WO2019174395A1

WO2019174395A1 - Method and apparatus for information recommendation, and device

Info

Publication number: WO2019174395A1
Application number: PCT/CN2019/071880
Authority: WO
Inventors: 刘海旭
Original assignee: 阿里巴巴集团控股有限公司
Priority date: 2018-03-13
Filing date: 2019-01-16
Publication date: 2019-09-19
Also published as: CN113837804A; CN108596645B; TW201939404A; TWI777015B; CN108596645A

Abstract

A method and apparatus for information recommendation, and a device. In the method, static information and dynamic information can be obtained, wherein the static information comprises fixed information of users and/or fixed information corresponding to service information, and the dynamic information comprises records of operations performed on historical service information by each user and/or at least one piece of attribute information of the environment where the user is currently located (S100); the obtained static information can be input to the Wide linear model in pre-trained Wide&Deep models, and the obtained dynamic information can be input to the Deep learning model in the Wide&Deep models, so as to obtain the output results of the Wide&Deep models (S102); and then according to the output results, determining service information to be recommended to the user, and recommending the service information to the user (S104).

Description

Method, device and device for information recommendation

Technical field

The present specification relates to the field of computer technology, and in particular, to a method, device and device for information recommendation.

Background technique

In order to provide users with better service, at present, many service providers can push various service information, such as advertisements and coupons, to users, so that users can perform more purposeful consumption according to the service information pushed by the service provider. .

In practical applications, the server usually recommends the service information such as advertisements, coupons, and the like related to the current location of the user according to the current location of the user. However, recommending service information to users based on a single dimension of geographic location may not be able to meet the actual needs of users.

Based on the prior art, a more accurate way of recommending information is needed.

Summary of the invention

The present specification provides a method for recommending information to solve the problem that the information recommendation method of the prior art cannot accurately and effectively recommend the required information to the user.

This manual provides a method of information recommendation, including:

Obtaining static information including dynamic information of the user and/or fixed information corresponding to each service information, the dynamic information including an operation record of each historical service information being operated by each user and/or the user At least one of environmental attribute information of an environment in which the current time is located;

Inputting the static information into a Wide linear model in a pre-trained Wide&Deep model, and inputting the dynamic information into a Deep Deep Learning Model in a pre-trained Wide&Deep model to obtain an output result of the Wide&Deep model;

Based on the output result, service information recommended to the user is determined and recommended to the user.

The present specification provides a device for recommending information to solve the problem that the information recommendation method of the prior art cannot accurately and effectively recommend the required information to the user.

This specification provides a means of information recommendation, including:

The information acquiring module acquires static information and dynamic information, where the static information includes fixed information of the user and/or fixed information corresponding to each service information, and the dynamic information includes an operation record of each historical service information being operated by each user and/or Or at least one of environmental attribute information of an environment in which the user is currently located;

Input module, inputting the static information into a Wide linear model in a pre-trained Wide&Deep model, and inputting the dynamic information into a Deep Deep Learning Model in a pre-trained Wide&Deep model to obtain an output of the Wide&Deep model result;

The information recommendation module determines service information recommended to the user according to the output result, and recommends to the user.

The present specification provides an information recommendation device comprising one or more memories and a processor that stores a program and is configured to perform the following steps by the one or more processors:

The above at least one technical solution adopted in the present specification can achieve the following beneficial effects:

In one or more embodiments of the present specification, static information and dynamic information may be acquired, where the static information includes fixed information of the user and/or fixed information corresponding to each service information, and the dynamic information includes each historical service information by each user. At least one of an operation record of performing an operation and/or environmental attribute information of an environment in which the user is currently at a time. Then, the obtained static information can be input into the Wide linear model in the pre-trained Wide&Deep model, and the acquired dynamic information is input into the Deep deep learning model in the Wide&Deep model to obtain the input result of the Wide&Deep model. Further, based on the output result, the service information recommended to the user is determined and recommended to the user.

It can be seen from the above method that static information such as fixed information of the user, fixed information corresponding to each service information, and environmental attribute information of the environment in which the user is currently located, and each historical service information can be operated by each user. The dynamic information such as the operation record is combined to determine the service information recommended to the user, and therefore, the manner of recommending the service information to the user is determined with respect to the information of the user through the single dimension, so that the user can be more accurately recommended to the user. Service information, which brings great convenience to users.

DRAWINGS

The drawings described herein are provided to provide a further understanding of the specification, and are intended to be a part of this specification. In the drawing:

FIG. 1 is a schematic diagram of a process of information recommendation provided by the present specification;

2 is a schematic diagram of processing static information and dynamic information acquired by a server through the Wide&Deep model according to the present specification, and obtaining an output result;

Figure 3 is a schematic diagram of the device recommended by the information provided in this specification;

FIG. 4 is a schematic diagram of an apparatus for information recommendation provided by the present specification.

detailed description

In order to make those skilled in the art better understand the technical solutions in one or more embodiments of the present specification, in the following one or more embodiments of the present specification, in one or more embodiments of the present specification, The technical solutions are described clearly and completely, and it is obvious that the described embodiments are only a part of the embodiments of the specification, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present specification without departing from the scope of the invention shall fall within the scope of the present disclosure.

In the present specification, the execution subject of the above-mentioned information recommendation method may be a server or a terminal device, and for convenience of description, the method of information recommendation provided in this specification will be described below only by using the server as the execution subject.

FIG. 1 is a schematic diagram of a process of information recommendation provided by the present specification, which specifically includes the following steps:

S100: Acquire static information and dynamic information, where the static information includes fixed information of the user and/or fixed information corresponding to each service information, where the dynamic information includes an operation record and/or a place where each historical service information is operated by each user. At least one of environmental attribute information of an environment in which the user is currently located.

In this specification, the server can combine the information of multiple dimensions to determine the service information required by the user, and recommend it to the user for viewing. The information of the various dimensions mentioned here can be roughly divided into two categories, one can be attributed to static information, and the other can be attributed to dynamic information.

The static information mentioned here may be information that can effectively indicate the inherent characteristics of the user or the service information, and does not change frequently. For example, for a user's age, education, gender, address, etc., the time interval for change is usually long, and the information can effectively characterize some inherent characteristics of the user. For another example, for the service information, the area where the service provider that issued the service information is located, and the effective time of the service information (generally, once the service information is released, the effective time will not change), and the service is released. The number of stores corresponding to the service provider of the information, the change period of the information such as the merchant size of the service provider is usually long, and therefore, the fixed information corresponding to the service information can be reduced to static information.

In this specification, the user's fixed information can effectively reflect the user's actual needs, preferences, and information that may be of interest to a certain extent. For example, for users of different ages, the information they pay attention to often differs greatly. The younger users tend to pay more attention to some fashionable topics, while the older users tend to pay more attention to some health topics. For another example, the degree of education of a user can often reflect the income status of the user to a certain extent. For users with lower academic qualifications, the income status is often poor. Correspondingly, users with higher education tend to earn income. Higher. Users with lower incomes may pay more attention to the promotion of some products, while users with higher incomes may pay more attention to the information of some high-end products.

It can be seen from the above examples that the fixed information of the user can reflect the actual needs of the user and the information of some types of information to a certain extent. Therefore, the fixed information of the user can be used as static information to determine the recommended information to the user. What kind of service information. In the present specification, the user's fixed information may include information such as a service category, a credit, a member's credit, and the like of a service that the user has performed, in addition to the several examples mentioned above, and will not be exemplified herein.

In the same way, in practical applications, what kind of service information users prefer, is often related to some inherent characteristics of the service information itself. For example, for service information recommended by some well-known service providers, users tend to pay more attention; for example, coupons recommended by different service providers to users are often displayed in different forms, and some service providers display coupon templates that are novel and beautiful. The coupons they display are more popular, and correspondingly, the number of users receiving coupons is also high. However, some service providers display coupon templates that are more common and cannot effectively attract users' attention. Therefore, the popularity is low, and the number of users who receive coupons is relatively small.

Therefore, in the present specification, the fixed information corresponding to each service information can be used as static information to determine which service information is more popular with the user, and thus recommended to the user. The fixed information corresponding to each service information mentioned herein may be other forms of information in addition to the above examples, for example, the service category corresponding to each service information (for example, for each restaurant, some restaurants are provided to the user. It offers coupons for Western food, while others offer coupons for Chinese food. Therefore, whether the coupons are Western or Chinese, it can be called the service category of the coupon. For example, the mall launched The coupons for purchasing apparel and the coupons for purchasing electronic products belong to different categories. Therefore, the coupons can be provided with clothing discounts or electronic product discounts, which are called service categories of coupons. The attention of the service provider of the service information, etc., will not be described in detail here.

The above-mentioned dynamic information may refer to information that the change cycle is relatively short and can have a certain influence on the actual needs of the user. For example, when users view the coupons launched by each merchant at the current moment, they are usually only interested in coupons launched by merchants that are closer to their current location. For example, when users are viewing overseas coupons for overseas scenic spots, they often combine the current season (relative to the user's education, age, etc., the season change period is relatively short), and choose to travel in the current season. Coupon for the scenic area.

Therefore, in the present specification, the environment attribute information of the environment in which the user is currently located can be used as the dynamic information, and based on this, the service information that meets the actual needs of the user is determined and recommended to the user. The environment attribute information of the environment in which the user is currently located may include: weather information at the current moment, the location of the current moment of the user, and the distance between the service providers.

Of course, in addition to the several kinds of information exemplified above, the environment attribute information mentioned herein may also be other forms of information, such as the current month, season, holiday information, etc., which are not illustrated in detail herein.

For an operation record in which each history service information is operated by each user, the period of change of the operation record in the past period of time is usually short, and through the operation record, the user's sense of each service information can be understood sideways. The level of interest. For example, suppose that a user receives a dining coupon provided by 10 restaurants through a mobile phone, but only the dining coupons provided by four restaurants are actually used, and the remaining dining coupons received are not used. Therefore, from this level, it can be effectively concluded that the user may be more acceptable to the dining environment, dishes, and per capita consumption of the four restaurants, and the degree of interest is higher, and for the other six restaurants, The level of interest is relatively low.

Based on this, in the present specification, an operation record characterizing the operation status of the user operation history service information may be used to determine which service information is more popular with the user and enter the recommendation to the user. The operation record mentioned herein may include: the number of times the history service information is performed by the user, the number of times the history service information is used by the user, the number of times the history service information is performed by the user, and the like. For example, the number of times the coupon issued by the service provider is clicked by the user, the number of times the user is picked up, the number of times the user is used, and the like. The number of times that each historical service information mentioned herein is browsed by the user may refer to the total number of times each historical service information is viewed by each user in the past period of time, or may refer to each historical service information in the past. The number of times viewed in unit time (eg, one hour, one day, etc.).

Similarly, the number of times the historical service information is used by the user to perform the use operation (or the selection operation) may also refer to the total number of times each historical service information is used by each user in the past operation period (or the selection operation), or The number of times the usage operation (or selection operation) was performed in each unit time in the past.

It should be noted that, in this specification, for the fixed information of the user, the server may determine the fixed information of the user according to the user account registered by the user. Similarly, for the fixed information corresponding to each service information, the server may also determine the account information of each service provider.

For the environment attribute information of the environment in which the user is currently located, the server may determine that the current time meets the preset trigger condition, and obtain the environment attribute information. For example, when the server detects that the user logs in through the user account of the user, it can determine that the current time meets the preset trigger condition, and then triggers obtaining the environment attribute information of the environment where the user is currently located; for example, when the server detects When the current time reaches the preset time, it is determined that the current time meets the preset trigger condition, and then the environment attribute information of the environment in which the user is currently located is obtained.

For the operation record mentioned above, the server can record the operation status of the user operation service information at all times, and save it in the corresponding operation log, in the subsequent process, determine the operation to the user through the operation record in the saved operation log. Recommended service information. For example, when the user views the coupon sent by the server through the terminal, the server may record the operation performed by the user on the coupon, and the recorded content includes: whether the user clicks to browse the detailed preferential information of the coupon, and whether the user has received the coupon. Coupon, whether the coupon is used, etc.

Before determining the fixed information corresponding to each service information, the server may select some candidate service information by using a preset screening manner, and then further determine the fixed information corresponding to the candidate service information, and in the subsequent process. The fixed information determined is input into the pre-trained Wide&Deep model. Among them, there are many ways to filter the alternative service information mentioned here. For example, the server can determine the location where the user is currently at the moment, and then use the coupons provided by the service providers within the set distance from the location as Alternative service information; for example, the server may determine the number of coupons currently provided by each service provider, and then select the coupons that have been received by the set number as the alternative service information.

S102: input the static information into a Wide linear model in a pre-trained Wide&Deep model, and input the dynamic information into a Deep Deep Learning Model in a pre-trained Wide&Deep model to obtain an output result of the Wide&Deep model. .

In this specification, the server can determine which service information needs to be recommended to the user through the pre-trained Wide&Deep model. Among them, a feature of the Wide&Deep model is that it has both memory and generalization capabilities. The so-called memory ability refers to the ability to estimate data that is strongly correlated with historical data through historical data. For example, the server recommends the same type of coupon to the user by the type of coupon that the user has received in the past. Generalization refers to the ability to predict new data that has never been seen before, through historical data and data correlation. For example, the server analyzes that the user may be interested in other types of coupons through the types of coupons that the user has received in the past, and then recommends to the user. Among them, other types mentioned here may be types that have never appeared before.

Through the Wide&Deep model, it is possible to further enhance the diversity of service information in order to ensure that the user's favorite service information is recommended to the user, so as to provide the user with more choices, thereby bringing a good user experience to the user.

After the server determines the static information and the dynamic information described above, the static information may be input into the Wide linear model in the Wide&Deep model, and the dynamic information is input into the Deep deep learning model in the Wide&Deep model to The output of the Wide&Deep model is obtained, as shown in Figure 2.

FIG. 2 is a schematic diagram of processing the static information and the dynamic information acquired by the server through the Wide&Deep model provided by the present specification to obtain an output result.

The server can input the obtained static information and dynamic information into the Wide linear model and the Deep deep learning model in the Wide&Deep model, respectively, and then the results of the two models can be weighted and summed, and then weighted and summed. The obtained result is input into a preset loss function, and finally the output result of the Wide&Deep model is obtained. Among them, the loss function mentioned here has no specific limitation.

The static information is input into the Wide linear model in order to comprehensively analyze which (or which type of) service information the user is generally interested in through the static information and the memory capabilities of the Wide model. The above dynamic information is input into the Deep Deep Learning Model in order to enable the dynamic information and the generalization ability of the Deep Deep Learning Model to comprehensively analyze which (which kind of) service information the user may have. interest. In other words, in the present specification, it is considered that the above-mentioned several types of dynamic information have generalization conditions that enable the Deep Deep Learning Model to perform generalization analysis.

The output of the above Wide&Deep model can be in the form of a recommended rating. Specifically, after the server inputs the static information and the dynamic information into the Wide&Deep model, the Wide&Deep model can obtain the recommended scores for each service information by processing the information, so as to determine which needs to be obtained through the obtained recommendation scores. Service information is recommended to this user. The service information mentioned herein may be the service information provided by all the service providers at present time, or may be the candidate service information mentioned above.

It should be noted that the Wide&Deep model needs to be trained before using the above Wide&Deep model. The server can obtain historical data and split the acquired historical data into training samples and verification samples. The Wide&Deep model can be trained through the training sample, and then the trained Wide&Deep model is verified by the verification sample.

The historical data mentioned above may refer to: fixed information corresponding to a plurality of historical service information published in the past history, personal information of a plurality of users, environmental attribute information of the environment in which the plurality of users are located in the past history, and the plurality of historical service information. A history operation record that is operated by each user, service information actually selected by the plurality of users, and the like. Among them, several users mentioned here can be randomly selected by the server, and there is no specific limitation.

Correspondingly, the server may perform fixed information corresponding to a plurality of historical service information published in the past history, personal information of a plurality of users, environmental attribute information of the environment in which the plurality of users are located in the past history, and the plurality of historical service information are executed by each user. The historical operation of the operation records the information as a training sample, and the service information actually selected by the plurality of users is used as a verification sample to train the Wide&Deep model.

From the time dimension, the verification samples selected by the server may refer to historical data after the release time is set, and the training samples may refer to historical data before the set time. For example, suppose that the server can record the fixed information corresponding to each coupon provided by each service provider in March of the year A, the operation record of each coupon operated by N users in March A, and N users in the middle of March of the year A. The environmental attribute information of the environment and the fixed information of the N users are used as training samples. At the same time, the server can use the coupons actually used by the N users after March A as a verification sample to train the Wide&Deep model. The set time is the March of the year A mentioned here. In this way, the purpose of predicting future data through historical data is realized, thereby improving the usability of the service information recommended by the server to the user.

In this specification, the information recommendation ability of the Wide&Deep model can be evaluated by a certain evaluation method, and when the Wide&Deep model reaches the preset training target by the evaluation method, the Wide&Deep model is launched. Among them, there are various evaluation methods mentioned here. For example, the accuracy of the recommended service information of the Wide&Deep model can be tested by the training samples and verification samples mentioned above, and the accuracy of the recommended service information of the Wide&Deep model is reached. When setting the accuracy rate, it is determined that the Wide&Deep model has reached the preset training target. Of course, the Wide&Deep model can also be evaluated by an evaluation method such as a commonly used AUC model evaluation index, and will not be exemplified here.

S104: Determine, according to the output result, service information recommended to the user, and recommend the information to the user.

After obtaining the output result, the server can determine which service information needs to be recommended to the user through the output result, and then recommend the determined service information to the user for browsing and operation.

Specifically, after determining the recommended score of each service information by using the Wide&Deep model, the server may recommend the service information not less than the set recommendation score to the user. Of course, other ways can be used to determine which service information is recommended to the user. For example, after the server determines the recommendation score of each service information, each service information may be sorted according to each recommendation score, and the service information sorted before the set position is recommended to the user.

It should be noted that, in this specification, the server may use the information of the period (that is, the historical data mentioned above as the training sample and the verification sample) as the training sample and the verification sample, and the Wide&Deep model. The training is carried out, and the trained Wide&Deep model is updated online in time to ensure the accuracy of the recommended service information of the Wide&Deep model through this automatic learning method. For example, every 5 days, the server can use the information of the first 4 days of the 5 days as a training sample, and the information of the 5th day as a verification sample to train the Wide&Deep model so that the trained Wide&Deep model can be as Meet the needs of users to obtain service information in the near future.

The above is a method for recommending information provided by one or more embodiments of the present specification. Based on the same idea, the present specification further provides a device for recommending information, as shown in FIG. 3 .

FIG. 3 is a schematic diagram of a device for information recommendation provided by the present specification, which specifically includes:

The information obtaining module 301 acquires static information and dynamic information, where the static information includes fixed information of the user and/or fixed information corresponding to each service information, where the dynamic information includes an operation record of each historical service information being operated by each user. / or at least one of environmental attribute information of the environment in which the user is currently located;

The input module 302 inputs the static information into the Wide linear model in the pre-trained Wide&Deep model, and inputs the dynamic information into the Deep depth learning model in the pre-trained Wide&Deep model to obtain the Wide&Deep model. Output result

The information recommendation module 403 determines service information recommended to the user according to the output result, and recommends to the user.

The fixed information of the user includes: at least one of age, education, gender, address of the user, and a service category of the user performing the service;

The fixed information corresponding to each service information includes: a region where each service provider that issues the service information, an effective time of each service information, a number of stores corresponding to each service provider, and each service provider At least one of the corresponding merchant sizes;

The environment attribute information of the environment in which the user is located at the current time includes: at least one of weather information of the current time, a location of the user at the current time, and a distance between the service providers;

The operation record includes at least one of the number of times each history service information is browsed by each user, the number of times each history service information is used by the user, and the number of times each history service information is selected by the user.

The output result of the Wide&Deep model includes: a recommendation score obtained by the Wide&Deep model for each service information;

The information recommendation module 303 recommends, to the user, service information that is not less than a set recommendation score.

The device also includes:

The training module 304 splits the historical data into training samples and verification samples; trains the Wide&Deep model through the training samples, and verifies the trained Wide&Deep model through the verification samples.

The training module 304 uses historical data before the set time as a training sample, and takes historical data after the set time as a verification sample.

The service information includes: a coupon.

Based on the above-described method of information recommendation, the present specification also provides a device for information recommendation, as shown in FIG. The device includes one or more memories and a processor that stores the program and is configured to perform the following steps by the one or more processors:

In one or more embodiments of the present specification, static information and dynamic information may be acquired, where the static information includes fixed information of the user and/or fixed information corresponding to each service information, and the dynamic information includes each historical service information The operation record of the user performing the operation and/or at least one of the environment attribute information of the environment in which the user is currently at the moment. Then, the obtained static information can be input into the Wide linear model in the pre-trained Wide&Deep model, and the acquired dynamic information is input into the Deep deep learning model in the Wide&Deep model to obtain the input result of the Wide&Deep model. Further, based on the output result, the service information recommended to the user is determined and recommended to the user.

In the 1990s, improvements to a technology could clearly distinguish between hardware improvements (eg, improvements to circuit structures such as diodes, transistors, switches, etc.) or software improvements (for process flow improvements). However, as technology advances, many of today's method flow improvements can be seen as direct improvements in hardware circuit architecture. Designers almost always get the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that the improvement of a method flow cannot be implemented by hardware entity modules. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is an integrated circuit whose logic function is determined by the user programming the device. Designers program themselves to "integrate" a digital system on a single PLD without having to ask the chip manufacturer to design and fabricate a dedicated integrated circuit chip. Moreover, today, instead of manually making integrated circuit chips, this programming is mostly implemented using "logic compiler" software, which is similar to the software compiler used in programming development, but before compiling The original code has to be written in a specific programming language. This is called the Hardware Description Language (HDL). HDL is not the only one, but there are many kinds, such as ABEL (Advanced Boolean Expression Language). AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc., are currently the most commonly used VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. It should also be apparent to those skilled in the art that the hardware flow for implementing the logic method flow can be easily obtained by simply programming the method flow into the integrated circuit with a few hardware description languages.

The controller can be implemented in any suitable manner, for example, the controller can take the form of, for example, a microprocessor or processor and a computer readable medium storing computer readable program code (eg, software or firmware) executable by the (micro)processor. In the form of logic gates, switches, application specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers, examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, The Microchip PIC18F26K20 and the Silicone Labs C8051F320, the memory controller can also be implemented as part of the memory's control logic. Those skilled in the art will also appreciate that in addition to implementing the controller in purely computer readable program code, the controller can be logically programmed by means of logic gates, switches, ASICs, programmable logic controllers, and embedding. The form of a microcontroller or the like to achieve the same function. Such a controller can therefore be considered a hardware component, and the means for implementing various functions included therein can also be considered as a structure within the hardware component. Or even a device for implementing various functions can be considered as a software module that can be both a method of implementation and a structure within a hardware component.

The system, device, module or unit illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product having a certain function. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or A combination of any of these devices.

For the convenience of description, the above devices are described separately by function into various units. Of course, the functions of the various units may be implemented in one or more software and/or hardware in the implementation of the present specification.

Those skilled in the art will appreciate that embodiments of the present specification can be provided as a method, system, or computer program product. Accordingly, the description may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment in combination of software and hardware. Moreover, the description may take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) including computer usable program code.

This description is described with reference to flowchart illustrations and/or block diagrams of a method, a device (system), and a computer program product according to one or more embodiments of the present specification. It will be understood that each flow and/or block of the flowchart illustrations and/or FIG. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing device to produce a machine for the execution of instructions for execution by a processor of a computer or other programmable data processing device. Means for implementing the functions specified in one or more of the flow or in a block or blocks of the flow chart.

The computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture comprising the instruction device. The apparatus implements the functions specified in one or more blocks of a flow or a flow and/or block diagram of the flowchart.

These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device. The instructions provide steps for implementing the functions specified in one or more of the flow or in a block or blocks of a flow diagram.

In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include non-persistent memory, random access memory (RAM), and/or non-volatile memory in a computer readable medium, such as read only memory (ROM) or flash memory. Memory is an example of a computer readable medium.

Computer readable media includes both permanent and non-persistent, removable and non-removable media. Information storage can be implemented by any method or technology. The information can be computer readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read only memory. (ROM), electrically erasable programmable read only memory (EEPROM), flash memory or other memory technology, compact disk read only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, Magnetic tape cartridges, magnetic tape storage or other magnetic storage devices or any other non-transportable media can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary storage of computer readable media, such as modulated data signals and carrier waves.

It is also to be understood that the terms "comprises" or "comprising" or "comprising" or any other variations are intended to encompass a non-exclusive inclusion, such that a process, method, article, Other elements not explicitly listed, or elements that are inherent to such a process, method, commodity, or equipment. An element defined by the phrase "comprising a ..." does not exclude the presence of additional equivalent elements in the process, method, item, or device including the element.

This description can be described in the general context of computer-executable instructions executed by a computer, such as a program module. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform particular tasks or implement particular abstract data types. One or more embodiments of the present specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are connected through a communication network. In a distributed computing environment, program modules can be located in both local and remote computer storage media including storage devices.

The various embodiments in the specification are described in a progressive manner, and the same or similar parts between the various embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiment.

The foregoing description of the specific embodiments of the specification has been described. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order than the embodiments and still achieve the desired results. In addition, the processes depicted in the figures are not necessarily in a particular order or in a sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

The above description is only one or more embodiments of the present specification, and is not intended to limit the present specification. One or more embodiments of the present disclosure can be variously changed and changed by those skilled in the art. Any modifications, equivalents, improvements, etc. made within the spirit and scope of the present invention are intended to be included within the scope of the appended claims.

Claims

A method of information recommendation, including:

Obtaining static information including dynamic information of the user and/or fixed information corresponding to each service information, the dynamic information including an operation record of each historical service information being operated by each user and/or the user At least one of environmental attribute information of an environment in which the current time is located;

Inputting the static information into a Wide linear model in a pre-trained Wide&Deep model, and inputting the dynamic information into a Deep Deep Learning Model in a pre-trained Wide&Deep model to obtain an output result of the Wide&Deep model;

Based on the output result, service information recommended to the user is determined and recommended to the user.
The method according to claim 1, wherein the fixed information of the user comprises: at least one of age, education, gender, address of the user, and a service category of the user performing the service;

The fixed information corresponding to each service information includes: a region where each service provider that issues the service information, an effective time of each service information, a number of stores corresponding to each service provider, and each service provider At least one of the corresponding merchant sizes;

The environment attribute information of the environment in which the user is located at the current time includes: at least one of weather information of the current time, a location of the user at the current time, and a distance between the service providers;

The operation record includes at least one of the number of times each history service information is browsed by each user, the number of times each history service information is used by the user, and the number of times each history service information is selected by the user.
The method of claim 1, the output of the Wide&Deep model comprising: a recommendation score obtained by the Wide&Deep model for each service information;

Determining the service information recommended to the user according to the output result, specifically:

The service information not less than the set recommendation score is recommended to the user.
The method of claim 1, training the Wide&Deep model, specifically comprising:

Split historical data into training samples and verification samples;

The Wide&Deep model is trained by the training sample, and the trained Wide&Deep model is verified by the verification sample.
The method of claim 4, the historical data is divided into a training sample and a verification sample, specifically comprising:

The historical data located before the set time is taken as a training sample, and the historical data located after the set time is taken as a verification sample.
The method according to any one of claims 1 to 5, wherein the service information comprises: a coupon.
A device for information recommendation, comprising:

The information acquiring module acquires static information and dynamic information, where the static information includes fixed information of the user and/or fixed information corresponding to each service information, and the dynamic information includes an operation record of each historical service information being operated by each user and/or Or at least one of environmental attribute information of an environment in which the user is currently located;

Input module, inputting the static information into a Wide linear model in a pre-trained Wide&Deep model, and inputting the dynamic information into a Deep Deep Learning Model in a pre-trained Wide&Deep model to obtain an output of the Wide&Deep model result;

The information recommendation module determines service information recommended to the user according to the output result, and recommends to the user.
The apparatus according to claim 7, wherein the fixed information of the user comprises: at least one of age, education, gender, address of the user, and a service category in which the user performs a service;

The fixed information corresponding to each service information includes: a region where each service provider that issues the service information, an effective time of each service information, a number of stores corresponding to each service provider, and each service provider At least one of the corresponding merchant sizes;

The environment attribute information of the environment in which the user is located at the current time includes: at least one of weather information of the current time, a location of the user at the current time, and a distance between the service providers;

The operation record includes at least one of the number of times each history service information is browsed by each user, the number of times each history service information is used by the user, and the number of times each history service information is selected by the user.
The apparatus according to claim 7, wherein the output of the Wide&Deep model comprises: a recommendation score obtained by the Wide&Deep model for each service information;

The information recommendation module recommends, to the user, service information that is not less than a set recommendation score.
The device of claim 7 further comprising:

The training module splits the historical data into a training sample and a verification sample; trains the Wide&Deep model through the training sample, and verifies the trained Wide&Deep model through the verification sample.
The apparatus according to claim 10, wherein said training module uses historical data located before the set time as a training sample, and uses historical data located after said set time as a verification sample.
The apparatus according to any one of claims 7 to 11, wherein the service information comprises: a coupon.
An information recommendation device, the device comprising one or more memories and a processor, the memory storing a program and configured to perform the following steps by the one or more processors:

Obtaining static information including dynamic information of the user and/or fixed information corresponding to each service information, the dynamic information including an operation record of each historical service information being operated by each user and/or the user At least one of environmental attribute information of an environment in which the current time is located;

Inputting the static information into a Wide linear model in a pre-trained Wide&Deep model, and inputting the dynamic information into a Deep Deep Learning Model in a pre-trained Wide&Deep model to obtain an output result of the Wide&Deep model;

Based on the output result, service information recommended to the user is determined and recommended to the user.