Book recommendation method, electronic device and storage medium

Abstract

The disclosed embodiments relate to a book recommendation method, electronic device and storage medium. The book recommendation method includes: performing a first screening process on a first book data set based on a first network in a preset recommendation model to obtain a second book data set; performing a second screening process on a second book data set based on a second network in a preset recommendation model to obtain a target book data set; and pushing books to users based on the target book data set. According to the disclosed embodiments, the book data is quickly screened for the first time through the first network, and the second network performs a second screening based on the book data screened by the first network. Since the amount of data is reduced, the screening speed is greatly improved, the system overhead is reduced, the recommendation speed is improved while ensuring the accuracy of the recommendation, the real-time nature of the book recommendation is achieved, and the user's reading experience is improved.

Description

Technical Field

[0001] The disclosed embodiments relate to the field of information processing technology, and in particular to a book recommendation method, an electronic device, and a storage medium. Background Art

[0002] Books in the form of e-books are loved by a large number of users due to their advantages such as easy access. Book resources are massive. In order to facilitate users to choose books suitable for themselves, many book platforms provide book recommendation systems to recommend books to users. However, the sorting model used by the existing recommendation system is complex in structure, with high system overhead and poor real-time output results. It is impossible to achieve real-time book recommendations, resulting in poor book recommendation results. Therefore, a book recommendation method is urgently needed to achieve real-time book recommendations and improve the user's reading experience. Summary of the invention

[0003] In order to solve the above technical problems or at least partially solve the above technical problems, the embodiments of the present disclosure provide a book recommendation method, an electronic device and a storage medium.

[0004] A first aspect of an embodiment of the present disclosure provides a book recommendation method, the method comprising:

[0005] Performing a first screening process on the first book data set based on the first network in the preset recommendation model to obtain a second book data set;

[0006] Performing a second screening process on the second book dataset based on the second network in the preset recommendation model to obtain a target book dataset;

[0007] Based on the target book dataset, books are pushed to the user.

[0008] A second aspect of an embodiment of the present disclosure provides an electronic device, including a processor and a memory, wherein the memory is used to store executable instructions, and the executable instructions enable the processor to perform the following operations:

[0009] Performing a first screening process on the first book data set based on the first network in the preset recommendation model to obtain a second book data set;

[0010] Performing a second screening process on the second book dataset based on the second network in the preset recommendation model to obtain a target book dataset;

[0011] Based on the target book dataset, push books to users.

[0012] A third aspect of an embodiment of the present disclosure provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the processor implements the book recommendation method described in the first aspect.

[0013] Compared with the prior art, the technical solution provided by the embodiments of the present disclosure has the following advantages:

[0014] The disclosed embodiment performs a first screening process on the first book data set based on the first network in the preset recommendation model to obtain a second book data set; performs a second screening process on the second book data set based on the second network in the preset recommendation model to obtain a target book data set; and pushes books to users based on the target book data set. According to the disclosed embodiment, the book data is quickly screened for the first time through the first network, and the second network performs a second screening based on the book data screened by the first network. Due to the reduction in the amount of data, the screening speed is greatly improved, the system overhead is reduced, the recommendation speed is improved while ensuring the accuracy of the recommendation, the real-time nature of the book recommendation is achieved, and the user's reading experience is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0017] Figure 1 is a flow chart of a book recommendation method provided by an embodiment of the present disclosure;

[0018] Figure 2 is a flow chart of another book recommendation method provided by an embodiment of the present disclosure;

[0019] Figure 3 It is a schematic diagram of the principle of a deep decomposition machine DeepFM model provided by an embodiment of the present disclosure;

[0020] Figure 4 It is a structural schematic diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0021] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0022] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0023] Figure 1 is a flow chart of a book recommendation method provided by an embodiment of the present disclosure. The method can be executed by an electronic device, which can be understood as any device with computing and processing capabilities. Figure 1 As shown, the book recommendation method provided by the embodiment of the present disclosure includes the following steps:

[0024] Step 101: Perform a first screening process on a first book dataset based on a first network in a preset recommendation model to obtain a second book dataset.

[0025] The preset recommendation model referred to in the embodiment of the present disclosure can be understood as a deep factorization machine (Deep Factorization Machines, DeepFM) model, which includes a factorization machine (Factorization Machine, FM) network and a deep Deep network. Among them, the FM network is a machine learning algorithm based on matrix decomposition, which uses feature vectors to transform the factor decomposition network, and expresses the interaction relationship between factors through the inner product between low-order relationships. The FM network has the advantage of automatically combining features, which can greatly reduce the work of feature engineering. The Deep network can be understood as a deep neural network (Deep Neural Networks, DNN), which can take into account both high-order feature combinations and low-order feature combinations. The DeepFM model can combine the advantages of both FM networks and DNN networks, and can automatically generate low-order features and high-order features, thereby improving the processing efficiency of feature engineering.

[0026] The main functional modules of the common DeepFM model include sparse feature processing network, FM network, Deep network, and output network.

[0027] The book data set referred to in the embodiment of the present disclosure can be understood as a collection of book data of multiple books recalled by a preset recommendation model using a multi-way recall strategy during the recall phase. The multi-way recall strategy can be item collaborative recall, user collaborative recall, etc., which is not limited here.

[0028] The first network referred to in the embodiment of the present disclosure can be understood as an FM network, which has a simple structure and can quickly complete screening.

[0029] The book data in the embodiments of the present disclosure may be understood as a collection of multiple feature data of a book, which may include book feature data, user feature data, and feature data of the user's interaction behavior with the book, wherein the book feature data may include the book's identification information (such as the book's name, author, introduction, and other information), the book's publishing information, the book's category, and other feature data; the user feature data may include the identification information of users who have interacted with the book, the user's reading history list, the book's preference category, and other feature data; the user's interaction behavior feature data with the book may include the user's interaction duration with the book, the user's marking of the book, and other feature data.

[0030] In the disclosed embodiment, the first book data set is first screened based on the first network in the preset recommendation model. The DeepFM model can be used to first perform sparse feature processing on the book data of each book in the first book data set to obtain the feature vector of the book data of each book; the feature vector is input into the first network for embedding processing, the feature vector is cross-combined with each other, the dot product (also called dot product, dot multiplication) value between the cross-combined feature vectors is calculated, the dot product value is linearly processed to obtain the first score of the book data of each book, the book data in the first book data set is sorted in descending order according to the first score, and the first score sorting result is obtained; based on the book data in the first score sorting result whose ranking is higher than the first preset threshold, the second book data set is generated. It should be noted that the first screening processing method of the first book data set is only described here by taking the scoring method as an example, and it is not the only description.

[0031] Step 102: Perform a second screening process on the second book dataset based on the second network in the preset recommendation model to obtain a target book dataset.

[0032] The second network referred to in the embodiments of the present disclosure can be understood as a Deep network.

[0033] It should be noted that the screening accuracy of the first network is lower than that of the second network.

[0034] In the embodiment of the present disclosure, the step of performing a second screening process on the second book dataset based on the second network in the preset recommendation model to obtain the target book dataset may include steps S11-S12:

[0035] Step S11: score and sort the book data in the second book data set based on the second network to obtain a second scoring and sorting result.

[0036] In the disclosed embodiment, the book data of each book in the first book data set can be processed with sparse features and embedded through the first network in the DeepFM model to obtain the feature vector of the book data of each book, and the first book data set can be processed with first screening based on the feature vector to obtain the second book data set. The feature vector of the book data of each book in the second book data set can be fully connected through a plurality of hidden layers through the second network in the DeepFM model to obtain the second score of the book data of each book. The book data of the books in the second book data set can be sorted in descending order according to the second score to obtain the second score sorting result.

[0037] Step S12: Generate a target book data set based on the book data ranked higher than a second preset threshold in the second scoring and ranking result.

[0038] In the disclosed embodiment, book data ranked higher than a second preset threshold in the second scoring and ranking result is selected to generate a target book data set.

[0039] Step 103: Push books to users based on the target book dataset.

[0040] In the disclosed embodiment, book data with a second score ranking higher than a second preset threshold may be pushed to the user according to the second score ranking result of the target book data set.

[0041] Compared with the prior art, the technical solution provided by the embodiments of the present disclosure has the following advantages:

[0042] The disclosed embodiment performs a first screening process on the first book data set based on the first network in the preset recommendation model to obtain a second book data set; performs a second screening process on the second book data set based on the second network in the preset recommendation model to obtain a target book data set; and pushes books to users based on the target book data set. According to the disclosed embodiment, the book data is quickly screened for the first time through the first network, and the second network performs a second screening based on the book data screened by the first network. Due to the reduction in the amount of data, the screening speed is greatly improved, the system overhead is reduced, the recommendation speed is improved while ensuring the accuracy of the recommendation, the real-time nature of the book recommendation is achieved, and the user's reading experience is improved.

[0043] Figure 2 is a flowchart of another book recommendation method provided by an embodiment of the present disclosure. The method can be executed by an electronic device, which can be understood as any device with computing and processing capabilities. Figure 2 As shown, the book recommendation method provided by the embodiment of the present disclosure includes the following steps:

[0044] Step 201: score and sort the book data in the first book data set based on the first network to obtain a first scoring and sorting result.

[0045] In the disclosed embodiment, before scoring and sorting the book data in the first book data set based on the first network, the book data of each book in the first book data set must first be sparsely processed to obtain a feature vector of the book data of each book. The feature vector is input into the first network for embedding processing, the feature vectors are cross-combined with each other, the dot product (also called dot product, dot multiplication) value between the cross-combined feature vectors is calculated, the dot product value is linearly processed to obtain the first score of the book data of each book, and the book data in the first book data set is sorted in descending order according to the first score to obtain the first score sorting result.

[0046] Step 202: Generate a second book data set based on the book data ranked higher than a first preset threshold in the first scoring and ranking result.

[0047] In the disclosed embodiment, the book data ranked higher than the first preset threshold in the first scoring and sorting result is selected to generate the second book data set. For example, assuming that there are book data of 100 books in the first book data set, the book data of the 100 books are respectively scored by the first scoring in step 201, and the scoring range is 0-1. The book data of the 100 books in the first book data set are sorted in descending order according to the first scores. Assuming that the first preset threshold is 60, the book data ranked from 1 to 60 in the first book data set are added to the data set to obtain the second book data set. It can be understood that this is only an exemplary description of generating the second book data set, and not the only description.

[0048] Step 203: score and sort the book data in the second book data set based on the second network to obtain a second scoring and sorting result.

[0049] The embodiments of the present disclosure can refer to Figure 1 Step S11 is not described here any more.

[0050] Step 204: Generate a target book data set based on the book data ranked higher than a second preset threshold in the second scoring and ranking result.

[0051] In the disclosed embodiment, the book data ranked higher than the second preset threshold in the second scoring sorting result is selected to generate a target book data set. For example, the book data of 60 books are screened out from the book data of 100 books in the first book data set through the above steps to form the second book data set, and the book data of the 60 books are respectively given a second score through step 203, with a score range of 0-1, and the book data of the 60 books in the second book data set are sorted in descending order according to the second score. Assuming that the second preset threshold is 20, the book data ranked from 1 to 20 in the second book data set are added to the data set to obtain the target book data set. It can be understood that this is only an exemplary description of generating a target book data set, and not the only description.

[0052] Step 205 : For the book data in the target book data set, the score of the first network corresponding to the book data and the score of the second network corresponding to the book data are summed to obtain the total score of the book data.

[0053] In the disclosed embodiment, the scores of the first network and the second network corresponding to the book data in the target book data set are input into the output network of the DeepFM model, and the summation is performed in combination with the activation function Sigmoid in the model to obtain the total score of the book data. For example, the first score of book data A in the target book data set is 0.7 points through the first network, and the second score through the second network is 0.9 points. The first score and the second score of book data A are summed to obtain a total score of book data A of 1.6 points. It can be understood that this is only an exemplary description of obtaining the total score of book data, not the only description.

[0054] Step 206: Sort the book data in the target book data set based on the total scores of the book data in the target book data set.

[0055] The book data in the target book data set are sorted in descending order according to the total scores of the book data in the target book data set to obtain a total score sorting result of the book data in the target book data set.

[0056] Step 207: Push the book data whose total score ranking is higher than the third preset threshold to the user.

[0057] In the disclosed embodiment, the book data in the target book data set whose total score ranking is higher than the third preset threshold value is pushed to the user. For example, there are book data of 20 books in the target book data set. The book data of the 20 books are respectively subjected to the summation of the first scoring and the second scoring through the above steps to obtain the total score of the book data of each book, and the book data of the 20 books in the target book data set are sorted in descending order of the total score. Assuming that the third preset threshold value is 5, the book data of the books ranked 1 to 5 in the target book data set are pushed to the user. It can be understood that this is only an exemplary description of pushing book data to users based on the total score, and not the only description.

[0058] Compared with the prior art, the technical solution provided by the embodiments of the present disclosure has the following advantages:

[0059] The disclosed embodiment scores and sorts the book data in the first book data set based on the first network to obtain a first scoring and sorting result; generates a second book data set based on the book data ranked higher than the first preset threshold in the first scoring and sorting result; scores and sorts the book data in the second book data set based on the second network to obtain a second scoring and sorting result; generates a target book data set based on the book data ranked higher than the second preset threshold in the second scoring and sorting result; for the book data in the target book data set, sums the score of the first network corresponding to the book data and the score of the second network corresponding to the book data to obtain the total score of the book data; sorts the book data in the target book data set based on the total score of the book data in the target book data set; pushes the book data ranked higher than the third preset threshold to the user. According to the disclosed embodiment, the book data is first quickly screened by the first network, and the second network performs a second screening based on the book data screened by the first network. Due to the reduction in the amount of data, the screening speed is greatly improved, and the results of the two screenings are combined to recommend books to users, which reduces system overhead, improves the recommendation speed while ensuring the accuracy of the recommendation, realizes the real-time nature of the book recommendation, and improves the reading experience of the user.

[0060] Figure 3 It is a schematic diagram of the principle of a deep decomposition machine DeepFM model provided in an embodiment of the present disclosure. The model 300 mainly includes a sparse feature processing network 310, an FM network 320, a Deep network 330, and an output network 340.

[0061] The sparse feature processing network 310 is mainly used to implement sparse feature processing of the book data of each book in the book data set to obtain a feature vector of the book data.

[0062] The FM network 320 includes an embedding processing subnetwork and an FM processing subnetwork. The FM network is mainly used to implement the first network, embedding the feature vectors of the book data, cross-combining the feature vectors in pairs, calculating the dot product (also called dot product, dot multiplication) value between the cross-combined feature vectors, performing linear processing on the dot product value to obtain the first score of the book data of each book, sorting the book data in the first book data set in descending order according to the first score, outputting the first score sorting result, and generating the second book data set based on the book data ranked higher than the first preset threshold in the first score sorting result. The FM network has a simple structure and can quickly complete the screening.

[0063] The Deep network 330 is mainly used for the implementation of the second network. The feature vectors of the book data of each book in the second book data set are fully connected through a plurality of hidden layers to obtain a second score of the book. The book data in the second book data set are sorted in descending order according to the second score to obtain a second score sorting result. The target book data set is generated based on the book data ranked higher than a second preset threshold in the second score sorting result.

[0064] The output network 340 can be understood as the top-level network of the DeepFM model, which is mainly based on the scores of the first network and the second network corresponding to the book data in the target book data set, combined with the activation function Sigmoid for summing up, to obtain the total score of the book data, and based on the total score of the book data in the target book data set, sort the book data in the target book data set, and push the book data with a total score ranking higher than the third preset threshold to the user.

[0065] The following is an explanation of the model training method of the DeepFM model. The method can be executed by a computing device, which can be understood as any device with computing functions and processing capabilities.

[0066] Firstly, the book data of multiple books are obtained from the set of book data recalled by the multi-way recall strategy, the book data of each book is labeled, and the recommendation value label of the book data of each book is obtained. The book data and the recommendation value label corresponding to the book data of each book in the book data are used as training samples, and the multiple training samples are divided into a training set, a validation set and a test set according to a preset ratio. The model is trained based on the training set, and the accuracy of the model is verified on the validation set.

[0067] In the disclosed embodiment, a training set can be used to simultaneously train the FM network and the Deep network in the DeepFM model, the first book data set is input into the FM network to obtain a first score of the book data and a second book data set, the second book data set is input into the Deep network to obtain a second score of the book data and a target book data set, the total score of the book data is obtained based on the first score and the second score, after multiple iterative training, the model parameters are continuously optimized through the loss function, and the trained DeepFM model is output.

[0068] In the disclosed embodiment, each time after the first score of the book data is obtained, the first score of the book data is verified, and the FM network is optimized and trained based on the verification result and the book data. Specifically, the first score of the book data is input into the first loss function, the first loss value of the model on the verification set is calculated, and the parameters of the FM network are continuously optimized based on the first loss value. When the first loss value of the model on the verification set is less than or equal to the first preset loss threshold, the training is stopped and the final parameters of the FM network are determined.

[0069] In another embodiment of the present disclosure, the total score of the book data can be verified at predetermined time intervals, and the entire DeepFM model can be optimized and trained based on the verification results and the book data. Specifically, the total score of the book data is input into the second loss function, the second loss value of the model on the verification set is calculated, and the parameters of the entire DeepFM model are continuously optimized based on the second loss value. When the second loss value of the model on the verification set is less than or equal to the second preset loss threshold, the training is stopped and the final parameters of the DeepFM model are determined.

[0070] Compared with the prior art, the technical solution provided by the embodiments of the present disclosure has the following advantages:

[0071] Based on the existing parallel DeepFM model, the disclosed embodiment improves the DeepFM model into a serial solution, trains the serial DeepFM model, uses the FM network to perform a first rapid screening of book data, and then transmits the first screening result to the Deep network for a second screening, that is, the calculation result of the FM network is used as the input of the Deep network. Due to the reduction in the amount of data, the screening speed is greatly improved, the system overhead is reduced, the recommendation speed is improved while ensuring the accuracy of the recommendation, the real-time nature of the book recommendation is achieved, and the user's reading experience is improved.

[0072] Figure 4 It is a structural schematic diagram of an electronic device provided in an embodiment of the present disclosure. The electronic device can be understood as any device with computing functions and processing capabilities.

[0073] The electronic devices provided by the embodiments of the present disclosure may include, but are not limited to, mobile terminals such as smart phones, tablet computers, and laptop computers, as well as fixed terminals such as digital TVs and desktop computers.

[0074] It should be noted that Figure 4 The electronic device 400 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0075] The electronic device 400 conventionally includes a processor 410 and a memory 420. The memory 420 may be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read-only memory), an EPROM, a hard disk, or a ROM. The memory 420 has a storage space 421 for executable instructions (or program codes) 4211 for executing any method steps in the above-mentioned book recommendation method. For example, the storage space 421 for executable instructions may include individual executable instructions 4211 respectively used to implement various steps in the above-mentioned book recommendation method. These executable instructions may be read from or written to one or more computer program products. These computer program products include program code carriers such as a hard disk, a compact disk (CD), a memory card, or a floppy disk. Such a computer program product is typically a portable or fixed storage unit. The storage unit may have a storage medium similar to Figure 4 The memory 420 in the electronic device 400 is similarly arranged as a storage segment or storage space. The executable instructions can be compressed, for example, in an appropriate form. Generally, the storage unit includes executable instructions for executing the steps of the book recommendation method according to the present invention, that is, codes that can be read by a processor such as the processor 410, and when these codes are executed by the electronic device 400, the electronic device 400 executes the various steps in the book recommendation method described above.

[0076] Of course, to simplify, Figure 4 Only some of the components related to the present invention in the electronic device 400 are shown, and components such as a bus, an input / output interface, an input device, an output device, etc. are omitted. In addition, according to specific application conditions, the electronic device 400 may also include any other appropriate components.

[0077] The memory 420 in the electronic device 400 is used to store executable instructions 4211, and the executable instructions 4211 enable the processor 420 to perform the following operations:

[0078] Performing a first screening process on the first book data set based on the first network in the preset recommendation model to obtain a second book data set;

[0079] Performing a second screening process on the second book dataset based on the second network in the preset recommendation model to obtain a target book dataset;

[0080] Based on the target book dataset, push books to users.

[0081] Optionally, the screening accuracy of the first network in the preset recommendation model in the electronic device 400 is lower than that of the second network.

[0082] Optionally, when the electronic device 400 performs a first screening process on the first book data set based on the first network in the preset recommendation model to obtain the second book data set, the executable instruction 4211 further causes the processor 420 to execute:

[0083] Scoring and sorting the book data in the first book data set based on the first network to obtain a first scoring and sorting result;

[0084] A second book data set is generated based on the book data ranked higher than a first preset threshold in the first scoring and sorting results.

[0085] Optionally, when the electronic device 400 performs a second screening process on the second book data set based on the second network in the preset recommendation model to obtain the target book data set, the executable instruction 4211 further causes the processor 420 to execute:

[0086] Scoring and sorting the book data in the second book data set based on the second network to obtain a second scoring and sorting result;

[0087] A target book data set is generated based on the book data ranked higher than a second preset threshold in the second scoring and ranking results.

[0088] Optionally, when the electronic device 400 pushes books to the user based on the target book dataset, the executable instruction 4211 further enables the processor 420 to execute:

[0089] For the book data in the target book data set, the score of the first network corresponding to the book data and the score of the second network corresponding to the book data are summed to obtain the total score of the book data;

[0090] sorting the book data in the target book data set based on the total scores of the book data in the target book data set;

[0091] The book data with a total score ranking higher than the third preset threshold is pushed to the user.

[0092] Optionally, the preset recommendation model in the electronic device 400 is a deep factorization machine DeepFM model, the first network is an FM network, and the second network is a Deep network.

[0093] The electronic device provided by the embodiment of the present disclosure is capable of executing the above-mentioned book recommendation method, and its execution method and beneficial effects are similar, which will not be repeated here.

[0094] An embodiment of the present disclosure provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the processor implements the above-mentioned book recommendation method. The execution method and beneficial effects are similar and will not be repeated here.

[0095] The above-mentioned computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable 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 above.

[0096] The computer program may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the disclosed embodiments, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0097] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0098] The above description is only a specific embodiment of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A book recommendation method, characterized in that: The method comprises: Performing a first screening process on the first book data set based on the first network in the preset recommendation model to obtain a second book data set; Performing a second screening process on the second book dataset based on the second network in the preset recommendation model to obtain a target book dataset; Pushing books to users based on the target book dataset; The first network in the preset recommendation model performs a first screening process on the first book data set to obtain a second book data set, including: Scoring and sorting the book data in the first book data set based on the first network to obtain a first scoring and sorting result; Generate a second book data set based on the book data ranked higher than a first preset threshold in the first scoring and ranking results; The performing a second screening process on the second book dataset based on the second network in the preset recommendation model to obtain a target book dataset includes: Scoring and sorting the book data in the second book data set based on the second network to obtain a second scoring and sorting result; Generate a target book data set based on the book data ranked higher than a second preset threshold in the second scoring and sorting results; The pushing books to the user based on the target book dataset includes: For the book data in the target book data set, summing the scores of the first network corresponding to the book data and the scores of the second network corresponding to the book data to obtain a total score of the book data; sorting the book data in the target book data set based on the total scores of the book data in the target book data set; The book data with a total score ranking higher than the third preset threshold is pushed to the user.

2. The method according to claim 1, characterized in that The screening accuracy of the first network is lower than that of the second network.

3. The method according to claim 1, characterized in that The preset recommendation model is a deep factorization machine DeepFM model, the first network is an FM network, and the second network is a Deep network.

4. An electronic device, characterized in that: The invention comprises a processor and a memory, wherein the memory is used to store executable instructions, and the executable instructions enable the processor to perform the following operations: Performing a first screening process on the first book data set based on the first network in the preset recommendation model to obtain a second book data set; Performing a second screening process on the second book dataset based on the second network in the preset recommendation model to obtain a target book dataset; Pushing books to users based on the target book dataset; When performing a first screening process on the first book data set based on the first network in the preset recommendation model to obtain a second book data set, the executable instruction further causes the processor to execute: Scoring and sorting the book data in the first book data set based on the first network to obtain a first scoring and sorting result; Generate a second book data set based on the book data ranked higher than a first preset threshold in the first scoring and ranking results; When performing a second screening process on the second book dataset based on the second network in the preset recommendation model to obtain a target book dataset, the executable instructions further cause the processor to execute: Scoring and sorting the book data in the second book data set based on the second network to obtain a second scoring and sorting result; Generate a target book data set based on the book data ranked higher than a second preset threshold in the second scoring and sorting results; When pushing books to the user based on the target book dataset, the executable instructions further cause the processor to execute: For the book data in the target book data set, summing the scores of the first network corresponding to the book data and the scores of the second network corresponding to the book data to obtain a total score of the book data; sorting the book data in the target book data set based on the total scores of the book data in the target book data set; The book data with a total score ranking higher than the third preset threshold is pushed to the user.

5. The electronic device according to claim 4, characterized in that: The screening accuracy of the first network is lower than that of the second network.

6. The electronic device according to claim 4, characterized in that: The preset recommendation model is a deep factorization machine DeepFM model, the first network is an FM network, and the second network is a Deep network.

7. 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 processor implements the book recommendation method according to any one of claims 1 to 3.

Images