Recommendation methods, devices, electronic devices, and storage media for objects.

Abstract

The embodiments of the present invention propose an object recommendation method, device, electronic device and storage medium. The object recommendation method of the embodiments of the present invention includes: obtaining user data authorized by the user; determining the short-term interest characteristics of the current user based on the behavior information of the current user; determining a user set and an object set similar to the current user based on the user information and historical object information; determining the long-term interest characteristics of the current user based on the user set and the object set; inputting the short-term interest characteristics and the long-term interest characteristics into a recommendation model to determine a recommended object, the recommendation model being pre-trained based on multiple user data; and sending the recommended object. The embodiments of the present invention can improve the accuracy of the recommended object by determining the recommended object based on two dimensions: short-term interest characteristics and long-term interest characteristics.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of computer technology, and in particular, to an object recommendation method, device, electronic device and storage medium. Background Art

[0002] As people's living standards continue to improve, cars have become a common means of transportation for people. However, due to the long service life and high price of cars, people buy cars less frequently, resulting in less data related to user-car transactions. For most users, cars are purchased only 1-2 times. In the process of predicting the user's purchase intention, it is impossible to predict the user's related purchase intention based on the user's only purchase information. It is impossible to accurately push the car that the user wants to the user, resulting in the user being unable to efficiently obtain the required car information. Therefore, it is necessary to provide a more accurate recommendation method to improve the accuracy of car recommendations. Summary of the invention

[0003] The embodiment of the present invention proposes a new technical solution based on the above-mentioned problems, by determining the short-term interest characteristics of the current user based on the behavior information of the current user, determining the long-term interest characteristics of the user based on the historical object information of multiple user sets similar to the current user, and then determining the recommended objects based on the short-term interest characteristics, the long-term interest characteristics and the recommendation model, thereby facilitating improving the accuracy of the recommended objects.

[0004] In view of this, according to a first aspect of an embodiment of the present invention, a method for recommending an object is proposed, and the method for recommending an object includes:

[0005] Obtaining user data authorized by the user, wherein the user data includes user information, historical object information, and related object information;

[0006] Determining the short-term interest characteristics of the current user according to the behavior information of the current user;

[0007] Determine a user set and an object set similar to the current user according to the user information and the historical object information, wherein the user set includes users with similar attribute information, and the object set includes multiple objects similar to the historical objects;

[0008] Determining the long-term interest characteristics of the current user according to the user set and the object set;

[0009] Inputting the short-term interest feature and the long-term interest feature into a recommendation model to determine a recommendation object, wherein the recommendation model is pre-trained based on a plurality of user data; and

[0010] The recommended object is sent.

[0011] Preferably, determining the short-term interest characteristics of the current user according to the user's behavior information includes:

[0012] A graph embedding algorithm is used to determine the short-term interest characteristics of the user based on the behavior information.

[0013] Preferably, determining the long-term interest feature of the current user according to the historical object information of similar users in the similar user set includes:

[0014] A graph embedding algorithm is used to determine the long-term interest characteristics of the user based on the user set and the object set.

[0015] Preferably, determining a user set and an object set similar to the current user according to the user information and the historical object information comprises:

[0016] Using a clustering algorithm to determine a set of similar users with similar user information; and

[0017] A clustering algorithm is used to determine a set of similar objects that are similar to the historical object information.

[0018] Preferably, the recommendation model takes short-term interest features and long-term interest features as inputs and historical object information as outputs, and is trained using a logistic regression algorithm or an extreme gradient boosting algorithm.

[0019] Preferably, before obtaining the user data authorized by the user, the object recommendation method further includes:

[0020] Preprocessing user information and historical object information of multiple users and saving them to a first database; and

[0021] Preprocessing the user information and behavior information of the current user and saving them to the second database;

[0022] The user data for obtaining user authorization includes:

[0023] Acquire user information and historical object information of multiple users from the first database; and

[0024] The user information and behavior information of the current user are obtained from the second database.

[0025] According to a second aspect of an embodiment of the present invention, an object recommendation device is provided, the device comprising:

[0026] A data acquisition unit, used to acquire user data authorized by the user, wherein the user data includes user information, historical object information and related object information;

[0027] A short-term feature determination unit, used to determine the short-term interest feature of the current user according to the behavior information of the current user;

[0028] A clustering unit, configured to determine a user set and an object set similar to the current user according to the user information and the historical object information, wherein the user set includes users with similar attribute information, and the object set includes multiple objects similar to the historical object;

[0029] A long-term feature determination unit, configured to determine the long-term interest feature of the current user according to the user set and the object set;

[0030] A prediction unit, configured to input the short-term interest feature and the long-term interest feature into a recommendation model to determine a recommendation object, wherein the recommendation model is pre-trained based on a plurality of user data; and

[0031] A sending unit is used to send the recommended object.

[0032] Preferably, the short-term feature determination unit comprises:

[0033] The short-term feature determination subunit is used to determine the short-term interest features of the user according to the behavior information using a graph embedding algorithm.

[0034] Preferably, the long-term feature determination unit comprises:

[0035] The long-term feature determination subunit is used to determine the long-term interest features of the user according to the user set and the object set by adopting a graph embedding algorithm.

[0036] Preferably, the clustering unit comprises:

[0037] A first clustering subunit is used to determine a set of similar users having similar user information to the user by using a clustering algorithm; and

[0038] The second clustering subunit is used to determine a set of similar objects similar to the historical object information by using a clustering algorithm.

[0039] Preferably, the recommendation model takes short-term interest features and long-term interest features as inputs and historical object information as outputs, and is trained using a logistic regression algorithm or an extreme gradient boosting algorithm.

[0040] Preferably, the object recommendation device further comprises:

[0041] A first preprocessing unit, used to preprocess user information and historical object information of multiple users, and save them to a first database;

[0042] A second preprocessing unit, used to preprocess the user information and behavior information of the current user and save them to a second database;

[0043] The data acquisition unit comprises:

[0044] A first data acquisition subunit, used to acquire user information and historical object information of multiple users from a first database;

[0045] The second data acquisition subunit is used to acquire user information and behavior information of the current user from the second database.

[0046] According to a third aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, wherein when the computer program is executed by a processor, the method as described in any one of the first aspects is implemented.

[0047] According to a fourth aspect of an embodiment of the present invention, an electronic device is provided, wherein the mobile terminal includes a processor, and the processor is used to implement the method as described in any one of the first aspects when executing a computer program stored in a memory.

[0048] The object recommendation method of the embodiment of the present invention includes: obtaining user data authorized by the user; determining the short-term interest characteristics of the current user according to the behavior information of the current user; determining the user set and object set similar to the current user according to the user information and historical object information; determining the long-term interest characteristics of the current user according to the user set and the object set; inputting the short-term interest characteristics and the long-term interest characteristics into the recommendation model to determine the recommended object, and the recommendation model is pre-trained based on multiple user data; and sending the recommended object. The embodiment of the present invention can improve the accuracy of the recommended object by determining the recommended object based on the two dimensions of short-term interest characteristics and long-term interest characteristics. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The above and other objects, features and advantages of the present invention will become more apparent through the following description of the embodiments of the present invention with reference to the accompanying drawings, in which:

[0050] Figure 1 A schematic diagram of an object recommendation system provided by an embodiment of the present invention;

[0051] Figure 2 A schematic diagram of a flow chart of an object recommendation method according to an embodiment of the present invention;

[0052] Figure 3 is a schematic diagram of the structure of an object recommendation device according to an embodiment of the present invention;

[0053] Figure 4 is a schematic diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0054] The following describes the embodiments of the present invention based on the embodiments, but the embodiments of the present invention are not limited to these embodiments. In the detailed description of the embodiments of the present invention below, some specific details are described in detail. It is possible for those skilled in the art to fully understand the present invention without the description of these details. In order to avoid confusing the essence of the embodiments of the present invention, known methods, processes, flows, components and circuits are not described in detail.

[0055] In addition, persons of ordinary skill in the art will appreciate that the drawings provided herein are for illustration purposes and are not necessarily drawn to scale.

[0056] Unless the context clearly requires otherwise, words such as “include”, “including” and the like in the embodiments of the present invention should be interpreted as including rather than being exclusive or exhaustive; that is, as including but not limited to.

[0057] In the description of the embodiments of the present invention, it should be understood that the terms "first", "second", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance. In addition, in the description of the embodiments of the present invention, unless otherwise specified, "plurality" means two or more.

[0058] For objects with low purchase frequency, it is difficult to predict the parameters of the object that the user wants to buy based on the user's relatively small purchase data. The object parameters may include the brand and price of the object. In view of this, an embodiment of the present invention provides a recommendation method to improve the recommendation accuracy of objects with less purchase data.

[0059] It should be understood that the recommended object in the embodiment of the present invention may be a car, a house, or a large or medium-sized household appliance, etc. In the following description, the recommended object is illustrated as a car.

[0060] In an embodiment of the present invention, based on a pre-trained recommendation model, objects that a user may be interested in and recommended to the user can be determined through the human-computer interaction behavior and network access behavior data of multiple target users recorded by a terminal-side program.

[0061] Figure 1 Schematic diagram of a recommendation system provided by an embodiment of the present invention. Figure 1 As shown, the recommendation system includes: multiple terminal devices 1 and a server 2. The terminal device 1 can be a smart phone, a tablet computer or a personal computer (PC), etc. The server 2 can be a single server, a server cluster configured in a distributed manner, or a cloud server. The multiple terminal devices 1 and the server 2 exchange data through a network 3.

[0062] In an embodiment of the present invention, the terminal device 1 may be a smart phone or a PC used by a user. When a user uses the terminal device 1, human-computer interaction behavior and network access behavior data are often generated. By analyzing these data by the server 2 or the terminal device 1, the user's consumption habits and purchasing tendencies can be determined.

[0063] In some embodiments, the server 2 may include one or more processing engines (e.g., a single-core processing engine or a multi-core processor). As an example only, the server 2 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), an application specific instruction set processor (ASIP), a graphics processing unit (GPU), a physical processing unit (PPU), a digital signal processor (DSP), a field programmable gate array (FPGA), a programmable logic device (PLD), a controller, a microcontroller unit, a reduced instruction set computer (RISC), a microprocessor, etc. One or more combinations thereof.

[0064] In some embodiments, the network 3 may be any form of wired or wireless network, or any combination thereof. By way of example only, the network 3 may be a wired network, a fiber optic network, a telecommunications network, an intranet, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), a wide area network (WAN), a public switched telephone network (PSTN), a Bluetooth network, a ZigBee network, a near field communication (NFC) network, a global system for mobile communications (GSM) network, a code division multiple access (CDMA) network, a time division multiple access (TDMA) network, a general packet radio service (GPRS) network, an enhanced data rate for GSM evolution (EDS), or a wireless communication network. One or more combinations of the following: wireless communication (LTE) network, broadband code division multiple access (WCDMA) network, high speed downlink packet access (HSDPA) network, long term evolution (LTE) network, user datagram protocol (UDP) network, transmission control protocol / Internet protocol (TCP / IP) network, short message service (SMS) network, wireless application protocol (WAP) network, ultra-wideband (UWB) network, mobile communication (1G, 2G, 3G, 4G, 5G) network, Wi-Fi, Li-Fi, narrowband Internet of Things (NB-IoT), infrared communication, etc.

[0065] Figure 2 is a flow chart of the object recommendation method according to an embodiment of the present invention. Figure 2 As shown, the recommended method of the embodiment of the present invention includes the following steps:

[0066] Step S210, obtaining user data authorized by the user, wherein the user data includes user information, historical object information and behavior information.

[0067] In this embodiment, user data is raw data obtained with user authorization, pre-processed and stored in a database. The raw data may include user attribute information, behavior information and historical object information. The attribute information may be the user's personal information, including age, gender, income and whether there is a car. The behavior information may include user behavior data, including click, browse and search information. Specifically, it may include page information and browsing time of the object browsed by the user. The page information may include browsing the commodity homepage, video page, detail page, evaluation page and parameter page and other pages and the browsing time staying on the page. Specifically, in this embodiment, the object information may include information such as the brand, model and price of the car. The historical object information is the information of the car that the user has used. It can be obtained through the user's order record in the trading platform, or it can be obtained through the information entered by the user. In an optional implementation, the user's behavior information may also include information collected from multiple platforms authorized by the user, including information on behaviors such as taxi, car rental and driver. And evaluate the user's consumption habits based on this.

[0068] In an optional embodiment, before obtaining the user data authorized by the user, the method further includes:

[0069] Step S200: pre-processing the user data.

[0070] Specifically, the preprocessing includes:

[0071] User information and historical object information of multiple users are preprocessed and saved in the first database. User information includes user gender, age, income level, etc. Historical object information is the information of cars purchased by users in the past, which may include: car guide price, displacement, gearbox type, etc. Hive technology is specifically used to store offline data in the first database. Hive is a data warehouse tool based on Hadoop, which can be used for data extraction, transformation, and loading. In practical applications, Hive can be used for functions such as storage and preprocessing of big data, and the full-link processor can be used to store and preprocess basic data and performance data.

[0072] Preprocess the user information and behavior information of the current user and save them in the second database. User behavior information includes real-time click sequences: home page, business details page, video page, etc. Redis technology is used to store real-time click data in the second database. Redis is a commonly used cache management tool. Due to its memory reading characteristics, it is often used in high-concurrency projects with high performance requirements.

[0073] In this implementation, obtaining the user data authorized by the user specifically includes obtaining the user information and historical object information of multiple users from the first database; and obtaining the user information and behavior information of the current user from the second database.

[0074] Specifically, it is mainly divided into continuous data preprocessing, discrete data preprocessing and statistical feature generation. Continuous data preprocessing mainly processes similar features such as car cost price and transaction price, and processes abnormal values ​​(fills or treats them as vacant values). The discrete data preprocessing module mainly preprocesses time, attribute and other features, divides time features into boxes by day / week / month, and processes attribute features by encoding method or one-hot method; statistical feature generation specifically counts the number of times users have taken private cars, express cars, visited car purchase pages, etc. in the past period of time, and stores the processed structured data in real-time Redis devices and offline hive big data storage devices respectively.

[0075] Step S220, determining the short-term interest characteristics of the current user according to the behavior information of the current user.

[0076] Determining the short-term interest features of the current user based on the user's behavior information includes: using a graph embedding algorithm to determine the short-term interest features of the user based on the behavior information. Specifically, the user's short-term behavior features are obtained by performing graph embedding processing on the user's behavior data within a predetermined time. The graph embedding process specifically inputs the user data into a pre-trained graph embedding model. The graph embedding methods in the embodiments of the present invention include: Deep Walk, Node2Vec, and Large-scale Information Network Embedding (LINE), etc.

[0077] Step S230: Determine a user set and an object set similar to the current user based on the user information and the historical object information. The user set includes users with similar attribute information, and the object set includes multiple objects similar to the historical objects.

[0078] Specifically, similar users are determined by clustering based on the user's attribute information and the model clicked as the main features. Clustering is performed based on features such as the user's current behavior information and the user's attribute information. In an optional implementation, users with similar age, income and click behavior to the current user are determined as similar users based on a clustering algorithm. The specific click behavior can be browsing or purchasing a model similar to the one the user has recently browsed. A clustering algorithm is used to determine a set of objects similar to the model clicked by the current user. Clustering is performed based on the brand model or price of the vehicles purchased by similar users as the main features to determine similar objects.

[0079] In this step, low-frequency transaction records and irrelevant users can be removed through aggregation, which can reduce irrelevant data and improve prediction efficiency and accuracy.

[0080] The clustering algorithms may include K-means clustering algorithm (K-Means), mean shift clustering algorithm (Mean Shift), density-based spatial clustering of applications with noise (DBSCAN) and hierarchical clustering algorithm (Hierarchical Clustering).

[0081] Step S240: determining the long-term interest characteristics of the current user according to the user set and the object set.

[0082] Specifically, the information of each user in the user set is processed by the graph embedding algorithm and represented as a vector. Specifically, the user set is represented as an embedded vector by solving function (1). Where book is the positive set of click behaviors, neg is the negative set of click behaviors, and v c The user's click behavior over a period of time. is the user's click behavior at time t.

[0083]

[0084] The information of vehicles in the object set is represented as a vector through the graph embedding algorithm. Specifically, the object set is represented as an embedded vector by solving function (2). Where book is the positive set of click behaviors, neg is the negative set of click behaviors, and v c The user's click behavior over a period of time. The car is clicked at time t.

[0085]

[0086] The embedding vector determined by the similar user set and the embedding vector determined by the similar object set are connected to determine the long-term interest characteristics of the user. The long-term interest of a class of users in the car group is represented by a vector. The embodiment of the present invention uses a clustering method to express the long-term interest characteristics of the user group and the car group, which can avoid the problem of inaccurate prediction due to less data on users buying cars. At the same time, the dimension of the data is reduced, which can improve the prediction efficiency of the model.

[0087] Step S250, inputting the short-term interest feature and the long-term interest feature into a recommendation model to determine a recommendation object, wherein the recommendation model is pre-trained based on a plurality of user data.

[0088] In an optional implementation, the recommendation model takes short-term interest features and long-term interest features as input and historical objects as output, and is trained using a logistic regression (LR) algorithm or an extreme gradient boosting (GBDT) decision tree (GBDT) algorithm.

[0089] Among them, the LR algorithm is a commonly used model for classification tasks in machine learning. Its essence is a generalized linear regression analysis model. The model structure of the LR model is simple, the training speed is fast, and it has a good probability explanation for the output variable. Among them, the XGBoost algorithm is an extensible machine learning system that can be used as an open source software package. At the same time, the influence of the system has been widely recognized in a large number of machine learning and data mining challenges. In an embodiment of the present invention, as the amount of data continues to increase, XGBoost can play a good classification role. The historical object is an object that the user has clicked or browsed in history.

[0090] In an optional implementation, the prediction capability of the model is continuously improved by continuously re-inputting the user's actual car purchase data into the model for training.

[0091] Step S260: sending the recommended object.

[0092] Specifically, the recommended object is sent to the user terminal for display.

[0093] The embodiment of the present invention represents the short-term interest characteristics and long-term interest characteristics of users based on graph_embedding technology, and predicts the probability of users clicking on cars. In addition to being able to accurately recommend cars to users, the method of the embodiment of the present invention is also applicable to other low-frequency business recommendation scenarios and has universal applicability.

[0094] The embodiment of the present invention is based on graph_embedding technology, which not only takes into account the short-term interest characteristics of users, but also uses clustering methods to express the long-term interest characteristics of user sets and object sets, and repeatedly iterates and polishes the recommendation model with actual transaction data, and has strong generalization adaptability. From the early data collection to the graph_embedding representation of model features, and then combining short-term and long-term interest characteristics to use the supervision model to predict the click probability of user groups and car groups, finally forming an accurate recommendation object, and feeding back to the model whether the recommended object is clicked, and repeatedly polishing. Using this complete link to accurately and effectively represent the user's interest in cars, thereby recommending suitable cars to users, provides a new idea for selling cars in the existing car sales industry, and also provides a new idea for object recommendation in low-frequency trading industries.

[0095] The object recommendation method of the embodiment of the present invention includes: obtaining user data authorized by the user; determining the short-term interest characteristics of the current user according to the behavior information of the current user; determining the user set and object set similar to the current user according to the user information and historical object information; determining the long-term interest characteristics of the current user according to the user set and the object set; inputting the short-term interest characteristics and the long-term interest characteristics into a recommendation model to determine the recommended object, wherein the recommendation model is pre-trained based on multiple user data; and sending the recommended object. The embodiment of the present invention can improve the accuracy of the recommended object by inputting the short-term interest characteristics and the long-term interest characteristics into the recommendation model to determine the recommended object.

[0096] Figure 3 Schematic diagram of the structure of the object recommendation device according to an embodiment of the present invention. Figure 3 As shown, in an optional implementation, the object recommendation device of an embodiment of the present invention includes: a data acquisition unit 310, a short-term feature determination unit 320, a clustering unit 330, a long-term feature determination unit 340, a prediction unit 350 and a sending unit 360.

[0097] The data acquisition unit 310 is used to acquire user data authorized by the user, wherein the user data includes user information, historical object information and related object information.

[0098] The data acquisition unit 310 includes: a first data acquisition subunit and a second data acquisition subunit.

[0099] The first data acquisition subunit is used to acquire user information and historical object information of multiple users from the first database.

[0100] The second data acquisition subunit is used to acquire user information and behavior information of the current user from the second database.

[0101] The short-term feature determination unit 320 is used to determine the short-term interest features of the current user according to the behavior information of the current user.

[0102] The short-term feature determination unit 320 includes: a short-term feature determination subunit.

[0103] The short-term feature determination subunit is used to determine the short-term interest features of the user according to the behavior information by using a graph embedding algorithm.

[0104] The clustering unit 330 is used to determine a user set and an object set similar to the current user according to the user information and the historical object information, wherein the user set includes users with similar attribute information, and the object set includes multiple objects similar to the historical objects.

[0105] The clustering unit 330 includes: a first clustering subunit and a second clustering subunit.

[0106] The first clustering subunit is used to determine a set of similar users with similar user information by using a clustering algorithm.

[0107] The second clustering subunit is used to determine a set of similar objects similar to the historical object information by using a clustering algorithm.

[0108] The long-term feature determination unit 340 is used to determine the long-term interest feature of the current user according to the user set and the object set.

[0109] The long-term feature determination unit includes: a long-term feature determination subunit.

[0110] The long-term feature determination subunit is used to determine the long-term interest features of the user according to the user set and the object set by adopting a graph embedding algorithm.

[0111] The prediction unit 350 is used to input the short-term interest feature and the long-term interest feature into a recommendation model to determine a recommendation object, and the recommendation model is pre-trained based on multiple user data.

[0112] The recommendation model takes short-term interest features and long-term interest features as input and historical object information as output, and is trained using a logistic regression algorithm or an extreme gradient boosting algorithm.

[0113] The sending unit 360 is used to send the recommended object.

[0114] In another optional implementation, the object recommendation device further includes: a first preprocessing unit 300a and a second preprocessing unit 300b.

[0115] The first preprocessing unit 300a is used to preprocess the user information and historical object information of multiple users and save them to the first database.

[0116] The second preprocessing unit 300b is used to preprocess the user information and behavior information of the current user and save them to the second database.

[0117] Figure 4 Schematic diagram of an electronic device according to an embodiment of the present invention. Figure 4 As shown, Figure 4 The electronic device shown is an object recommendation device, which includes a general computer hardware structure, which includes at least a processor 401 and a memory 402. The processor 401 and the memory 402 are connected via a bus 403. The memory 402 is suitable for storing instructions or programs executable by the processor 401. The processor 401 can be an independent microprocessor or a collection of one or more microprocessors. Thus, the processor 401 executes the instructions stored in the memory 402, thereby executing the method flow of the embodiment of the present invention as described above to realize the processing of data and the control of other devices. The bus 403 connects the above-mentioned multiple components together, and at the same time connects the above-mentioned components to the display controller 404 and the display device and the input / output (I / O) device 405. The input / output (I / O) device 405 can be a mouse, a keyboard, a modem, a network interface, a touch input device, a somatosensory input device, a printer, and other devices known in the art. Typically, the input / output device 405 is connected to the system via an input / output (I / O) controller 406.

[0118] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, devices (equipment) or computer program products. Therefore, the embodiments of the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the embodiments of the present invention may take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0119] The present invention is described with reference to flowcharts of methods, apparatuses (devices) and computer program products according to embodiments of the present invention. It should be understood that each process in the flowchart can be implemented by computer program instructions.

[0120] These computer program instructions may be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device that implements the process Figure 1 A function specified in a process or multiple processes.

[0121] These computer program instructions may also be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce the instructions for implementing the process Figure 1 A device that specifies functions in a process or multiple processes.

[0122] Another embodiment of the present invention relates to a non-volatile storage medium for storing a computer-readable program, wherein the computer-readable program is used for a computer to execute part or all of the above method embodiments.

[0123] That is, those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including a number of instructions to enable a device (which can be a single-chip microcomputer, chip, etc.) or a processor to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0124] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An object recommendation method, applicable to automobile transactions, characterized in that: The object recommendation method comprises: Obtain user data authorized by the user, the user data including user information, historical object information and related object information, the related object information including behavior information, the historical object information including the brand, model and price of the corresponding car, and the behavior information including taxi-hailing, car rental and designated driver behavior; Determining the short-term interest characteristics of the current user according to the behavior information of the current user; Determine a user set and an object set similar to the current user according to the user information and the historical object information, wherein the user set includes users with similar attribute information, and the object set includes multiple objects similar to the historical objects; Determining the long-term interest characteristics of the current user according to the user set and the object set; Inputting the short-term interest feature and the long-term interest feature into a recommendation model to determine a recommendation object, wherein the recommendation model takes the short-term interest feature and the long-term interest feature as input and takes the historical object as output, and is trained by a logistic regression algorithm or an extreme gradient boosting algorithm; and sending the recommended object; Wherein, determining the long-term interest feature of the current user according to the user set and the object set includes: The information of each user in the user set is represented as an embedding vector through a graph embedding algorithm; The information of vehicles in the object set is represented as an embedding vector through a graph embedding algorithm; According to the embedding vector determined by the similar user set and the embedding vector determined by the similar object set, the long-term interest characteristics of the current user are determined.

2. The method according to claim 1, characterized in that Determining the short-term interest characteristics of the current user according to the user's behavior information includes: A graph embedding algorithm is used to determine the short-term interest characteristics of the user based on the behavior information.

3. The method according to claim 1, characterized in that Determining a user set and an object set similar to the current user according to the user information and the historical object information includes: Using a clustering algorithm to determine a set of similar users with similar user information; and A clustering algorithm is used to determine a set of similar objects that are similar to the historical object information.

4. The method according to claim 1, characterized in that: The recommendation model takes short-term interest features and long-term interest features as inputs and historical object information as outputs, and is trained using a logistic regression algorithm or an extreme gradient boosting algorithm.

5. The method according to claim 1, characterized in that Before obtaining the user data authorized by the user, the object recommendation method further includes: Preprocessing user information and historical object information of multiple users and saving them to a first database; and Preprocessing the user information and behavior information of the current user and saving them to the second database; The user data for obtaining user authorization includes: Acquire user information and historical object information of multiple users from the first database; and The user information and behavior information of the current user are obtained from the second database.

6. An object recommendation device, suitable for automobile transactions, characterized in that: The device comprises: A data acquisition unit, used to acquire user data authorized by the user, wherein the user data includes user information, historical object information and related object information, wherein the related object information includes behavior information, wherein the historical object information includes the brand, model and price of the corresponding car, and wherein the behavior information includes taxi-hailing, car-renting and designated driver behavior; A short-term feature determination unit, used to determine the short-term interest feature of the current user according to the behavior information of the current user; A clustering unit, configured to determine a user set and an object set similar to the current user according to the user information and the historical object information, wherein the user set includes users with similar attribute information, and the object set includes multiple objects similar to the historical object; A long-term feature determination unit, configured to determine the long-term interest feature of the current user according to the user set and the object set; a prediction unit, configured to input the short-term interest feature and the long-term interest feature into a recommendation model to determine a recommendation object, wherein the recommendation model takes the short-term interest feature and the long-term interest feature as input and takes a historical object as output, and is obtained by training using a logistic regression algorithm or an extreme gradient boosting algorithm; and A sending unit, configured to send the recommended object; Wherein, the long-term feature determination unit is further used for: The information of each user in the user set is represented as an embedding vector through a graph embedding algorithm; The information of vehicles in the object set is represented as an embedding vector through a graph embedding algorithm; According to the embedding vector determined by the similar user set and the embedding vector determined by the similar object set, the long-term interest characteristics of the current user are determined.

7. The device according to claim 6, characterized in that The short-term feature determination unit comprises: The short-term feature determination subunit is used to determine the short-term interest features of the user according to the behavior information using a graph embedding algorithm.

8. The device according to claim 6, characterized in that The clustering unit comprises: A first clustering subunit is used to determine a set of similar users having similar user information to the user by using a clustering algorithm; and The second clustering subunit is used to determine a set of similar objects similar to the historical object information by using a clustering algorithm.

9. The device according to claim 6, characterized in that The recommendation model takes short-term interest features and long-term interest features as inputs and historical object information as outputs, and is trained using a logistic regression algorithm or an extreme gradient boosting algorithm.

10. The device according to claim 6, characterized in that The object recommendation device further comprises: A first preprocessing unit, used to preprocess user information and historical object information of multiple users, and save them to a first database; A second preprocessing unit, used to preprocess the user information and behavior information of the current user and save them to a second database; The data acquisition unit comprises: A first data acquisition subunit, used to acquire user information and historical object information of multiple users from a first database; The second data acquisition subunit is used to acquire user information and behavior information of the current user from the second database.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

12. An electronic device, characterized in that: The electronic device comprises a processor, and the processor is configured to implement the method according to any one of claims 1 to 5 when executing a computer program stored in a memory.

13. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images