Training method of preference prediction model and prediction method of user preference

Abstract

The embodiment of the present application provides a method for training a preference prediction model and a method for predicting user preferences. The method includes: obtaining historical evaluation data of users on an object; determining whether the amount of the historical evaluation data meets a preset condition; constructing a training set based on whether the amount of the historical evaluation data meets the preset condition; training a preference prediction model based on the training set, wherein the preference prediction model is established based on a restricted Boltzmann machine model. The preference prediction model trained based on this solution can better extract the user's hidden features to perform preference prediction and improve the accuracy of user preference prediction.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method for training a preference prediction model and a method for predicting user preferences. Background Art

[0002] At present, various products are emerging one after another. The number of active users is an important indicator for evaluating products. How to maintain the rapid growth of active users has become a problem that many companies need to think about and solve. Predicting user preferences and catering to their needs is a good solution.

[0003] One way to predict user preferences is based on observed features (Feature-based) method. The idea of ​​this method is to directly use the observed features of users as input, and predict user preferences by establishing models such as linear regression, Bayesian network, support vector machine and decision tree.

[0004] The method of predicting preferences based on observed features has a certain bottleneck, that is, the prediction granularity is limited. For example, when predicting users' preferences for different videos, the prediction cannot be accurate to the (user, video) pair. For example, if the observable features of two users are the same, the prediction results for the same video are the same, but this prediction result may not be reasonable, and the user's preferences may actually be different. Therefore, it is urgent to provide a method that can improve the accuracy of user preference prediction. Summary of the invention

[0005] The purpose of this application is to solve at least one of the above technical defects. The technical solution adopted by this application is as follows:

[0006] In a first aspect, an embodiment of the present application provides a method for training a preference prediction model, the method comprising:

[0007] Obtain historical evaluation data of users on objects;

[0008] Determine whether the amount of historical evaluation data meets the preset conditions;

[0009] Construct a training set based on whether the amount of historical evaluation data meets the preset conditions;

[0010] The preference prediction model is trained based on the training set, and the preference prediction model is established based on the restricted Boltzmann machine model.

[0011] Optionally, based on whether the amount of historical evaluation data meets a preset condition, a training set is constructed, including:

[0012] If the amount of historical evaluation data does not meet the preset conditions, a training set is constructed based on the historical evaluation data and observable features;

[0013] If the amount of historical evaluation data meets the preset conditions, a training set is constructed based on the historical evaluation data.

[0014] Optionally, training a preference prediction model based on a training set includes:

[0015] Based on the training set, the preference prediction model is trained according to the collaborative filtering algorithm.

[0016] Optionally, the observable feature includes a user feature, and the user feature includes at least one of the following:

[0017] User age;

[0018] User's gender;

[0019] User occupation.

[0020] Optionally, the observable feature further includes an object feature. If the object is a video, the object feature includes at least one of the following:

[0021] Keywords extracted from the video name of the video;

[0022] Time information taken from the year the video was released.

[0023] Optionally, if the amount of historical evaluation data does not meet the preset condition, after training the preference prediction model, the method further includes:

[0024] According to the preset retry rules, the historical evaluation data of users on the object is repeatedly obtained until the amount of historical evaluation data obtained meets the preset conditions, the training set is updated based on the historical evaluation data, and the preference prediction model is trained based on the updated training set.

[0025] In a second aspect, an embodiment of the present application provides a method for predicting user preferences, the method comprising:

[0026] Obtain historical evaluation data of users on objects;

[0027] The historical evaluation data is input into a pre-trained preference prediction model to obtain a user's preference prediction result, wherein the preference prediction model is trained according to the above-mentioned preference prediction model training method.

[0028] In a third aspect, an embodiment of the present application provides a training device for a preference prediction model, the device comprising:

[0029] A data acquisition module is used to obtain historical evaluation data of users on objects;

[0030] A data volume comparison module is used to determine whether the data volume of historical evaluation data meets the preset conditions;

[0031] A training set construction module, used to construct a training set based on whether the amount of historical evaluation data meets preset conditions;

[0032] The model training module is used to train the preference prediction model based on the training set, and the preference prediction model is established based on the restricted Boltzmann machine model.

[0033] Optionally, the training set construction module is specifically used for:

[0034] If the amount of historical evaluation data does not meet the preset conditions, a training set is constructed based on the historical evaluation data and observable features;

[0035] If the amount of historical evaluation data meets the preset conditions, a training set is constructed based on the historical evaluation data.

[0036] Optionally, the model training module is specifically used for:

[0037] Based on the training set, the preference prediction model is trained according to the collaborative filtering algorithm.

[0038] Optionally, the observable feature includes a user feature, and the user feature includes at least one of the following:

[0039] User age;

[0040] User's gender;

[0041] User occupation.

[0042] Optionally, the observable feature further includes an object feature. If the object is a video, the object feature includes at least one of the following:

[0043] Keywords extracted from the video name of the video;

[0044] Time information taken from the year the video was released.

[0045] Optionally, if the amount of historical evaluation data does not meet the preset conditions, the above model training module is also used to:

[0046] After the preference prediction model is trained, the historical evaluation data of users on the object is repeatedly obtained according to the preset retry rules until the amount of historical evaluation data obtained meets the preset conditions, the training set is updated based on the historical evaluation data, and the preference prediction model is trained based on the updated training set.

[0047] In a fourth aspect, an embodiment of the present application provides a device for predicting user preferences, the device comprising:

[0048] An evaluation data acquisition module is used to obtain historical evaluation data of users on an object;

[0049] The preference prediction module is used to input historical evaluation data into a pre-trained preference prediction model to obtain a user's preference prediction result, wherein the preference prediction model is trained according to the above-mentioned preference prediction model training method.

[0050] In a fifth aspect, an embodiment of the present application provides an electronic device, the electronic device comprising: a processor and a memory;

[0051] A memory, used for storing operation instructions;

[0052] The processor is used to execute the method shown in any implementation of the first aspect or any implementation of the second aspect of the present application by calling an operation instruction.

[0053] In a sixth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method shown in any implementation of the first aspect or any implementation of the second aspect of the present application.

[0054] The beneficial effects of the technical solution provided by the embodiment of the present application are:

[0055] The solution provided in the embodiment of the present application obtains the historical evaluation data of the user on the object, determines whether the data volume of the historical evaluation data meets the preset conditions, constructs a training set based on whether the data volume of the historical evaluation data meets the preset conditions, trains the preference prediction model based on the training set, and the preference prediction model is established based on the restricted Boltzmann machine model. The preference prediction model trained based on this solution can better extract the hidden features of the user to perform preference prediction, thereby improving the accuracy of user preference prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in describing the embodiments of the present application are briefly introduced below.

[0057] Figure 1 A flowchart of a method for training a preference prediction model provided in an embodiment of the present application;

[0058] Figure 2 A flowchart of a method for predicting user preferences provided in an embodiment of the present application;

[0059] Figure 3 A schematic diagram of the structure of a training device for a preference prediction model provided in an embodiment of the present application;

[0060] Figure 4 A schematic diagram of the structure of a user preference prediction device provided in an embodiment of the present application;

[0061] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0062] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present invention.

[0063] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be an intermediate element. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0064] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0065] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0066] Figure 1 A flow chart of a method for training a preference prediction model provided in an embodiment of the present application is shown, such as Figure 1 As shown, the method may mainly include:

[0067] Step S110: Obtaining historical evaluation data of the user on the object;

[0068] Step S120: Determine whether the amount of historical evaluation data meets a preset condition;

[0069] Step S130: constructing a training set based on whether the amount of historical evaluation data meets a preset condition;

[0070] Step S140: training a preference prediction model based on the training set, where the preference prediction model is established based on a restricted Boltzmann machine model.

[0071] In the embodiment of the present application, the object may be a product, such as a video, etc., and the historical evaluation data may be a user's rating of the object.

[0072] Hidden features play a very important role in predicting user preferences. For example, two users may give similar ratings to videos, but in reality, they have similar hidden features, such as both like comedies but not horror movies. Therefore, in solving the problem of predicting user preferences, extracting hidden features of users based on observed user preferences is a critical part of the problem.

[0073] In an embodiment of the present application, the preset condition may be a preset amount of data. If the historical evaluation data is small, the trained prediction model may have poor prediction effect. Therefore, before training the prediction model based on the historical evaluation data, it is possible to determine whether the amount of the historical evaluation data meets the preset condition. A training set is constructed based on whether the amount of the historical evaluation data meets the preset condition to ensure the prediction effect of the trained prediction model.

[0074] Restricted Boltzmann Machine (RBM) can abstract features in depth through superposition and can flexibly handle probability generation models. Therefore, in order to better extract the model of user hidden features to predict user preferences, the embodiment of the present application predicts user preferences by establishing a restricted Boltzmann machine model.

[0075] The method provided in the embodiment of the present application obtains the historical evaluation data of the user on the object, determines whether the data volume of the historical evaluation data meets the preset conditions, constructs a training set based on whether the data volume of the historical evaluation data meets the preset conditions, and trains the preference prediction model based on the training set. The preference prediction model is established based on the restricted Boltzmann machine model. The preference prediction model obtained by training based on this solution can better extract the hidden features of the user to perform preference prediction, thereby improving the accuracy of user preference prediction.

[0076] In an optional manner of the embodiment of the present application, a training set is constructed based on whether the amount of historical evaluation data meets a preset condition, including:

[0077] If the amount of historical evaluation data does not meet the preset conditions, a training set is constructed based on the historical evaluation data and observable features;

[0078] If the amount of historical evaluation data meets the preset conditions, a training set is constructed based on the historical evaluation data.

[0079] In an embodiment of the present application, before training a prediction model based on historical evaluation data, it can be determined whether the data amount of the historical evaluation data meets the preset conditions. If the data amount of the historical evaluation data meets the preset conditions, it can be considered that the data amount of the historical evaluation data used for model training can meet the needs of model training, and a training set can be constructed based on the historical evaluation data.

[0080] If the amount of historical evaluation data does not meet the preset conditions, it can be considered that the amount of historical evaluation data used for model training cannot meet the needs of model training. In order to ensure the predictive effect of the trained prediction model, observable features can be added for model training, and a training set can be constructed together with observable features and historical evaluation data.

[0081] Since observable features are introduced, even if the amount of historical evaluation data is small, hidden features can be better predicted.

[0082] In an optional manner of the embodiment of the present application, training a preference prediction model based on a training set includes:

[0083] Based on the training set, the preference prediction model is trained according to the collaborative filtering algorithm.

[0084] Since in the historical evaluation data, if all users have rated all the products, it can be concluded that this RBM has M visible nodes. However, in actual situations, most of the ratings in the rating matrix are missing. In order to cope with the situation of missing ratings, the RBM is trained according to the collaborative filtering algorithm in the embodiment of the present application. Specifically, an RBM can be trained for each user, and all these RBMs have the same number of hidden nodes, but the number of visible nodes of each RBM is the number of products in a user's rating. In this way, although all RBMs have only one training example, the deviations and weights of the same hidden nodes and videos are tied together for all RBMs. For example, if two users rate the same video, then for both RBMs, the weight value between the video and the hidden node is the same.

[0085] In the embodiment of the present application, the historical evaluation data can be binarized, that is, all nodes in the RBM have only two states: 0 and 1. Constructing a binarized RBM can simplify the training and prediction process.

[0086] In an optional manner of the embodiment of the present application, the observable feature includes a user feature, and the user feature includes at least one of the following:

[0087] User age;

[0088] User's gender;

[0089] User occupation.

[0090] In an embodiment of the present application, user-related information may be used as observable features, such as user age, user gender, and user occupation as user features.

[0091] In an optional manner of the embodiment of the present application, the observable feature further includes an object feature. If the object is a video, the object feature includes at least one of the following:

[0092] Keywords extracted from the video name of the video;

[0093] Time information taken from the year the video was released.

[0094] In an embodiment of the present application, observable features may also include object features. If the object is a video, the object features may be extracted from relevant information of the video, such as extracting some keywords that can represent the features of the video from the video name, or obtaining time information from the year of release.

[0095] In an optional manner of the embodiment of the present application, if the amount of historical evaluation data does not meet the preset condition, after training the preference prediction model, the method further includes:

[0096] According to the preset retry rules, the historical evaluation data of users on the object is repeatedly obtained until the amount of historical evaluation data obtained meets the preset conditions, the training set is updated based on the historical evaluation data, and the preference prediction model is trained based on the updated training set.

[0097] In an embodiment of the present application, after the preference prediction model is trained, attempts can be made to repeatedly obtain historical evaluation data, and the amount of historical evaluation data obtained can be detected. When it is determined that the amount of historical evaluation data obtained meets a preset condition, a training set can be reconstructed based on the currently obtained historical evaluation data, and the preference prediction model can be retrained based on the reconstructed training set.

[0098] As an example, the observable features can be preprocessed to digitize the user features and video features. In order to facilitate the calculation in the model, in this example, both the user features and the video features are mapped to binary vectors. The specific relationship is as follows:

[0099] Age: In this example, age is divided into five intervals [-17][18-24][25-34][35-44][44-], and a binary vector of length 5 is used to represent the user's age interval;

[0100] Gender: Use a binary vector of length 2 to represent gender: (1, 0) for male and (0, 1) for female.

[0101] Occupation: It is represented by a binary vector of length 21. If the user's occupation is the i-th, the i-th element in the vector is 1, and the other elements are 0;

[0102] Movie type: It is represented by a binary vector of length 19. If the movie type of the video is the jth one, the jth element in the vector is 1, and the other elements are 0.

[0103] After data preprocessing, a vector with a length of 28 can be obtained as an observable feature for each user; a vector with a length of 19 can be obtained as an observable feature for each video.

[0104] Figure 2 A schematic diagram of a user preference prediction method provided in an embodiment of the present application is shown. Figure 2 As shown, the method may mainly include:

[0105] Step S210: Obtaining historical evaluation data of the user on the object;

[0106] Step S220: inputting the historical evaluation data into a pre-trained preference prediction model to obtain a user preference prediction result, wherein the preference prediction model is trained according to the above-mentioned preference prediction model training method.

[0107] In the embodiment of the present application, the object may be a product, such as a video, etc., and the historical evaluation data may be a user's rating of the object.

[0108] When predicting user preferences, the user's historical evaluation data can be input into the pre-trained preference prediction model. Since the prediction model is based on the restricted Boltzmann machine model, it can perform deep abstraction of features through superposition, and better extract the model of user hidden features to predict user preferences, making the predicted user preferences more accurate.

[0109] The method provided in the embodiment of the present application obtains the historical evaluation data of the user on the object, inputs the historical evaluation data into the pre-trained preference prediction model, and obtains the user's preference prediction result. Based on this solution, since the user preference prediction is performed based on the preference prediction model established by the restricted Boltzmann machine model, the user's hidden features can be better extracted to perform preference prediction, thereby improving the accuracy of user preference prediction.

[0110] Based on Figure 1 The same principle as shown in the method, Figure 3 A schematic diagram of the structure of a training device for a preference prediction model provided in an embodiment of the present application is shown. Figure 3 As shown, the training device 30 of the preference prediction model may include:

[0111] The data acquisition module 310 is used to acquire the historical evaluation data of the user on the object;

[0112] A data volume comparison module 320 is used to determine whether the data volume of the historical evaluation data meets a preset condition;

[0113] A training set construction module 330, for constructing a training set based on whether the amount of historical evaluation data meets a preset condition;

[0114] The model training module 340 is used to train a preference prediction model based on a training set, where the preference prediction model is established based on a restricted Boltzmann machine model.

[0115] The device provided in the embodiment of the present application obtains the historical evaluation data of the user on the object, determines whether the data volume of the historical evaluation data meets the preset conditions, constructs a training set based on whether the data volume of the historical evaluation data meets the preset conditions, and trains the preference prediction model based on the training set, and the preference prediction model is established based on the restricted Boltzmann machine model. The preference prediction model trained based on this solution can better extract the hidden features of the user to perform preference prediction, thereby improving the accuracy of user preference prediction.

[0116] Optionally, the training set construction module is specifically used for:

[0117] If the amount of historical evaluation data does not meet the preset conditions, a training set is constructed based on the historical evaluation data and observable features;

[0118] If the amount of historical evaluation data meets the preset conditions, a training set is constructed based on the historical evaluation data.

[0119] Optionally, the model training module is specifically used for:

[0120] Based on the training set, the preference prediction model is trained according to the collaborative filtering algorithm.

[0121] Optionally, the observable feature includes a user feature, and the user feature includes at least one of the following:

[0122] User age;

[0123] User's gender;

[0124] User occupation.

[0125] Optionally, the observable feature further includes an object feature. If the object is a video, the object feature includes at least one of the following:

[0126] Keywords extracted from the video name of the video;

[0127] Time information taken from the year the video was released.

[0128] Optionally, if the amount of historical evaluation data does not meet the preset conditions, the above model training module is also used to:

[0129] After the preference prediction model is trained, the historical evaluation data of users on the object is repeatedly obtained according to the preset retry rules until the amount of historical evaluation data obtained meets the preset conditions, the training set is updated based on the historical evaluation data, and the preference prediction model is trained based on the updated training set.

[0130] It can be understood that the above modules of the training device for the preference prediction model in this embodiment have the function of realizing Figure 1 The functions of the corresponding steps of the training method of the preference prediction model in the embodiment shown in . The function can be implemented by hardware, or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions. The above modules can be software and / or hardware, and the above modules can be implemented separately or integrated with multiple modules. For the functional description of each module of the training device of the preference prediction model, please refer to Figure 1 The corresponding description of the training method of the preference prediction model in the embodiment shown in will not be repeated here.

[0131] Based on Figure 2 The same principle as shown in the method, Figure 4 A schematic diagram of the structure of a user preference prediction device provided in an embodiment of the present application is shown. Figure 4 As shown, the user preference prediction device 40 may include:

[0132] Evaluation data acquisition module 410, used to obtain historical evaluation data of users on the object;

[0133] The preference prediction module 420 is used to input the historical evaluation data into the pre-trained preference prediction model to obtain the user's preference prediction result, wherein the preference prediction model is trained according to the above-mentioned preference prediction model training method.

[0134] The device provided in the embodiment of the present application obtains the historical evaluation data of the user on the object, inputs the historical evaluation data into the pre-trained preference prediction model, and obtains the user's preference prediction result. Based on this solution, since the user preference prediction is performed based on the preference prediction model established by the restricted Boltzmann machine model, the user's hidden features can be better extracted to perform preference prediction, thereby improving the accuracy of user preference prediction.

[0135] It can be understood that the above modules of the user preference prediction device in this embodiment have the function of realizing Figure 2The functions of the corresponding steps of the method for predicting user preferences in the embodiment shown in . The functions can be implemented by hardware, or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions. The above modules can be software and / or hardware, and the above modules can be implemented separately or integrated with multiple modules. For the functional description of each module of the above user preference prediction device, please refer to Figure 2 The corresponding description of the method for predicting user preferences in the embodiment shown in is not repeated here.

[0136] An embodiment of the present application provides an electronic device, including a processor and a memory;

[0137] A memory, used for storing operation instructions;

[0138] The processor is used to execute the method provided in any embodiment of the present application by calling an operation instruction.

[0139] As an example, Figure 5 A schematic diagram of the structure of an electronic device applicable to the embodiment of the present application is shown. Figure 5 As shown, the electronic device 2000 includes: a processor 2001 and a memory 2003. The processor 2001 and the memory 2003 are connected, such as through a bus 2002. Optionally, the electronic device 2000 may also include a transceiver 2004. It should be noted that in actual applications, the transceiver 2004 is not limited to one, and the structure of the electronic device 2000 does not constitute a limitation on the embodiments of the present application.

[0140] The processor 2001 is applied in the embodiment of the present application to implement the method shown in the above method embodiment. The transceiver 2004 may include a receiver and a transmitter. The transceiver 2004 is applied in the embodiment of the present application to implement the function of the electronic device of the embodiment of the present application communicating with other devices when executed.

[0141] Processor 2001 may be a CPU (Central Processing Unit), a general purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 2001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0142] The bus 2002 may include a path to transmit information between the above components. The bus 2002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 2002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0143] The memory 2003 can be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.

[0144] Optionally, the memory 2003 is used to store application code for executing the solution of the present application, and the execution is controlled by the processor 2001. The processor 2001 is used to execute the application code stored in the memory 2003 to implement the method provided in any embodiment of the present application.

[0145] The electronic device provided in the embodiment of the present application is applicable to any embodiment of the above method and will not be described in detail here.

[0146] The embodiment of the present application provides an electronic device, which, compared with the prior art, obtains the historical evaluation data of the user on the object, determines whether the data volume of the historical evaluation data meets the preset conditions, constructs a training set based on whether the data volume of the historical evaluation data meets the preset conditions, trains a preference prediction model based on the training set, and the preference prediction model is established based on the restricted Boltzmann machine model. The preference prediction model trained based on this solution can better extract the hidden features of the user to perform preference prediction, thereby improving the accuracy of user preference prediction.

[0147] An embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the method shown in the above method embodiment is implemented.

[0148] The computer-readable storage medium provided in the embodiments of the present application is applicable to any embodiment of the above method and will not be described in detail here.

[0149] The embodiment of the present application provides a computer-readable storage medium. Compared with the prior art, by obtaining the historical evaluation data of the user on the object, determining whether the data volume of the historical evaluation data meets the preset conditions, constructing a training set based on whether the data volume of the historical evaluation data meets the preset conditions, training a preference prediction model based on the training set, and establishing the preference prediction model based on the restricted Boltzmann machine model. The preference prediction model trained based on this solution can better extract the hidden features of the user to perform preference prediction, thereby improving the accuracy of user preference prediction.

[0150] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0151] The above are only some embodiments of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for training a preference prediction model, characterized in that: include: Obtain historical evaluation data of users on objects; Determining whether the amount of the historical evaluation data meets a preset condition; Constructing a training set based on whether the amount of the historical evaluation data meets a preset condition; Training a preference prediction model based on the training set, wherein the preference prediction model is established based on a restricted Boltzmann machine model; The step of constructing a training set based on whether the amount of the historical evaluation data meets a preset condition comprises: If the amount of the historical evaluation data does not meet the preset conditions, constructing a training set based on the historical evaluation data and observable features; If the amount of the historical evaluation data meets a preset condition, constructing a training set based on the historical evaluation data; The observable features include at least one of the following: user age; user gender; user occupation; the observable features also include object features. If the object is a video, the object features include at least one of the following: keywords extracted from the video name of the video; time information obtained from the year of release of the video.

2. The method according to claim 1, characterized in that: The training preference prediction model based on the training set includes: Based on the training set, a preference prediction model is trained according to a collaborative filtering algorithm.

3. The method according to claim 1, characterized in that If the amount of the historical evaluation data does not meet the preset condition, after the preference prediction model is trained, the method further includes: According to the preset retry rules, the historical evaluation data of the user on the object is repeatedly obtained until the amount of the historical evaluation data obtained meets the preset conditions, the training set is updated based on the historical evaluation data, and the preference prediction model is trained based on the updated training set.

4. A method for predicting user preferences, characterized in that: include: Obtain historical evaluation data of users on objects; The historical evaluation data is input into a pre-trained preference prediction model to obtain the preference prediction result of the user, wherein the preference prediction model is trained according to the training method of the preference prediction model according to any one of claims 1 to 3.

5. A training device for a preference prediction model, characterized in that: include: A data acquisition module is used to obtain historical evaluation data of users on objects; A data volume comparison module, used to determine whether the data volume of the historical evaluation data meets a preset condition; A training set construction module, used to construct a training set based on whether the data volume of the historical evaluation data meets a preset condition; A model training module, used for training a preference prediction model based on the training set, wherein the preference prediction model is established based on a restricted Boltzmann machine model; The step of constructing a training set based on whether the amount of the historical evaluation data meets a preset condition comprises: If the amount of the historical evaluation data does not meet the preset conditions, constructing a training set based on the historical evaluation data and observable features; If the amount of the historical evaluation data meets a preset condition, constructing a training set based on the historical evaluation data; The observable features include at least one of the following: user age; user gender; user occupation; the observable features also include object features. If the object is a video, the object features include at least one of the following: keywords extracted from the video name of the video; time information obtained from the year of release of the video.

6. A device for predicting user preferences, characterized in that: include: An evaluation data acquisition module is used to obtain historical evaluation data of users on an object; A preference prediction module is used to input the historical evaluation data into a pre-trained preference prediction model to obtain the preference prediction result of the user, wherein the preference prediction model is trained according to the training method of the preference prediction model according to any one of claims 1 to 3.

7. An electronic device, characterized in that: including a processor and a memory; The memory is used to store operation instructions; The processor is used to execute the method according to any one of claims 1 to 4 by calling the operation instruction.

8. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images