CN109190044B - Personalized recommendation method, device, server and medium - Google Patents
Personalized recommendation method, device, server and medium Download PDFInfo
- Publication number
- CN109190044B CN109190044B CN201811052567.8A CN201811052567A CN109190044B CN 109190044 B CN109190044 B CN 109190044B CN 201811052567 A CN201811052567 A CN 201811052567A CN 109190044 B CN109190044 B CN 109190044B
- Authority
- CN
- China
- Prior art keywords
- user
- intention
- behavior
- recommendation
- demand
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention discloses a personalized recommendation method, a personalized recommendation device, a server and a medium, wherein the method comprises the following steps: predicting a demand user group with demands for the recommendation object from an original user group by utilizing a pre-trained demand intention model according to the user portrait; predicting the behavior intention of each user by utilizing a pre-trained intention recognition model and the user portrait of each user in the demand user group; and matching the behavior intention of each user with the attribute characteristics of the recommended object, and performing personalized recommendation for each user according to the matching result. The embodiment of the invention solves the problem of low personalized recommendation precision for the user, improves the recommendation precision, further improves the response rate of the recommended user, and improves the conversion rate of goods or services.
Description
Technical Field
The embodiment of the invention relates to the technical field of computers, in particular to a personalized recommendation method, a personalized recommendation device, a server and a medium.
Background
A user representation is a tagged user model that is abstracted based on information such as the user's social attributes, lifestyle, and consumption behavior. Through user portrait, thereby can carry out user statistics, analysis potential user and realize accurate marketing, can also carry out big data mining and analysis based on user portrait to perfect the product operation, promote quality of service. In the large environment of the internet, personalized recommendation based on user portraits is applied in many fields.
In the prior art, the content of interest of a user is generally analyzed based on a user portrait, and then relevant goods or services are recommended for the user. However, since the user representation is obtained according to the historical network data of the user, the obtained content that the user is interested in is only the content that the user is interested in once, thereby affecting the accuracy of personalized recommendation.
Disclosure of Invention
The embodiment of the invention provides a personalized recommendation method, a personalized recommendation device, a server and a medium, and aims to achieve the effect of improving the accuracy of personalized recommendation.
In a first aspect, an embodiment of the present invention provides a personalized recommendation method, where the method includes:
predicting a demand user group with demands for the recommendation object from an original user group by utilizing a pre-trained demand intention model according to the user portrait;
predicting the behavior intention of each user by utilizing a pre-trained intention recognition model and the user portrait of each user in the demand user group;
and matching the behavior intention of each user with the attribute characteristics of the recommended object, and performing personalized recommendation for each user according to the matching result.
In a second aspect, an embodiment of the present invention further provides a personalized recommendation apparatus, where the apparatus includes:
the demand user group prediction module is used for predicting a demand user group with demand for the recommendation object from an original user group by utilizing a pre-trained demand intention model according to the user portrait;
the behavior intention prediction module is used for predicting the behavior intention of each user by utilizing a pre-trained intention recognition model and the user portrait of each user in the demand user group;
and the recommending module is used for matching the behavior intention of each user with the attribute characteristics of the recommended object and performing personalized recommendation for each user according to the matching result.
In a third aspect, an embodiment of the present invention further provides a server, including:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement a personalized recommendation method as described in any of the embodiments of the invention.
In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements a personalized recommendation method according to any embodiment of the present invention.
According to the embodiment of the invention, a demand user group with demands for a recommended object is predicted by utilizing a pre-trained demand intention model according to the user portrait; then, predicting the behavior intention of each user by using a pre-trained intention recognition model and the user portrait of each user in the demand user group; and finally, matching the behavior intention of each user with the attribute characteristics of the recommended object, and performing personalized recommendation for each user according to the matching result. According to the embodiment of the invention, through the recommendation processing process based on the dual model, the problem of low personalized recommendation accuracy for the user is solved, the recommendation accuracy is improved, the response rate of the recommended user is further improved, and the conversion rate of goods or services is improved.
Drawings
Fig. 1 is a flowchart of a personalized recommendation method according to an embodiment of the present invention;
fig. 2 is a flowchart of a personalized recommendation method according to a second embodiment of the present invention;
fig. 3 is a schematic structural diagram of a personalized recommendation device according to a third embodiment of the present invention;
fig. 4 is a schematic structural diagram of a server according to a fourth embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
Example one
Fig. 1 is a flowchart of a personalized recommendation method according to an embodiment of the present invention, where the embodiment is applicable to a case of performing personalized recommendation for a user, and the method may be executed by a personalized recommendation device, and the device may be implemented in a software and/or hardware manner and may be integrated on a server. As shown in fig. 1, the method may include:
s110, predicting a demand user group with demand for the recommendation object from the original user group by utilizing a pre-trained demand intention model according to the user portrait.
The user representation is a tagged representation of user information. And predicting a demand user group aiming at the recommendation object from the original user group by using a demand intention model, namely judging whether the user has recommended qualification or not, and realizing primary screening of the user group.
Optionally, the user representation includes at least natural attributes, social attributes, interest preferences, address location information, and behavioral attributes of the user.
The natural attributes of the user, namely the personal information of the user, including sex, age range, age, constellation and the like; the social attributes of the users, namely the social tags to which the users belong, comprise life stages, asset conditions, industries, education levels, occupation categories, income levels, consumption levels, quality consumption levels, consumption willingness, marital conditions, business retail attributes, terminal application lists, equipment information and the like; the interest preference comprises long-term interest and short-term interest of the user and interest labels covering various industries; the address location information is the location information of the user, including the frequently-visited city list, the location longitude and latitude, the address, the building name and the like of the user, and the information can be obtained by positioning; the behavior attribute is a behavior related to the network operation of the user, and includes a living habit of surfing the internet, a normal activity coordinate, a real-time action scene and the like of the user, such as an online shopping behavior, an online payment behavior, a web browsing behavior and the like. The more extensive the user attribute categories involved in the user representation, the more user information is included.
Optionally, the training process of the demand-intention model includes:
acquiring a user portrait of a user who has a demand on a recommended object from historical data;
two classifiers are obtained by utilizing user portrait training as a demand intention model.
In the process of training the demand intention model, a sample data set of model training, namely data information of a historical demand user, needs to be acquired based on a data statistics technology, then the user portrait of the historical demand user is used as input, and the demand intention model is obtained through training. The two classifiers available in the training process include but are not limited to: logistic regression, SVM (support vector machine), GBDT (gradient boosting tree), or DNN (deep neural network).
And S120, predicting the behavior intention of each user by using the pre-trained intention recognition model and the user portrait of each user in the demand user group.
At least one user behavior intention is included in the user behavior intentions predicted in real time according to the user portrait, and the predicted behavior intention is changed along with the updating of the user portrait.
For example, according to the behavior attributes in the user representation, clustering analysis is performed on the user behavior characteristics of the user A in a recent month, and it is found that the user frequently uses the online shopping mall C to purchase cosmetics in the month, so that the behavior intention of the user in a future period can be predicted to include online shopping mall C shopping, cosmetic consumption and the like. Through behavior intention prediction, the basis of the recommendation process is no longer only the historical interests of the user.
Optionally, the training process of the intention recognition model includes:
acquiring at least one intention label of different users from historical data;
and training by using the user portrait of each user and at least one corresponding intention label to obtain a multi-classifier as an intention recognition model.
Similar to the process of training the demand intention model, in the process of training the intention recognition model, a sample data set of model training, namely an intention label and a user portrait of a user in network historical data, needs to be obtained based on a data statistics technology, and the intention recognition model is obtained through training. Wherein, the user portrait may correspond to multiple intention labels simultaneously, and the multiple classifiers available in the model training process include but are not limited to: softmax classifiers, decision trees, random forests, GBDTs (gradient boosting trees) or DNNs (deep neural networks), etc.
And S130, matching the behavior intention of each user with the attribute characteristics of the recommendation object, and performing personalized recommendation for each user according to the matching result.
When the behavior intention of the user is matched with the attribute characteristics of the recommendation object, the recommendation object can be recommended to the user. And if the behavior intention of the user comprises a plurality of intentions, recommending the recommendation object which is simultaneously matched with the plurality of intentions of the user to the user. The higher the matching degree between the user behavior intention and the attribute characteristics of the recommended object is, the higher the requirement of the user on the recommended object is, and the higher the recommendation accuracy is at the moment.
Specifically, the recommendation object may be any item related to a clothing and eating behavior, such as an audio/video, a book, a garment, a household article, an office article, an electronic device, various consumer cards, and the like, and the attribute feature may be representative information for representing the recommendation object.
Illustratively, the recommendation object belongs to a credit card, which includes three equity features in the attribute features of city X: fast food D preferential package, online shopping mall C shopping full activity 1 and pastry E consumption full activity 2; predicting, via the intent recognition model, behavioral intent of user A in Y City for a future period of time includes: shopping, cosmetic consumption and book consumption in the online shopping mall C; predicting, via the intent recognition model, behavioral intent of user B in city X over a future period of time includes: and (3) shopping in the online shopping mall C, fast food D consumption and order E consumption, and if the matching degree between the credit card equity characteristics and the behavior intention of the user B is higher, recommending the credit card to the user B, and correspondingly, according to the consumption requirement of the user B, the probability of handling the credit card is higher, namely the response rate of the recommended user is improved, and the conversion rate of credit card recommendation is improved.
According to the technical scheme, firstly, a pre-trained demand intention model is used for preliminarily screening a demand user group of a recommended object according to a user portrait, then, a pre-trained intention recognition model and a user portrait of each user in the demand user group are used for predicting the behavior intention of each user in real time, and finally, personalized recommendation is carried out on each user according to the matching result between the behavior intention of the user and the attribute characteristics of the recommended object. Namely, the problem of low personalized recommendation accuracy for the user is solved through the dual-model-based recommendation processing process in the embodiment, the recommendation accuracy is improved, the response rate of the recommended user is further improved, the conversion rate of goods or services is improved, the popularization and marketing costs can be saved, and the calculation efficiency is very high in the dual-model recommendation processing process based on machine learning.
Example two
Fig. 2 is a flowchart of a personalized recommendation method provided in the second embodiment of the present invention, and the second embodiment is further optimized based on the foregoing embodiments. As shown in fig. 2, the method may include:
s210, predicting a demand user group with demand for the recommendation object from the original user group by using a pre-trained demand intention model according to the user portrait.
And S220, predicting the behavior intention of each user by utilizing the pre-trained intention recognition model and the user portrait of each user in the demand user group, wherein the intention recognition model is a multi-classification model, the predicted behavior intention of each user comprises at least one intention arranged according to the size of the prediction score, and the higher the prediction score is, the stronger the intention is.
That is, the output results of the intent recognition model include the behavioral intent of the user and the predicted scores of the behavioral intent. The predicted score of the behavior intention can be determined according to the degree of association between each intention and the user image, and the greater the association between the behavior intention and the user image, the higher the corresponding predicted score, the stronger the intention of the user for executing the behavior. Illustratively, a user representation over a period of time includes the following feature tags: european, white collar, fashion, after 90 and cosmetic consumption, through behavioral prediction, the user behavioral intent over a future period of time includes: the corresponding prediction scores are 45%, 37%, 13% and 5% respectively, which shows that the intention of a user for online purchasing cosmetics is stronger and the occurrence probability is higher in the future period, while the intention of travel is weakest and the occurrence probability is lowest.
And S230, extracting behavior characteristics according to the prediction score corresponding to each behavior intention in the behavior intentions of each user, and representing the behavior intention of each user as a behavior intention vector according to the behavior characteristics.
The user's behavior intention vector is composed of features for characterizing the user's behavior intention over a future period of time, each vector element representing a behavior intention. By representing the behavior intention of the user in a vector form, the subsequent calculation of the matching degree between the behavior intention of the user and the recommended object can be facilitated.
Illustratively, the sequencing result of the predicted user behavior intentions is maintained, and the behavior intentions with the score larger than or equal to the prediction threshold value are extracted according to the prediction score of each intention, and are represented as 1, while the behavior intentions smaller than the prediction threshold value are represented as 0, so that the behavior intention vector of the user is obtained. Continuing with the above example as an example, the obtained user behavior intention and the prediction score are respectively: if 45% of the on-line cosmetics, 37% of the nail consumption, 13% of the clothing shopping, and 5% of the travel, which are kept in this order, and the behavior intention with the prediction score of more than 20% is represented as 1, and the rest is represented as 0, the behavior intention vector of the user may be represented as [1,1, 0,0 ]. Of course, on the basis of keeping the behavior intention ordering, the prediction score of the behavior intention can also be directly used as an element of the behavior intention vector, that is, the behavior intention vector of the user can also be expressed as [ 45%, 37%, 13%, 5% ]. On the basis of ensuring that the behavior intention of the user can be distinguished and represented, the embodiment does not specifically limit the representation form of the elements in the behavior intention vector.
S240, obtaining the attribute features of the recommended object, and representing the attribute features of the recommended object as attribute feature vectors according to at least one feature vector of the attribute features.
Wherein the length of the attribute feature vector is related to the attribute feature of the object type to which the recommended object belongs.
In an example, a recommended object belongs to a video, and the attribute features of the video include the following five features: high definition, fashion, makeup, teaching and brand, the eigenvector that each attribute characteristic corresponds is in proper order: [1,0,0,0,0], [0,1,0,0,0], [0,0,1,0,0], [0,0,0,1,0], and [0,0,0,0,1], the non-zero number 1 indicates that the video in the type has an attribute feature corresponding to the element position. The attribute features of the recommendable video 1 on the video website are high definition, fashion and make-up, and then the video can be represented as an attribute feature vector [1,1,1,0,0 ]; the recommendable video 2 on the video website has attribute features of fashion, make-up, and brand, then the video may be represented as an attribute feature vector [0,1,1,0,1 ].
Example two, the recommended object is a credit card of a certain business, and the credit card has four attribute features, which are: fast food package preference, electronic mall shopping full-reduced activity, nine-fold dining consumption and installment preference, the corresponding feature vectors are [1,0,0,0], [0,1,0,0], [0,0,1,0] and [0,0,0,1], respectively, and a non-zero number 1 indicates that the credit card has the attribute feature corresponding to the element position. The attribute characteristics of credit cards of this type are slightly different in different regions, for example in X1, and the specific attribute characteristics of the credit card are: the credit card in the X1 market can be represented as an attribute feature vector [0,1,1,1] according to the feature vector of the attribute feature if the electronic mall purchases full-subtractive activities, has nine discount for eating consumption and pays preferential benefit by stages for cosmetic consumption; in X2, the credit card has the following specific attributes: fast food package offers, nine folds for meal consumption and cosmetics consumption installment offers, then the credit card in X2 may be represented as attribute feature vector [1,0,1,1] according to the feature vector of attribute features.
And S250, calculating the similarity between the behavior intention vector and the attribute feature vector, and performing personalized recommendation according to the similarity value.
In the similarity calculation process, the lengths of the behavior intention vector of the user and the attribute feature vector of the recommendation object need to be the same, and if the lengths of the two vectors are different, the two vectors can be transformed based on the vectors so that the two vectors have the same length. The specific vector similarity calculation method comprises cosine similarity, jaccard (Jacard) similarity, Euclidean distance similarity, Mahalanobis distance and the like. The higher the calculated similarity value is, the higher the matching degree between the behavior intention of the user and the attribute characteristics of the recommendation object is. When the number of the recommendation objects is multiple, the recommendation object corresponding to the highest similarity score can be recommended to the user, or the recommendation objects ranked in the front by the preset number can be recommended to the user by ranking according to the similarity scores from high to low.
Continuing with the above example as an example, when the recommendation object belongs to the video category, through similarity calculation, the matching degree between the recommendable video 2 and the behavior intention of the user is the highest, and the recommendable video is recommended for 1 time, so that the videos 2 and 1 can be recommended to the user in sequence according to the ranking result of the similarity scores. Or, when the recommendation object belongs to a credit card, the similarity calculation shows that the matching degree between the credit card in the city of X1 and the behavior intention of the user is the highest, and the credit card in the city of X1 is recommended to the user.
Optionally, the recommendation object is a credit card, and the attribute characteristics of the recommendation object include multiple rights and interests characteristics of different credit cards;
correspondingly, personalized recommendation is performed according to the similarity score, and the personalized recommendation method comprises the following steps:
acquiring a credit score model of each user;
and determining a target user of the recommended object according to the credit score model from the users corresponding to the similarity scores meeting the preset conditions, and recommending the recommended object to the target user.
Wherein the interest characteristic of the credit card is personalized service for distinguishing the credit card, and the credit score model is used for evaluating the credit degree of the user. And after further screening the users in the required user group according to the similarity between the behavior intention vector of the user and the attribute feature vector of the recommended object, carrying out user screening again based on the user credit in the users corresponding to the similarity score meeting the preset condition, namely the users corresponding to the similarity threshold value or more, and determining the users with the credit greater than or equal to the credit threshold value as the target users of the recommended object. Through layer-by-layer screening, only when the user meets all the screening conditions, the recommendation information is received, accurate recommendation of the credit card is guaranteed, the recommendation conversion rate of the credit card is improved, and further useless work of the user can be avoided. In addition, through accurate recommendation of the credit card, the popularization and marketing cost of the credit card can be saved.
According to the technical scheme, firstly, a demand user group of a recommendation object is preliminarily screened out according to the user portrait, then, the behavior intention of each user is predicted by utilizing a pre-trained intention recognition model and the user portrait of each user in the demand user group, the behavior intention vector is expressed, secondly, the attribute feature of the recommendation object is obtained and expressed as the attribute feature vector, and finally, personalized recommendation is carried out according to the similarity score between the behavior intention vector and the attribute feature vector. According to the technical scheme, the problem of low personalized recommendation accuracy for the user is solved, the recommendation accuracy is improved, the response rate of the recommended user is improved, the conversion rate of goods or services is improved, and the popularization and marketing cost is saved.
EXAMPLE III
Fig. 3 is a schematic structural diagram of a personalized recommendation device according to a third embodiment of the present invention, which is applicable to a case of performing personalized recommendation for a user. The device can be realized in a software and/or hardware mode and can be configured on the server.
As shown in fig. 3, the apparatus includes a demand user group prediction module 310, a behavior intention prediction module 320, and a recommendation module 330, wherein:
a demand user group prediction module 310, configured to predict, according to the user representation, a demand user group having a demand for the recommended object from an original user group by using a pre-trained demand intention model;
a behavior intent prediction module 320 for predicting a behavior intent of each user using a pre-trained intent recognition model and a user profile of each user in the group of desired users;
and the recommending module 330 is configured to match the behavior intention of each user with the attribute features of the recommendation object, and perform personalized recommendation for each user according to a matching result.
Optionally, the user representation includes at least natural attributes, social attributes, interest preferences, address location information, and behavioral attributes of the user.
Optionally, the intention recognition model is a multi-classification model, and the predicted behavior intention of each user includes at least one intention arranged according to the size of the prediction score, wherein a higher prediction score indicates a stronger intention;
accordingly, the recommendation module 330 includes:
the behavior intention vector representing unit is used for extracting behavior characteristics according to the prediction score corresponding to each behavior intention in the behavior intentions of each user and representing the behavior intention of each user as a behavior intention vector according to the behavior characteristics;
the attribute feature vector representing unit is used for acquiring the attribute features of the recommended object and representing the attribute features of the recommended object as attribute feature vectors according to the feature vector of at least one attribute feature in the attribute features;
and the recommending unit is used for calculating the similarity between the behavior intention vector and the attribute feature vector and carrying out personalized recommendation according to the similarity value.
Optionally, the recommendation object is a credit card, and the attribute characteristics of the recommendation object include multiple rights and interests characteristics of different credit cards;
correspondingly, the recommending unit is specifically configured to:
acquiring a credit score model of each user;
and determining a target user of the recommended object according to the credit score model from the users corresponding to the similarity scores meeting the preset conditions, and recommending the recommended object to the target user.
Optionally, the apparatus further includes a requirement intention model training module, configured to:
acquiring a user portrait of a user who has a demand on a recommended object from historical data;
two classifiers are obtained by utilizing user portrait training as a demand intention model.
Optionally, the apparatus further includes an intention recognition model training module, configured to:
acquiring at least one intention label of different users from historical data;
and training by using the user portrait of each user and at least one corresponding intention label to obtain a multi-classifier as an intention recognition model.
The personalized recommendation device provided by the embodiment of the invention can execute the personalized recommendation method provided by any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method. The technical details not elaborated in this embodiment can be explained with reference to the content of any method embodiment of the invention.
Example four
Fig. 4 is a schematic structural diagram of a server according to a fourth embodiment of the present invention. FIG. 4 illustrates a block diagram of an exemplary server 412 suitable for use in implementing embodiments of the present invention. The server 412 shown in fig. 4 is only an example and should not bring any limitations to the function and scope of use of the embodiments of the present invention.
As shown in FIG. 4, server 412 is in the form of a general purpose server. Components of server 412 may include, but are not limited to: one or more processors 416, a storage device 428, and a bus 418 that couples the various system components including the storage device 428 and the processors 416.
A program/utility 440 having a set (at least one) of program modules 442 may be stored, for instance, in storage 428, such program modules 442 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which examples or some combination thereof may comprise an implementation of a network environment. The program modules 442 generally perform the functions and/or methodologies of the described embodiments of the invention.
The server 412 may also communicate with one or more external devices 414 (e.g., keyboard, pointing terminal, display 424, etc.), with one or more terminals that enable a user to interact with the server 412, and/or with any terminals (e.g., network card, modem, etc.) that enable the server 412 to communicate with one or more other computing terminals. Such communication may occur via input/output (I/O) interfaces 422. Further, server 412 may communicate with one or more networks (e.g., a Local Area Network (LAN), Wide Area Network (WAN), and/or a public Network, such as the Internet) via Network adapter 420. As shown in FIG. 4, network adapter 420 communicates with the other modules of server 412 via bus 418. It should be appreciated that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the server 412, including but not limited to: microcode, end drives, Redundant processors, external disk drive Arrays, RAID (Redundant Arrays of Independent Disks) systems, tape drives, and data backup storage systems, among others.
The processor 416 executes various functional applications and data processing by executing programs stored in the storage device 428, for example, implementing a personalized recommendation method provided by any embodiment of the present invention, and the method may include:
predicting a demand user group with demands for the recommendation object from an original user group by utilizing a pre-trained demand intention model according to the user portrait;
predicting the behavior intention of each user by utilizing a pre-trained intention recognition model and the user portrait of each user in the demand user group;
and matching the behavior intention of each user with the attribute characteristics of the recommended object, and performing personalized recommendation for each user according to the matching result.
EXAMPLE five
An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements a personalized recommendation method according to any embodiment of the present invention, where the method may include:
predicting a demand user group with demands for the recommendation object from an original user group by utilizing a pre-trained demand intention model according to the user portrait;
predicting the behavior intention of each user by utilizing a pre-trained intention recognition model and the user portrait of each user in the demand user group;
and matching the behavior intention of each user with the attribute characteristics of the recommended object, and performing personalized recommendation for each user according to the matching result.
Computer storage media for embodiments of the invention may employ any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or terminal. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (12)
1. A method for personalized recommendation, comprising:
predicting a demand user group with demands for the recommendation object from an original user group by utilizing a pre-trained demand intention model according to the user portrait;
predicting the behavior intention of each user by utilizing a pre-trained intention recognition model and the user portrait of each user in the demand user group;
matching the behavior intention of each user with the attribute characteristics of the recommended object, and performing personalized recommendation for each user according to the matching result;
the intention identification model is a multi-classification model, and the predicted behavior intention of each user comprises at least one intention which is arranged according to the size of a prediction score, wherein the higher the prediction score is, the stronger the intention is;
correspondingly, the matching the behavior intention of each user with the attribute features of the recommendation object, and performing personalized recommendation for each user according to the matching result includes:
extracting behavior characteristics according to the prediction score corresponding to each behavior intention in the behavior intentions of each user, and representing the behavior intention of each user as a behavior intention vector according to the behavior characteristics;
acquiring attribute features of the recommended object, and representing the attribute features of the recommended object as attribute feature vectors according to at least one feature vector of the attribute features;
and calculating the similarity between the behavior intention vector and the attribute feature vector, and performing personalized recommendation according to the similarity value.
2. The method of claim 1, wherein the user representation includes at least natural attributes, social attributes, interest preferences, address location information, and behavioral attributes of the user.
3. The method of claim 1, wherein the recommendation object is a credit card, and the attribute features of the recommendation object include multiple entitlement features of different credit cards;
correspondingly, the personalized recommendation according to the similarity score includes:
acquiring a credit score model of each user;
and determining a target user of the recommended object according to the credit score model from the users corresponding to the similarity scores meeting the preset conditions, and recommending the recommended object to the target user.
4. The method of claim 1, wherein the training process of the demand-intent model comprises:
acquiring a user portrait of a user who has a demand on the recommended object from historical data;
and training by utilizing the user portrait to obtain a two-classifier which is used as the demand intention model.
5. The method of claim 1, wherein the training process of the intent recognition model comprises:
acquiring at least one intention label of different users from historical data;
and training a plurality of classifiers by using the user portrait of each user and at least one corresponding intention label as the intention recognition model.
6. A personalized recommendation device, comprising:
the demand user group prediction module is used for predicting a demand user group with demand for the recommendation object from an original user group by utilizing a pre-trained demand intention model according to the user portrait;
the behavior intention prediction module is used for predicting the behavior intention of each user by utilizing a pre-trained intention recognition model and the user portrait of each user in the demand user group;
the recommendation module is used for matching the behavior intention of each user with the attribute characteristics of the recommendation object and carrying out personalized recommendation on each user according to the matching result;
the intention recognition model is a multi-classification model, and the predicted behavior intention of each user comprises at least one intention which is arranged according to the size of a prediction score, wherein the higher the prediction score is, the stronger the intention is;
correspondingly, the recommendation module comprises:
the behavior intention vector representing unit is used for extracting behavior characteristics according to the prediction score corresponding to each behavior intention in the behavior intentions of each user and representing the behavior intention of each user as a behavior intention vector according to the behavior characteristics;
the attribute feature vector representing unit is used for acquiring the attribute features of the recommended object and representing the attribute features of the recommended object as attribute feature vectors according to the feature vector of at least one attribute feature in the attribute features;
and the recommending unit is used for calculating the similarity between the behavior intention vector and the attribute feature vector and carrying out personalized recommendation according to the similarity value.
7. The apparatus of claim 6, wherein the user representation includes at least natural attributes, social attributes, interest preferences, address location information, and behavioral attributes of the user.
8. The apparatus of claim 6, wherein the recommendation object is a credit card, and the attribute feature of the recommendation object comprises multiple entitlement features of different credit cards;
correspondingly, the recommending unit is specifically configured to:
acquiring a credit score model of each user;
and determining a target user of the recommended object according to the credit score model from the users corresponding to the similarity scores meeting the preset conditions, and recommending the recommended object to the target user.
9. The apparatus of claim 6, further comprising a demand intent model training module to:
acquiring a user portrait of a user who has a demand on the recommended object from historical data;
and training by utilizing the user portrait to obtain a two-classifier which is used as the demand intention model.
10. The apparatus of claim 6, further comprising an intent recognition model training module to:
acquiring at least one intention label of different users from historical data;
and training a plurality of classifiers by using the user portrait of each user and at least one corresponding intention label as the intention recognition model.
11. A server, comprising:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement the personalized recommendation method of any of claims 1-5.
12. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out a personalized recommendation method according to any one of claims 1 to 5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811052567.8A CN109190044B (en) | 2018-09-10 | 2018-09-10 | Personalized recommendation method, device, server and medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811052567.8A CN109190044B (en) | 2018-09-10 | 2018-09-10 | Personalized recommendation method, device, server and medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109190044A CN109190044A (en) | 2019-01-11 |
CN109190044B true CN109190044B (en) | 2020-11-20 |
Family
ID=64915644
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811052567.8A Active CN109190044B (en) | 2018-09-10 | 2018-09-10 | Personalized recommendation method, device, server and medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109190044B (en) |
Families Citing this family (28)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109711950A (en) * | 2019-01-18 | 2019-05-03 | 深圳壹账通智能科技有限公司 | Recommended method, device, storage medium and the computer equipment of marketing product |
CN111666503B (en) * | 2019-03-05 | 2023-08-18 | 百度在线网络技术(北京)有限公司 | Travel mode recommendation method and device, computer equipment and readable storage medium |
CN111695960A (en) * | 2019-03-12 | 2020-09-22 | 阿里巴巴集团控股有限公司 | Object recommendation system, method, electronic device and storage medium |
CN111695962B (en) * | 2019-03-13 | 2023-04-25 | 阿里巴巴集团控股有限公司 | Cloud product recommendation method and device, computing equipment and storage medium |
CN110135912B (en) * | 2019-05-17 | 2022-05-13 | 北京百度网讯科技有限公司 | Information pushing method and device, server and storage medium |
CN110177152B (en) * | 2019-06-11 | 2022-03-29 | 秒针信息技术有限公司 | Information pushing method and device |
CN110348892A (en) * | 2019-06-27 | 2019-10-18 | 上海淇馥信息技术有限公司 | Customized information generation method, device and electronic equipment based on user characteristics |
CN110363582A (en) * | 2019-06-29 | 2019-10-22 | 上海淇馥信息技术有限公司 | Financial service favor information promotion method, device and electronic equipment based on user intention |
CN110489639B (en) * | 2019-07-15 | 2022-10-25 | 北京奇艺世纪科技有限公司 | Content recommendation method and device |
CN110532309B (en) * | 2019-07-15 | 2022-05-03 | 浙江工业大学 | Generation method of college library user portrait system |
CN110443639A (en) * | 2019-07-16 | 2019-11-12 | 深圳市数位汇聚科技有限公司 | A kind of marketing strategy recommended method and device based on position positioning under line |
CN110413888B (en) * | 2019-07-24 | 2024-05-10 | 腾讯科技(深圳)有限公司 | Book recommendation method and device |
CN110489642B (en) * | 2019-07-25 | 2020-05-22 | 山东大学 | Commodity recommendation method, system, equipment and medium based on behavior feature analysis |
CN110688573A (en) * | 2019-09-24 | 2020-01-14 | 支付宝(杭州)信息技术有限公司 | Method and device for recommending issue order in credit project |
CN112860980A (en) * | 2019-11-12 | 2021-05-28 | 阿里巴巴集团控股有限公司 | Content presentation and recommendation processing method and device, storage medium and electronic equipment |
CN111079013B (en) * | 2019-12-13 | 2023-12-29 | 北京金山数字娱乐科技有限公司 | Information recommendation method and device based on recommendation model |
CN111310009A (en) * | 2020-01-16 | 2020-06-19 | 珠海格力电器股份有限公司 | User classification method and device, storage medium and computer equipment |
CN111292168B (en) * | 2020-02-06 | 2022-08-02 | 腾讯科技(深圳)有限公司 | Data processing method, device and equipment |
CN111428994B (en) * | 2020-03-20 | 2022-06-03 | 支付宝(杭州)信息技术有限公司 | Service processing method and device and electronic equipment |
CN111445308A (en) * | 2020-03-26 | 2020-07-24 | 四川旅游学院 | Tourist attraction recommendation method based on user portrait |
CN113240049B (en) * | 2020-12-18 | 2022-03-25 | 龙马智芯(珠海横琴)科技有限公司 | Intention recognition method, device, server and computer readable storage medium |
CN112598047A (en) * | 2020-12-18 | 2021-04-02 | 龙马智芯(珠海横琴)科技有限公司 | Intention recognition model generation method and device, server and readable storage medium |
CN112782982A (en) * | 2020-12-31 | 2021-05-11 | 海南大学 | Intent-driven essential computation-oriented programmable intelligent control method and system |
CN112818227B (en) * | 2021-01-29 | 2024-03-19 | 北京百度网讯科技有限公司 | Content recommendation method and device, electronic equipment and storage medium |
CN112905899B (en) * | 2021-03-31 | 2024-06-07 | 中国工商银行股份有限公司 | Intelligent recommending method and device for entity card information |
CN113436001A (en) * | 2021-06-25 | 2021-09-24 | 中国工商银行股份有限公司 | Credit card pushing method, device, equipment, storage medium and program product |
CN113722371B (en) * | 2021-08-31 | 2024-04-12 | 深圳平安智慧医健科技有限公司 | Medicine recommendation method, device, equipment and storage medium based on decision tree |
CN117407571B (en) * | 2023-12-14 | 2024-02-09 | 天津畅玩网络科技有限公司 | Information technology consultation service method and system based on correlation analysis |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105956146A (en) * | 2016-05-12 | 2016-09-21 | 腾讯科技(深圳)有限公司 | Article information recommending method and device |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2014026152A2 (en) * | 2012-08-10 | 2014-02-13 | Assurerx Health, Inc. | Systems and methods for pharmacogenomic decision support in psychiatry |
CN103412882B (en) * | 2013-07-18 | 2016-11-23 | 百度在线网络技术(北京)有限公司 | A kind of method and device identifying that consumption is intended to |
WO2018000210A1 (en) * | 2016-06-28 | 2018-01-04 | 深圳狗尾草智能科技有限公司 | User portrait-based skill package recommendation device and method |
CN107403335A (en) * | 2017-06-19 | 2017-11-28 | 北京至信普林科技有限公司 | A kind of drawn a portrait based on depth user carries out the system and implementation method of precision marketing |
CN107451217A (en) * | 2017-07-17 | 2017-12-08 | 广州特道信息科技有限公司 | Information recommends method and device |
CN107491486A (en) * | 2017-07-17 | 2017-12-19 | 广州特道信息科技有限公司 | User's portrait construction method and device |
CN107909473A (en) * | 2017-12-27 | 2018-04-13 | 中国银行股份有限公司 | A kind of Web bank's marketing method and device based on user behavior analysis |
-
2018
- 2018-09-10 CN CN201811052567.8A patent/CN109190044B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105956146A (en) * | 2016-05-12 | 2016-09-21 | 腾讯科技(深圳)有限公司 | Article information recommending method and device |
Non-Patent Citations (1)
Title |
---|
一种知识驱动的个性化文献推荐方法;郐弘智 等;《计算机应用研究》;20171212;第35卷(第12期);3603-3608 * |
Also Published As
Publication number | Publication date |
---|---|
CN109190044A (en) | 2019-01-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109190044B (en) | Personalized recommendation method, device, server and medium | |
KR20200123015A (en) | Information recommendation method, apparatus, device and medium | |
JP6134444B2 (en) | Method and system for recommending information | |
Miao et al. | Context‐based dynamic pricing with online clustering | |
US8650141B2 (en) | System and method of segmenting and tagging entities based on profile matching using a multi-media survey | |
US10354184B1 (en) | Joint modeling of user behavior | |
US11243992B2 (en) | System and method for information recommendation | |
US11004123B2 (en) | Method for trading goods on a computer network based on image processing for interest, emotion and affinity detection | |
US20240095490A1 (en) | Aspect Pre-selection using Machine Learning | |
JP6527275B1 (en) | Harmonious search method based on harmony of multiple objects in image, computer apparatus and computer program | |
CN111783039B (en) | Risk determination method, risk determination device, computer system and storage medium | |
CN112085568B (en) | Commodity and rich media aggregation display method and equipment, electronic equipment and medium | |
CN111292168B (en) | Data processing method, device and equipment | |
CN113327151A (en) | Commodity object recommendation method and device, computer equipment and storage medium | |
CN113111264B (en) | Interface content display method and device, electronic equipment and storage medium | |
CN116089745A (en) | Information recommendation method, device, electronic equipment and computer readable storage medium | |
CN113327132A (en) | Multimedia recommendation method, device, equipment and storage medium | |
CN113254824B (en) | Content determination method, device, medium, and program product | |
US11120476B2 (en) | Systems and methods for generating personalized advertisements | |
CN114492313A (en) | Encoder training method, resource recommendation method and device | |
CN113947431A (en) | User behavior quality evaluation method, device, equipment and storage medium | |
US20230099904A1 (en) | Machine learning model prediction of interest in an object | |
US20230349710A1 (en) | Method, computer device, and non-transitory computer-readable recording medium for providing optimal path | |
CN116049531A (en) | Method and device for determining associated application, and method and device for determining recommended content | |
CN117036024A (en) | Financial object recommendation method, device, equipment and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |