Object recommendation method and device, equipment and medium

Abstract

The invention provides an object recommendation method and device, equipment and a medium, and relates to the technical field of computers, in particular to the technical field of artificial intelligence and recommendation. According to the implementation scheme, prediction evaluation information, aiming at multiple candidate objects, of a target user is obtained, and the prediction evaluation information is generated based on user features of the target user, user features of multiple target evaluation users and historical evaluation information, aiming at the multiple candidate objects, of the multiple target evaluation users; and based on the prediction evaluation information of the target user for the plurality of candidate objects, determining a recommendation object for the target user from the plurality of candidate objects.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, in particular to the field of artificial intelligence and recommendation technology, and specifically to an object recommendation method, device, electronic device, and computer-readable storage medium. Background Art

[0002] With the development of Internet technology, people's consumption, entertainment, learning, travel and other behaviors in life are closely related to the Internet. In the operation process of Internet platforms, it is usually necessary to use various channels to recommend users in order to improve the corresponding business performance.

[0003] The methods described in this section are not necessarily methods that have been previously conceived or employed. Unless otherwise indicated, it should not be assumed that any method described in this section is considered to be prior art simply because it is included in this section. Similarly, unless otherwise indicated, the issues mentioned in this section should not be considered to have been recognized in any prior art. Summary of the invention

[0004] The present disclosure provides an object recommendation method, device, electronic device, and computer-readable storage medium.

[0005] According to one aspect of the present disclosure, there is provided an object recommendation method, comprising: obtaining predicted evaluation information of a target user for multiple candidate objects, wherein the predicted evaluation information is generated based on user characteristics of the target user, user characteristics of multiple target evaluation users, and historical evaluation information of the multiple target evaluation users for multiple candidate objects; and determining a recommended object for the target user from the multiple candidate objects based on the predicted evaluation information of the target user for the multiple candidate objects.

[0006] According to another aspect of the present disclosure, an object recommendation device is provided, including: an acquisition unit, configured to acquire predicted evaluation information of a target user for multiple candidate objects, wherein the predicted evaluation information is generated based on user characteristics of the target user, user characteristics of multiple target evaluation users, and historical evaluation information of the multiple target evaluation users for multiple candidate objects; and a determination unit, configured to determine a recommended object for the target user from the multiple candidate objects based on the predicted evaluation information of the target user for the multiple candidate objects.

[0007] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned object recommendation method.

[0008] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable a computer to execute the above object recommendation method.

[0009] According to one or more embodiments of the present disclosure, evaluation information can be generated based on the user characteristics of the target user, the user characteristics of multiple target evaluation users, and the historical evaluation information of multiple target evaluation users, and the predicted evaluation information of the target user for multiple candidate objects can be obtained while ensuring accuracy, and object recommendations can be made for the target user based on the predicted evaluation information to improve the recommendation effect.

[0010] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The accompanying drawings exemplarily illustrate the embodiments and constitute a part of the specification, and together with the text description of the specification, are used to explain the exemplary implementation of the embodiments. The embodiments shown are for illustrative purposes only and do not limit the scope of the claims. In all drawings, the same reference numerals refer to similar but not necessarily identical elements.

[0012] Figure 1 A flowchart of an object recommendation method according to an exemplary embodiment of the present disclosure is shown;

[0013] Figure 2 A schematic diagram showing a process of acquiring user characteristics according to an exemplary embodiment of the present disclosure;

[0014] Figure 3 A schematic diagram showing a process of determining an evaluation feature vector according to an exemplary embodiment of the present disclosure;

[0015] Figure 4 A schematic diagram showing a process of determining user similarity according to an exemplary embodiment of the present disclosure;

[0016] Figure 5 A schematic diagram showing a process of determining predicted evaluation information according to an exemplary embodiment of the present disclosure.

[0017] Figure 6 A flowchart of an object recommendation method according to an exemplary embodiment of the present disclosure is shown;

[0018] Figure 7 A structural block diagram of an object recommendation device according to an exemplary embodiment of the present disclosure is shown;

[0019] Figure 8 A block diagram of an electronic device according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0020] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope of the present disclosure. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0021] In the present disclosure, unless otherwise specified, the use of the terms "first", "second", etc. to describe various elements is not intended to limit the positional relationship, timing relationship, or importance relationship of these elements, and such terms are only used to distinguish one element from another element. In some examples, the first element and the second element may refer to the same instance of the element, and in some cases, based on the description of the context, they may also refer to different instances.

[0022] The terms used in the description of various examples in this disclosure are only for the purpose of describing specific examples and are not intended to be limiting. Unless the context clearly indicates otherwise, if the number of elements is not specifically limited, the element can be one or more. In addition, the term "and / or" used in this disclosure covers any one of the listed items and all possible combinations.

[0023] In the related art, object recommendations are usually made based on existing historical evaluation information of users regarding the recommended objects. When the historical evaluation information of users is less, the recommendation effect will be greatly affected.

[0024] Based on this, the present disclosure provides an object recommendation method, which generates evaluation information based on the user characteristics of the target user, the user characteristics of multiple target evaluation users and the historical evaluation information of multiple target evaluation users, so that the historical evaluation information of other users can be used to obtain the target user's predicted evaluation information for multiple candidate objects, and object recommendations are made for the target user based on the predicted evaluation information to improve the recommendation accuracy.

[0025] The embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.

[0026] Figure 1 FIG. 1 is a flowchart of an object recommendation method 100 according to an exemplary embodiment of the present disclosure. Figure 1 As shown, the method 100 includes:

[0027] Step S101, obtaining predicted evaluation information of a target user for multiple candidate objects, wherein the predicted evaluation information is generated based on user characteristics of the target user, user characteristics of multiple target evaluation users, and historical evaluation information of multiple target evaluation users for multiple candidate objects; and

[0028] Step S102: Determine a recommended object for the target user from the multiple candidate objects based on the predicted evaluation information of the target user for the multiple candidate objects.

[0029] By applying method 100, the evaluation data sets of other users can be used to specifically expand the evaluation data set of the target user. Specifically, the correlation between the target user and the target evaluation users can be fully considered by obtaining the user characteristics of the target user and the user characteristics of multiple target evaluation users. On this basis, the historical evaluation information of other users can be used to generate the predicted evaluation information of the target user, and data enhancement can be achieved while ensuring accuracy. Then, object recommendations can be made based on the predicted evaluation information to improve the recommendation accuracy.

[0030] In some examples, the above method can be applied in a cold start scenario for making recommendations for new users, so that when the historical evaluation information of the target user is insufficient, data enhancement can be performed based on the historical evaluation information of other users, thereby achieving more accurate object recommendations.

[0031] In some examples, candidate objects may include various types of objects to be recommended in a variety of application scenarios. For example, when method 100 is applied to e-commerce scenarios such as online shopping, candidate objects may include different types of goods or various types of goods in the scenario, such as consumer coupons, etc. For another example, when method 100 is applied to operating scenarios of Internet platforms such as video playback and e-book reading, candidate objects may include various videos, e-books, etc. to be recommended. It should be understood that method 100 can also be applied to other scenarios, and candidate objects may include corresponding objects to be recommended in different scenarios. For example, in travel service scenarios, different types of travel modes (such as taxis, shared bicycles, etc.) can be recommended to users, and this disclosure is not limited to this.

[0032] In some examples, historical evaluation information and predicted evaluation information may include various content types, such as rating information, evaluation text, etc. The rating information may be discrete rating information or continuous rating information, as long as it can represent the user's evaluation of the candidate object. The present disclosure does not limit this.

[0033] In some examples, the user characteristics of the target user may include a user portrait for indicating information in multiple dimensions such as the user's personal characteristics and behavioral characteristics. For example, the user portrait of the target user, such as gender, age group, occupation, and resident area, may also include the historical behavioral characteristics of the target user in the application scenario corresponding to the candidate object. For example, when method 100 is applied to the field of e-commerce, the historical behavioral characteristics may include the characteristics of the user's historical purchase of goods, such as purchase frequency, types of purchased goods, average order value, etc. For another example, when method 100 is applied to a video playback website, the historical behavioral characteristics may include the characteristics of the user's historical video viewing, such as viewing frequency, types of videos viewed, viewing duration, etc.

[0034] In some examples, the user features may be user feature vectors obtained using a specific encoding method, so that the user features may be learned using a machine learning method to achieve more efficient and accurate object recommendations.

[0035] Figure 2 A schematic diagram of a process of acquiring user features according to an exemplary embodiment of the present disclosure is shown. In this example, the user features may be user feature vectors obtained by using a user feature extraction network.

[0036] See also Figure 2 As shown, in this example, the user feature extraction network can be obtained based on the training process of the user classification model including the cascade network layer, the fully connected layer, the user feature extraction network, the multi-layer perceptron, and the output layer. For example, the user classification model can be supervisedly trained using a sample user data set labeled with real user categories, and then the user feature extraction network included in the trained user classification model can be used to encode the user data to obtain a user feature vector.

[0037] In the above example, the user classification task is used to train the user feature extraction network. In some examples, the user feature extraction network can also be trained using labeled sample data corresponding to other tasks to obtain a user feature extraction network that can efficiently and accurately encode user feature data and output user feature vectors.

[0038] According to some embodiments, the process of generating predicted evaluation information includes: clustering multiple target evaluation users based on user features of multiple target evaluation users to obtain multiple evaluation user groups; and determining group user features of each evaluation user group based on user features of target evaluation users in the same evaluation user group; and generating predicted evaluation information of the target user for multiple candidate objects based on user features of the target user, group user features of multiple evaluation user groups, and historical evaluation information of multiple target evaluation users for multiple candidate objects. Thus, multiple target evaluation users can be clustered, and calculations can be performed based on the group user features of the clustered evaluation user groups, which can reduce the amount of calculations and improve efficiency.

[0039] In some examples, a K-means clustering algorithm may be used to obtain multiple evaluation user groups. The clustering process may include, for example: determining the target number of multiple evaluation user groups, which step may be determined manually according to actual needs to reduce the amount of calculation and save computing resources while ensuring the accuracy of the calculation; randomly selecting the user features of a target evaluation user as the initial group center; calculating the similarity between the user features of each target evaluation user and the initial group center; determining the user features of the target evaluation user with the lowest similarity to the initial group center as another group center, and repeating this step based on the group center until the number of the determined multiple group centers reaches the target number; based on the similarity between the user features of each target evaluation user and the multiple group centers, determining the evaluation user groups to which the multiple target evaluation users belong respectively, so that the sum of the similarities between the user features of all target evaluation users and the group centers of the evaluation user groups to which they belong is maximized.

[0040] In some examples, when the user feature is in the form of a user feature vector, the similarity between the user feature of the target evaluation user and the corresponding group center can be determined by calculating the similarity between two user feature vectors (for example, cosine similarity, Euclidean similarity, etc.).

[0041] According to some embodiments, based on the user characteristics of the target user, the group user characteristics of multiple evaluation user groups, and the historical evaluation information of multiple target evaluation users for multiple candidate objects, generating the target user's predicted evaluation information for multiple candidate objects includes: determining the group evaluation characteristics of each evaluation user group based on the historical evaluation information of the target evaluation users in the same evaluation user group; and generating the target user's evaluation information for multiple candidate objects based on the user characteristics of the target user, the group user characteristics of multiple evaluation user groups, and the group evaluation characteristics of multiple evaluation user groups. Thus, on the basis of clustering multiple target evaluation users, the group evaluation characteristics of the evaluation user groups can be further determined, and calculations can be performed based on the group evaluation characteristics, thereby further reducing the amount of calculations and improving efficiency.

[0042] According to some embodiments, based on the historical evaluation information of the target evaluation user in the same evaluation user group, determining the group evaluation features of each evaluation user group includes: determining the evaluation feature vectors corresponding to the historical evaluation information of multiple target evaluation users; and performing pooling calculations based on the evaluation feature vectors corresponding to the historical evaluation information of the target evaluation user in the same evaluation user group to obtain the group evaluation features of each evaluation user group. For example, the group evaluation features of each evaluation user group may be obtained by a mean pooling calculation method. By encoding the historical evaluation information into an evaluation feature vector and performing a pooling calculation, the group evaluation features of the evaluation user group can be obtained more easily and accurately, thereby improving the calculation efficiency.

[0043] According to some embodiments, the historical evaluation information includes at least one evaluation tag, and the at least one evaluation tag is determined from a plurality of preset tags. Thus, the evaluation tags in a limited set can be used to indicate the characteristics of the user's evaluation content, avoiding the defects of insufficient feature encoding accuracy and large encoding calculation in the open domain, so as to improve the accuracy and efficiency of recommendation.

[0044] In some examples, the evaluation tag may include multiple sub-tags for indicating different contents, such as for indicating whether the evaluation information is a positive evaluation or a negative evaluation.

[0045] In some examples, when the historical evaluation information includes evaluation text, the evaluation label can be obtained by extracting preset keywords in the evaluation text. For another example, the evaluation label can be determined by using a label prediction model for the evaluation text. In this case, the label prediction model can be obtained by supervised training using sample evaluation texts annotated with reference labels, so as to more accurately and efficiently obtain evaluation labels that can indicate the characteristics of the evaluation content.

[0046] Figure 3 A schematic diagram of the determination process of the evaluation feature vector according to an exemplary embodiment of the present disclosure is shown. In this example, the historical evaluation information includes the evaluation text, and then the evaluation text can be first input into the language model to encode the evaluation text to obtain the evaluation text feature vector. The language model can be obtained by pre-training using a large-scale corpus, for example, it can be a bidirectional encoder representation method model (Bidirectional Encoder Representations from Transformers, BERT) or a knowledge integration enhanced table model (Enhanced Representation through Knowledge Integration, ERNIE). By inputting the evaluation text feature vector into the label prediction model to obtain the label prediction result, and then encoding the label prediction result to obtain the evaluation label feature vector, the evaluation label feature vector can be used to more efficiently and accurately indicate the user's evaluation characteristics.

[0047] In some examples, a pooling calculation may be performed based on each evaluation label feature vector corresponding to the evaluation user group to obtain the group evaluation feature.

[0048] According to some embodiments, based on the user characteristics of the target user, the user characteristics of multiple target evaluation users, and the historical evaluation information of the multiple target evaluation users for multiple candidate objects, generating the target user's predicted evaluation information for multiple candidate objects includes: determining the user similarity between the target user and the multiple target evaluation users based on the user characteristics of the target user and the user characteristics of the multiple target evaluation users; and generating the target user's predicted evaluation information for multiple candidate objects based on the user similarity between the target user and the multiple target evaluation users and the historical evaluation information of the multiple target evaluation users for multiple candidate objects. Thus, the predicted evaluation information can be generated based on the user similarity between the target user and the target evaluation users to obtain more accurate predicted evaluation information.

[0049] In some examples, the predicted evaluation information may also be generated by other means. For example, multiple target evaluation users may be further screened based on the user characteristics of the target user and the user characteristics of multiple target evaluation users, and the predicted evaluation information may be generated based on the historical evaluation information of at least one screened target evaluation user to improve efficiency and convenience.

[0050] For example, in one example, the user feature may include a user category, and the user category may be determined based on the user feature and a preset category standard. In this case, multiple target evaluation users may be further screened based on the user category of the target user and the user categories of multiple target evaluation users to further improve efficiency.

[0051] According to some embodiments, based on the user features of the target user and the user features of the multiple target evaluation users, determining the user similarity between the target user and the multiple target evaluation users includes: calculating the attention weights between the user features of the multiple target evaluation users and the user features of the target user based on the attention mechanism to obtain the user similarity between the target user and the multiple target evaluation users. Thus, the attention weights can be calculated based on the attention mechanism as user similarity information to improve accuracy.

[0052] In some examples, the attention mechanism based on the Transformer network may include multiple types such as a cross-attention mechanism and a multi-head attention mechanism. The cross-attention mechanism is a calculation mechanism for the attention weight applied to cross-sequence data. The cross-sequence data may be, for example, user feature data of different users as described in the exemplary embodiment of the present disclosure. In the mutual attention mechanism, the calculation is usually based on two sets of input data, the query sequence and the key value sequence. By calculating the similarity between each element in the query sequence and each element in the key value sequence, the attention weight of the query sequence can be obtained. Similarly, the user feature data of different users can also be used as different input sequences of the multi-head attention mechanism. In the multi-head attention mechanism, the weights can be calculated independently using multiple heads, and the final attention weights can be obtained by splicing the weights of multiple heads. The multi-head attention mechanism interacts with each other, and the correlation between different input sequences (i.e., user features of different users) can be better captured.

[0053] Figure 4 A schematic diagram of the process of determining user similarity according to an exemplary embodiment of the present disclosure is shown. In this example, a weight calculation network including a Softmax activation layer, a linear layer, a Tanh mapping layer, and a Matmal multiplication layer can be constructed based on the attention mechanism of the Transformer network. After clustering multiple target evaluation users and determining the group user features of multiple evaluation user groups, the group user features of the multiple evaluation user groups and the user features of the target user are input into the weight calculation network to obtain the weight information of each evaluation user group relative to the target user, and the weight information characterizes the similarity between each evaluation user group and the target user.

[0054] According to some embodiments, based on the user similarity between the target user and multiple target evaluation users and the historical evaluation information of the multiple target evaluation users for multiple candidate objects, generating the target user's predicted evaluation information for multiple candidate objects includes: determining the user weights of the multiple target evaluation users based on the user similarity between the target user and the multiple target evaluation users; and performing weighted calculation on the historical evaluation information of the multiple target evaluation users based on the user weights of the multiple target evaluation users to obtain the target user's predicted evaluation information for the multiple candidate objects. Thus, the user similarity can be used as weight data, and the evaluation information of the target user can be obtained more accurately and efficiently through weighted calculation.

[0055] In some examples, the predicted evaluation information may be obtained by other means, for example, after further screening multiple target evaluation users, the predicted evaluation information may be obtained by performing pooling calculation based on the screened historical evaluation information. For another example, the historical evaluation information of the target evaluation user who meets the preset conditions may be directly used as the predicted evaluation information of the target user to further improve efficiency.

[0056] Figure 5 A schematic diagram of a process for determining predicted evaluation information according to an exemplary embodiment of the present disclosure is shown. In this example, the predicted evaluation information of a target user for multiple candidate objects may be calculated based on the weight information output by the weight calculation network as described above, so that the group evaluation characteristics of the evaluation user group that is more similar to the target user can affect the calculation result to a greater extent, so as to obtain more accurate predicted evaluation information.

[0057] In some examples, the process of generating the above-mentioned predicted evaluation information may have been completed before executing the recommendation method 100. By storing the predicted evaluation information of the target user in an offline database, the predicted evaluation information can be directly obtained during the online object recommendation process, thereby reducing the amount of online calculations and improving the recommendation efficiency.

[0058] In some examples, in step S103, based on the target user's predicted evaluation information for multiple candidate objects, determining the recommended object for the target user from multiple candidate objects may include: inputting the target user's predicted evaluation information for multiple candidate objects into the recommendation model to obtain the recommended object output by the recommendation model. The recommendation model may, for example, be obtained by supervised training using sample data labeled with reference recommendation results, and the sample data may specifically include predicted evaluation information for multiple candidate objects. By inputting the sample data into the initial recommendation model and obtaining the predicted recommendation results, the loss value can be calculated based on the predicted results and the labeled reference recommendation results, and then the initial recommendation model can be adjusted based on this. By repeating the training and parameter adjustment steps multiple times, a recommendation model with optimized performance can be obtained.

[0059] In some examples, the recommended object may be determined based on more information in step S103. For example, the recommended object may be determined by further combining object features of multiple candidate objects to further improve accuracy.

[0060] In some examples, the above-mentioned predictive evaluation information can also be applied to the training process of the recommendation model. By labeling the sample data set containing the predictive evaluation information and using it as the supervisory information for the recommendation model training, a larger and more accurate sample data set can be used for training to obtain a more accurate recommendation model.

[0061] Figure 6FIG. 6 is a flowchart of an object recommendation method 600 according to an exemplary embodiment of the present disclosure. Figure 6 As shown, the method 600 includes:

[0062] Step S601: Obtain user characteristics of a target user and user characteristics of multiple target evaluation users for multiple candidate objects.

[0063] Step S602: Obtain at least one historical evaluation information of each evaluation user for multiple candidate objects.

[0064] Step S603: Determine evaluation feature vectors corresponding to historical evaluation information of multiple target evaluation users.

[0065] Step S604: clustering the multiple target evaluation users based on their user characteristics to obtain multiple evaluation user groups.

[0066] Step S605: Determine the group user characteristics of each evaluation user group based on the user characteristics of at least one target evaluation user in the evaluation user group.

[0067] Step S606: Determine the group evaluation feature of the evaluation user group based on the evaluation feature vector corresponding to each historical evaluation information in each evaluation user group.

[0068] Step S607: Determine the similarity between each evaluation user group and the target user based on the group user characteristics of the multiple evaluation user groups and the user characteristics of the target user.

[0069] Step S608: Based on the similarities between the multiple evaluation user groups and the target user, weighted calculation is performed on the group evaluation features of the multiple evaluation user groups to obtain predicted evaluation information of the target user for the multiple candidate objects.

[0070] Step S609: Determine a recommended object for the target user from the multiple candidate objects based on the predicted evaluation information of the target user for the multiple candidate objects.

[0071] By applying the above method 600 , it is possible to accurately and efficiently expand the target user's evaluation data set for the candidate object, thereby achieving more accurate object recommendation.

[0072] In some examples, the above-mentioned steps S601-S608 and step S609 can be implemented using different execution entities respectively. After completing the above-mentioned steps S601-S608, the target user's predicted evaluation information for multiple candidate objects can be stored in an offline database, so that the predicted evaluation information can be directly obtained during the process of executing object recommendation to improve the recommendation efficiency.

[0073] According to one aspect of the present disclosure, a device for recommending an object is also provided. Figure 7 FIG. 7 shows a structural block diagram of an object recommendation device 700 according to an exemplary embodiment of the present disclosure. Figure 7 As shown, the apparatus 700 includes:

[0074] An acquisition unit 701 is configured to acquire predicted evaluation information of a target user for a plurality of candidate objects, wherein the predicted evaluation information is generated based on a user feature of the target user, user features of a plurality of target evaluation users, and historical evaluation information of the plurality of target evaluation users for a plurality of candidate objects;

[0075] The determination unit 702 is configured to determine a recommended object for the target user from the multiple candidate objects based on the predicted evaluation information of the target user for the multiple candidate objects.

[0076] According to some embodiments, the predicted evaluation information in the acquisition unit 701 is generated by using a predicted evaluation information generating device, which includes: a clustering module, configured to cluster multiple target evaluation users based on user characteristics of multiple target evaluation users to obtain multiple evaluation user groups; a first determination module, configured to determine the group user characteristics of each evaluation user group based on the user characteristics of the target evaluation users in the same evaluation user group; a first generation module, configured to generate predicted evaluation information of the target user for multiple candidate objects based on the user characteristics of the target user, the group user characteristics of multiple evaluation user groups and the historical evaluation information of multiple target evaluation users for multiple candidate objects.

[0077] According to some embodiments, the first generation module includes: a determination submodule, configured to determine the group evaluation characteristics of each evaluation user group based on the historical evaluation information of the target evaluation user in the same evaluation user group; and a generation submodule, configured to generate the target user's evaluation information for multiple candidate objects based on the user characteristics of the target user, the group user characteristics of multiple evaluation user groups, and the group evaluation characteristics of multiple evaluation user groups.

[0078] According to some embodiments, the determination submodule is configured to: determine the evaluation feature vectors corresponding to the historical evaluation information of multiple target evaluation users; and perform pooling calculations based on the evaluation feature vectors corresponding to the historical evaluation information of the target evaluation users in the same evaluation user group to obtain the group evaluation features of each evaluation user group.

[0079] According to some embodiments, the predicted evaluation information generating device includes: a second determination module, configured to determine the user similarity between the target user and multiple target evaluation users based on the user characteristics of the target user and the user characteristics of multiple target evaluation users; and a second generation module, configured to generate predicted evaluation information of the target user for multiple candidate objects based on the user similarity between the target user and the multiple target evaluation users and the historical evaluation information of the multiple target evaluation users for multiple candidate objects.

[0080] According to some embodiments, the second determination module is configured to: calculate the attention weights between user features of multiple target evaluation users and user features of the target user based on the attention mechanism to obtain user similarities between the target user and the multiple target evaluation users.

[0081] According to some embodiments, the second generation module is configured to: determine the user weights of multiple target evaluation users based on the user similarities between the target user and the multiple target evaluation users; and based on the user weights of the multiple target evaluation users, perform weighted calculation on the historical evaluation information of the multiple target evaluation users to obtain the target user's predicted evaluation information for multiple candidate objects.

[0082] According to some embodiments, the historical evaluation information includes at least one evaluation tag, and the at least one evaluation tag is determined from a plurality of preset tags.

[0083] According to another aspect of the present disclosure, an electronic device is also provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned object recommendation method.

[0084] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is further provided, wherein the computer instructions are used to enable a computer to execute the above-mentioned object recommendation method.

[0085] See also Figure 8 , a block diagram of an electronic device 800 that can be used as the present disclosure will now be described, which is an example of a hardware device that can be applied to various aspects of the present disclosure. The electronic device can be a computer device of different types, such as a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0086] Figure 8 1 shows a block diagram of an electronic device according to an embodiment of the present disclosure. Figure 8 As shown, the electronic device 800 may include at least one processor 801 , a working memory 802 , an I / O device 804 , a display device 805 , a storage device 806 , and a communication interface 807 that can communicate with each other via a system bus 803 .

[0087] The processor 801 may be a single processing unit or multiple processing units, all of which may include a single or multiple computing units or multiple cores. The processor 801 may be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuits, and / or any device that manipulates signals based on operating instructions. The processor 801 may be configured to obtain and execute computer-readable instructions stored in a working memory 802, a storage device 806, or other computer-readable media, such as program codes of an operating system 802a, program codes of an application program 802b, and the like.

[0088] The working memory 802 and the storage device 806 are examples of computer-readable storage media for storing instructions, which are executed by the processor 801 to implement the various functions described above. The working memory 802 may include both volatile memory and non-volatile memory (e.g., RAM, ROM, etc.). In addition, the storage device 806 may include a hard disk drive, a solid-state drive, a removable medium, including external and removable drives, a memory card, a flash memory, a floppy disk, an optical disk (e.g., a CD, a DVD), a storage array, a network attached storage, a storage area network, etc. The working memory 802 and the storage device 806 may all be collectively referred to herein as memory or computer-readable storage media, and may be a non-transitory medium capable of storing computer-readable, processor-executable program instructions as computer program code, which may be executed by the processor 801 as a specific machine configured to implement the operations and functions described in the examples herein.

[0089] The I / O device 804 may include an input device and / or an output device. The input device may be any type of device capable of inputting information to the electronic device 800, and may include but is not limited to a mouse, a keyboard, a touch screen, a track pad, a track ball, a joystick, a microphone, and / or a remote controller. The output device may be any type of device capable of presenting information, and may include but is not limited to a video / audio output terminal, a vibrator, and / or a printer.

[0090] The communication interface 807 allows the electronic device 800 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks, and may include but is not limited to a modem, a network card, an infrared communication device, a wireless communication transceiver and / or a chipset, such as a Bluetooth™ device, an 802.11 device, a WiFi device, a WiMax device, a cellular communication device and / or the like.

[0091] The application 802b in the working register 802 can be loaded to execute the various methods and processes described above, such as Figure 1 In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 800 via the storage device 806 and / or the communication interface 807. When the computer program is loaded and executed by the processor 801, one or more steps of the object recommendation method described above may be performed.

[0092] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0093] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0094] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0095] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0096] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0097] A computing system may include clients and servers. Clients and servers are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship to each other.

[0098] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0099] Although the embodiments or examples of the present disclosure have been described with reference to the accompanying drawings, it should be understood that the above-mentioned methods, systems and devices are merely exemplary embodiments or examples, and the scope of the present invention is not limited by these embodiments or examples, but only by the claims after authorization and their equivalent scope. Various elements in the embodiments or examples can be omitted or replaced by their equivalent elements. In addition, each step can be performed in an order different from that described in the present disclosure. Further, the various elements in the embodiments or examples can be combined in various ways. It is important that with the evolution of technology, many elements described herein can be replaced by equivalent elements that appear after the present disclosure.

Claims

1. An object recommendation method, comprising: Acquire predicted evaluation information of a target user for a plurality of candidate objects, wherein the predicted evaluation information is generated based on a user feature of the target user, user features of a plurality of target evaluation users, and historical evaluation information of the plurality of target evaluation users for the plurality of candidate objects; Based on the predicted evaluation information of the target user for the multiple candidate objects, a recommended object for the target user is determined from the multiple candidate objects.

2. The method according to claim 1, wherein the process of generating the prediction evaluation information comprises: Clustering the multiple target evaluation users based on user characteristics of the multiple target evaluation users to obtain multiple evaluation user groups; Determining group user characteristics of each evaluation user group based on user characteristics of target evaluation users in the same evaluation user group; Based on the user characteristics of the target user, the group user characteristics of the plurality of evaluation user groups, and the historical evaluation information of the plurality of target evaluation users for the plurality of candidate objects, predicted evaluation information of the target user for the plurality of candidate objects is generated.

3. The method of claim 2, wherein: The generating the predicted evaluation information of the target user for the multiple candidate objects based on the user characteristics of the target user, the group user characteristics of the multiple evaluation user groups, and the historical evaluation information of the multiple target evaluation users for the multiple candidate objects comprises: Determining group evaluation features of each evaluation user group based on historical evaluation information of target evaluation users in the same evaluation user group; and Based on the user characteristics of the target user, the group user characteristics of the plurality of evaluation user groups, and the group evaluation characteristics of the plurality of evaluation user groups, evaluation information of the target user for the plurality of candidate objects is generated.

4. The method of claim 3, wherein: The determining of the group evaluation features of each evaluation user group based on the historical evaluation information of the target evaluation users in the same evaluation user group includes: Determining evaluation feature vectors corresponding to the historical evaluation information of the plurality of target evaluation users; and Based on the evaluation feature vectors corresponding to the historical evaluation information of the target evaluation users in the same evaluation user group, pooling calculation is performed to obtain the group evaluation features of each evaluation user group.

5. The method according to any one of claims 1 to 4, wherein: The generation process of the prediction evaluation information includes: Determining user similarities between the target user and the multiple target evaluation users based on the user characteristics of the target user and the user characteristics of the multiple target evaluation users; and Based on the user similarities between the target user and the plurality of target evaluation users and the historical evaluation information of the plurality of target evaluation users for the plurality of candidate objects, predicted evaluation information of the target user for the plurality of candidate objects is generated.

6. The method of claim 5, wherein: The determining the user similarity between the target user and the multiple target evaluation users based on the user characteristics of the target user and the user characteristics of the multiple target evaluation users comprises: The attention weights between the user features of the multiple target evaluation users and the user features of the target user are calculated based on the attention mechanism to obtain the user similarity between the target user and the multiple target evaluation users.

7. The method of claim 5, wherein: The generating the predicted evaluation information of the target user for the multiple candidate objects based on the user similarity between the target user and the multiple target evaluation users and the historical evaluation information of the multiple target evaluation users for the multiple candidate objects comprises: determining user weights of the multiple target evaluation users based on user similarities between the target user and the multiple target evaluation users; and Based on the user weights of the multiple target evaluation users, weighted calculation is performed on the historical evaluation information of the multiple target evaluation users to obtain the predicted evaluation information of the target users for the multiple candidate objects.

8. The method according to any one of claims 1 to 4, wherein: The historical evaluation information includes at least one evaluation tag, and the at least one evaluation tag is determined from a plurality of preset tags.

9. An object recommendation device, comprising: an acquisition unit configured to acquire predicted evaluation information of a target user for a plurality of candidate objects, wherein the predicted evaluation information is generated based on a user feature of the target user, user features of a plurality of target evaluation users, and historical evaluation information of the plurality of target evaluation users for the plurality of candidate objects; The determining unit is configured to determine a recommended object for the target user from the multiple candidate objects based on the predicted evaluation information of the target user for the multiple candidate objects.

10. An electronic device, comprising: at least one processor; as well as a memory communicatively coupled to the at least one processor; in The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 8.

11. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-8.

Images