between terminal and product
The accuracy rate of property product mix.
Also, for the cluster operation for realizing terminal and product, the application personalizes product object, by the spy of terminal
The characteristic set characteristic dimension having the same with product is closed in collection, realizes that terminal object and product object are accordingly to be regarded as same type
Object, and then can be realized the cluster operation of terminal and product.
User and product are executed cluster operation by extended meeting after the application, so that it is determined that the personalized product being most adapted to user
Combination, and then push to the corresponding terminal of user.
Referring to Fig. 3, the application also provides a kind of personalized product combination driving means, comprising:
Construction unit 31, for extracting multiple terminal characteristic informations and multiple product features letter from historical trading data
Breath, building include the characteristic information set of multiple terminal characteristic informations and multiple Product Feature Informations；Wherein, the terminal feature
Characteristic dimension in information is identical as the characteristic dimension in Product Feature Information；
Cluster cell 32 obtains multiple clusters for executing cluster operation to the characteristic information set；Wherein each cluster packet
At least one Product Feature Information is included, and, one or more terminal characteristic informations；
Push unit 33, for the corresponding terminal push personalized product combination of terminal characteristic information, institute into each cluster
Stating personalized product combination includes the corresponding product of one or more Product Feature Information in the cluster.
Wherein the cluster cell 32 includes:
First determines cluster center cell 321, for determining K Product Feature Information from the characteristic information set, point
Cluster center not as K cluster；
Division unit 322, for dividing remaining characteristic information in the characteristic information set, to minimum with cluster centre distance
Cluster in, obtain K cluster；
Second determines cluster center cell 324, for redefining the cluster center of K cluster；Wherein, the cluster center of each cluster is
A Product Feature Information in the cluster；
Judging unit 325, for judging whether the cluster center of K cluster changes；
Enter division unit 322 if the cluster center of K cluster changes；
Cluster unit 326 is determined, if the cluster center for K cluster does not change, it is determined that cluster operation terminates, by K
Cluster is determined as multiple clusters of cluster operation acquisition.
Wherein, second determine that cluster center cell 324 includes:
Following processes are executed for each cluster in K cluster: being calculated the true cluster center of cluster, calculated separately each production in cluster
Product characteristic information at a distance from the true cluster center, by cluster with the true the smallest Product Feature Information of cluster centre distance, really
It is set to the cluster center of cluster.
Wherein, construction unit 31 includes:
Extraction unit 311 extracts and multiple terminals multiple terminal features correspondingly for from historical trading data
It set and extracts and multiple products multiple product feature set correspondingly；Wherein, the feature dimensions in terminal feature set
It spends identical as the characteristic dimension in product feature set；
Weight unit 312 is determined, for determining the corresponding weight of each characteristic dimension；
Aggregation units 313 are handled, for executing vector to multiple terminal feature set and the multiple product feature set
Change operation and normalization operation, obtains multiple terminal characteristic informations and multiple Product Feature Informations；
Aggregation units 314 are determined, for by multiple terminal characteristic informations and multiple Product Feature Informations, determination to be characterized letter
Breath set.
Wherein, the characteristic dimension in terminal feature set and the characteristic dimension in product feature set, comprising:
Assets class, transaction class, profit and loss class, interest class and ability class.
About the specific implementation of personalized product combination driving means, embodiment shown in Fig. 2 may refer to, herein no longer
It repeats.
Referring to Fig. 1, the application provides a kind of personalized product combination supplying system, comprising:
Server, for extracting multiple terminal characteristic informations and multiple Product Feature Informations, structure from historical trading data
Build the characteristic information set comprising multiple terminal characteristic informations and multiple Product Feature Informations；Wherein, the terminal characteristic information
In characteristic dimension it is identical as the characteristic dimension in Product Feature Information；Cluster operation is executed to the characteristic information set to obtain
Multiple clusters；Wherein each cluster includes at least one Product Feature Information, and, one or more terminal characteristic informations；To each cluster
The corresponding terminal push personalized product combination of middle terminal characteristic information, the personalized product combination include in the cluster one or
The corresponding product of multiple Product Feature Informations；
Terminal, for receiving and showing that personalized product combines.
About the specific implementation of personalized product combination supplying system, embodiment shown in Fig. 2 may refer to, herein no longer
It repeats.
The application also provides a kind of electronic equipment, comprising:
Processor；And
Memory, for storing the executable instruction of the processor；
Wherein, the processor is configured to execute personalized product shown in Fig. 2 via the executable instruction is executed
Combine method for pushing.The specific implementation that push amplification is combined about personalized product, may refer to embodiment shown in Fig. 2,
This is repeated no more.
The application also provides a kind of storage medium, and for the storage medium for storing software program, the software program is available
Personalized product described in realizing combines method for pushing.The specific implementation that push amplification is combined about personalized product, can be with
Embodiment shown in Figure 2, details are not described herein.
If function described in the present embodiment method is realized in the form of SFU software functional unit and as independent product pin
It sells or in use, can store in a storage medium readable by a compute device.Based on this understanding, the embodiment of the present application
The part of the part that contributes to existing technology or the technical solution can be embodied in the form of software products, this is soft
Part product is stored in a storage medium, including some instructions are used so that calculating equipment (it can be personal computer,
Server, mobile computing device or network equipment etc.) execute all or part of step of each embodiment the method for the application
Suddenly.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), deposits at random
The various media that can store program code such as access to memory (RAM, RandomAccess Memory), magnetic or disk.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with it is other
The difference of embodiment, same or similar part may refer to each other between each embodiment.
The foregoing description of the disclosed embodiments makes professional and technical personnel in the field can be realized or use the application.
Various modifications to these embodiments will be readily apparent to those skilled in the art, as defined herein
General Principle can be realized in other embodiments without departing from the spirit or scope of the application.Therefore, the application
It is not intended to be limited to the embodiments shown herein, and is to fit to and the principles and novel features disclosed herein phase one
The widest scope of cause.
Claims (10)
1. a kind of personalized product combines method for pushing characterized by comprising
Multiple terminal characteristic informations and multiple Product Feature Informations are extracted from historical trading data, building is special comprising multiple terminals
The characteristic information set of reference breath and multiple Product Feature Informations；Wherein, the characteristic dimension and production in the terminal characteristic information
Characteristic dimension in product characteristic information is identical；
Cluster operation is executed to the characteristic information set and obtains multiple clusters；Wherein each cluster includes at least one product feature letter
Breath, and, one or more terminal characteristic informations；
The corresponding terminal push personalized product combination of terminal characteristic information into each cluster, the personalized product, which combines, includes
The corresponding product of one or more Product Feature Information in the cluster.
2. the method as described in claim 1, which is characterized in that described to execute cluster operation acquisition to the characteristic information set
Multiple clusters, comprising:
K Product Feature Information is determined from the characteristic information set, respectively as the cluster center of K cluster；
It divides remaining characteristic information in the characteristic information set, to the smallest cluster of cluster centre distance, obtains K cluster；
Redefine the cluster center of K cluster；Wherein, the cluster center of each cluster is a Product Feature Information in the cluster；
Judge whether the cluster center of K cluster changes；If then entering step: it is special to divide remaining in the characteristic information set
Reference breath extremely and in the smallest cluster of cluster centre distance, obtains K cluster；
Determine that cluster operation terminates, and K cluster is determined as multiple clusters of cluster operation acquisition if not.
3. method according to claim 2, which is characterized in that the cluster center for redefining K cluster, comprising:
Following processes are executed for each cluster in K cluster:
The true cluster center for calculating cluster, calculating separately each Product Feature Information in cluster, will at a distance from the true cluster center
With the true the smallest Product Feature Information of cluster centre distance in cluster, it is determined as the cluster center of cluster.
4. method according to claim 2, which is characterized in that described to extract multiple terminal feature letters from historical trading data
Breath and multiple Product Feature Informations, building include the characteristic information collection of multiple terminal characteristic informations and multiple Product Feature Informations
It closes；Wherein, the characteristic dimension in the terminal characteristic information is identical as the characteristic dimension in Product Feature Information, comprising:
From historical trading data, extract and multiple terminal feature set and the extraction and multiple productions correspondingly of multiple terminals
Product multiple product feature set correspondingly；Wherein, in the characteristic dimension in terminal feature set and product feature set
Characteristic dimension is identical；
Determine the corresponding weight of each characteristic dimension；
To multiple terminal feature set and the multiple product feature set, vectorization operation and normalization operation are executed, is obtained
Multiple terminal characteristic informations and multiple Product Feature Informations；
By multiple terminal characteristic informations and multiple Product Feature Informations, it is determined as characteristic information set.
5. method as claimed in claim 4, which is characterized in that characteristic dimension and product feature set in terminal feature set
In characteristic dimension, comprising:
Assets class, transaction class, profit and loss class, interest class and ability class.
6. a kind of personalized product combines driving means characterized by comprising
Construction unit is constructed for extracting multiple terminal characteristic informations and multiple Product Feature Informations from historical trading data
Characteristic information set comprising multiple terminal characteristic informations and multiple Product Feature Informations；Wherein, in the terminal characteristic information
Characteristic dimension it is identical as the characteristic dimension in Product Feature Information；
Cluster cell obtains multiple clusters for executing cluster operation to the characteristic information set；Wherein each cluster includes at least
One Product Feature Information, and, one or more terminal characteristic informations；
Push unit, for the corresponding terminal push personalized product combination of terminal characteristic information, the individual character into each cluster
Changing product mix includes the corresponding product of one or more Product Feature Information in the cluster.
7. device as claimed in claim 6, which is characterized in that the cluster cell includes:
First determines cluster center cell, for determining K Product Feature Information from the characteristic information set, respectively as K
The cluster center of a cluster；
Division unit, for dividing remaining characteristic information in the characteristic information set, extremely and in the smallest cluster of cluster centre distance,
Obtain K cluster；
Second determines cluster center cell, for redefining the cluster center of K cluster；Wherein, the cluster center of each cluster is in the cluster
One Product Feature Information；
Judging unit, for judging whether the cluster center of K cluster changes；
Enter division unit if the cluster center of K cluster changes；
Determination unit, if the cluster center for K cluster does not change, it is determined that cluster operation terminates, and K cluster is determined as gathering
Multiple clusters that generic operation obtains.
8. a kind of personalized product combines supplying system characterized by comprising
Server, for extracting multiple terminal characteristic informations and multiple Product Feature Informations, building packet from historical trading data
Characteristic information set containing multiple terminal characteristic informations and multiple Product Feature Informations；Wherein, in the terminal characteristic information
Characteristic dimension is identical as the characteristic dimension in Product Feature Information；It is multiple that cluster operation acquisition is executed to the characteristic information set
Cluster；Wherein each cluster includes at least one Product Feature Information, and, one or more terminal characteristic informations；Eventually into each cluster
The corresponding terminal push personalized product combination of characteristic information is held, the personalized product combination includes one or more in the cluster
The corresponding product of Product Feature Information；
Terminal, for receiving and showing that personalized product combines.
9. a kind of electronic equipment characterized by comprising
Processor；And
Memory, for storing the executable instruction of the processor；
Wherein, the processor is configured to execute as described in any one of Claims 1 to 5 via the executable instruction is executed
Personalized product combine method for pushing.
10. a kind of storage medium, which is characterized in that for storing software program, which can be used for the storage medium
Realize personalized product combination method for pushing as claimed in any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811581395.3A CN109474703B (en) | 2018-12-24 | 2018-12-24 | Personalized product combination pushing method, device and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811581395.3A CN109474703B (en) | 2018-12-24 | 2018-12-24 | Personalized product combination pushing method, device and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109474703A true CN109474703A (en) | 2019-03-15 |
| CN109474703B CN109474703B (en) | 2021-08-31 |
Family
ID=65676783
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811581395.3A Active CN109474703B (en) | 2018-12-24 | 2018-12-24 | Personalized product combination pushing method, device and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109474703B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113112318A (en) * | 2020-01-10 | 2021-07-13 | 阿里巴巴集团控股有限公司 | Method, device, electronic equipment and computer storage medium for determining product combination |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080243815A1 (en) * | 2007-03-30 | 2008-10-02 | Chan James D | Cluster-based assessment of user interests |
| CN102254028A (en) * | 2011-07-22 | 2011-11-23 | 青岛理工大学 | Personalized commodity recommending method and system which integrate attributes and structural similarity |
| CN102609523A (en) * | 2012-02-10 | 2012-07-25 | 上海视畅信息科技有限公司 | Collaborative filtering recommendation algorithm based on article sorting and user sorting |
| CN102750647A (en) * | 2012-06-29 | 2012-10-24 | 南京大学 | Merchant recommendation method based on transaction network |
| US20140006552A1 (en) * | 2012-06-27 | 2014-01-02 | Ubiquiti Networks, Inc. | Method and apparatus for maintaining network connections between devices |
| CN103886003A (en) * | 2013-09-22 | 2014-06-25 | 天津思博科科技发展有限公司 | Collaborative filtering processor |
| CN105869001A (en) * | 2015-01-19 | 2016-08-17 | 苏宁云商集团股份有限公司 | Customized product recommendation guiding method and system |
| CN105931066A (en) * | 2015-09-24 | 2016-09-07 | 中国银联股份有限公司 | Transaction data processing method and device |
| CN106951489A (en) * | 2017-03-13 | 2017-07-14 | 杭州师范大学 | A kind of personalized recommendation method and device for sparse big data |
| CN107657500A (en) * | 2016-12-03 | 2018-02-02 | 平安证券股份有限公司 | Stock recommends method and server |
-
2018
- 2018-12-24 CN CN201811581395.3A patent/CN109474703B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080243815A1 (en) * | 2007-03-30 | 2008-10-02 | Chan James D | Cluster-based assessment of user interests |
| CN102254028A (en) * | 2011-07-22 | 2011-11-23 | 青岛理工大学 | Personalized commodity recommending method and system which integrate attributes and structural similarity |
| CN102609523A (en) * | 2012-02-10 | 2012-07-25 | 上海视畅信息科技有限公司 | Collaborative filtering recommendation algorithm based on article sorting and user sorting |
| US20140006552A1 (en) * | 2012-06-27 | 2014-01-02 | Ubiquiti Networks, Inc. | Method and apparatus for maintaining network connections between devices |
| CN102750647A (en) * | 2012-06-29 | 2012-10-24 | 南京大学 | Merchant recommendation method based on transaction network |
| CN103886003A (en) * | 2013-09-22 | 2014-06-25 | 天津思博科科技发展有限公司 | Collaborative filtering processor |
| CN105869001A (en) * | 2015-01-19 | 2016-08-17 | 苏宁云商集团股份有限公司 | Customized product recommendation guiding method and system |
| CN105931066A (en) * | 2015-09-24 | 2016-09-07 | 中国银联股份有限公司 | Transaction data processing method and device |
| CN107657500A (en) * | 2016-12-03 | 2018-02-02 | 平安证券股份有限公司 | Stock recommends method and server |
| CN106951489A (en) * | 2017-03-13 | 2017-07-14 | 杭州师范大学 | A kind of personalized recommendation method and device for sparse big data |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113112318A (en) * | 2020-01-10 | 2021-07-13 | 阿里巴巴集团控股有限公司 | Method, device, electronic equipment and computer storage medium for determining product combination |
| CN113112318B (en) * | 2020-01-10 | 2022-04-29 | 阿里巴巴集团控股有限公司 | Method, device, electronic equipment and computer storage medium for determining product combination |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109474703B (en) | 2021-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2018086401A1 (en) | Cluster processing method and device for questions in automatic question and answering system | |
| CN111859960A (en) | Semantic matching method and device based on knowledge distillation, computer equipment and medium | |
| CN110472154B (en) | Resource pushing method and device, electronic equipment and readable storage medium | |
| CN110032583B (en) | Fraudulent party identification method and device, readable storage medium and terminal equipment | |
| CN109784474A (en) | A kind of deep learning model compression method, apparatus, storage medium and terminal device | |
| CN104391879B (en) | The method and device of hierarchical clustering | |
| CN103455555B (en) | Recommendation method and recommendation apparatus based on mobile terminal similarity | |
| CN110689136B (en) | Deep learning model obtaining method, device, equipment and storage medium | |
| CN108509793A (en) | A kind of user's anomaly detection method and device based on User action log data | |
| US11935049B2 (en) | Graph data processing method and apparatus, computer device, and storage medium | |
| CN110347724A (en) | Abnormal behaviour recognition methods, device, electronic equipment and medium | |
| CN112466314A (en) | Emotion voice data conversion method and device, computer equipment and storage medium | |
| CN110223095A (en) | Determine the method, apparatus, equipment and storage medium of item property | |
| CN109474703A (en) | Personalized product combines method for pushing, apparatus and system | |
| CN110443633A (en) | Matching process, device, computer equipment and the storage medium of excited data | |
| CN109308637A (en) | Entity consumption scene oriented marketing method, device and system | |
| CN115187345A (en) | Intelligent household building material recommendation method, device, equipment and storage medium | |
| CN106570756A (en) | Identification method and device of business objects | |
| CN112071331A (en) | Voice file repairing method and device, computer equipment and storage medium | |
| CN112215441A (en) | Prediction model training method and system | |
| CN113205217B (en) | Data processing method, device, equipment and storage medium | |
| Singh | Banking 4.0-era of innovation | |
| CN113782033B (en) | Voiceprint recognition method, voiceprint recognition device, voiceprint recognition equipment and storage medium | |
| CN113535887B (en) | Formula similarity detection method and device | |
| CN111105020B (en) | Feature representation migration learning method and related device |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20210811 Address after: Room 323, 3 / F, Hang Seng building, 3588 Jiangnan Avenue, Binjiang District, Hangzhou City, Zhejiang Province 310053 Applicant after: ZHEJIANG JINGTENG NETWORK TECHNOLOGY Co.,Ltd. Address before: 310053 8th floor, building 2, No. 3588 Jiangnan Avenue, Binjiang District, Hangzhou, Zhejiang Applicant before: HANGZHOU YUNJI NETWORK TECHNOLOGY Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |