Hotspot event pushing method and device, electronic equipment and storage medium

Abstract

The present disclosure provides a hotspot event pushing method and device, electronic equipment and storage medium, relates to the technical field of computers, and particularly relates to the technical field of media content pushing. The specific implementation scheme is as follows: determining a target hotspot event based on a hotspot event preference vector and a candidate hotspot event, wherein the hotspot event preference vector is obtained by performing inter-event feature cross learning on hotspot events in a historical hotspot event watch list of a user; and pushing the target hotspot event to the user. The present disclosure pushes the hotspot event individually, and improves the pushing accuracy.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, specifically to the field of media content push technology, and in particular to a method, apparatus, electronic device, and storage medium for pushing trending events. Background Technology

[0002] With the development of mobile internet, online media platforms have become an important channel for people to obtain and exchange information.

[0003] The act of users sharing their opinions and breaking news on online media platforms generates a large number of trending topics every day. Driven by factors such as increasing user engagement and expanding platform influence, online media platforms automatically push trending topics to users. However, because different users have different information preferences, personalized push notifications play a significant role in improving push efficiency. Summary of the Invention

[0004] This disclosure provides a method, apparatus, electronic device, and storage medium for pushing trending events, enabling personalized push notifications of trending events and improving the accuracy of the push notifications.

[0005] According to a first aspect of this disclosure, a method for pushing trending events is provided, comprising: determining a target trending event based on a trending event preference vector and candidate trending events, wherein the trending event preference vector is obtained by performing cross-feature learning between trending events in a user's historical trending event viewing list; and pushing the target trending event to the user.

[0006] According to a second aspect of this disclosure, a hot topic event push device is provided, comprising: a determining module configured to determine a target hot topic event based on a hot topic event preference vector and candidate hot topic events, wherein the hot topic event preference vector is obtained by performing cross-feature learning between hot topic events in a user's historical hot topic event viewing list; and a push module configured to push the target hot topic event to the user.

[0007] According to a third aspect of this 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, the instructions being executed by the at least one processor to enable the at least one processor to perform the hotspot event push method provided in the first aspect.

[0008] According to a fourth aspect of this disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause the computer to execute the hotspot event push method provided in the first aspect.

[0009] According to a fifth aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the hot topic event push method provided in the first aspect.

[0010] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0011] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0012] Figure 1 This is an exemplary system architecture diagram to which this disclosure can be applied;

[0013] Figure 2 This is a flowchart of an embodiment of the hot topic push method according to this disclosure;

[0014] Figure 3 This is a flowchart of yet another embodiment of the hot topic push method according to this disclosure;

[0015] Figure 4 This is a schematic diagram of the training process of the first feature cross-learning model according to this disclosure;

[0016] Figure 5 This is a flowchart of yet another embodiment of the hot topic push method according to this disclosure;

[0017] Figure 6 This is a schematic diagram of the structure of one embodiment of the hot topic push device according to the present disclosure;

[0018] Figure 7 This is a block diagram of an electronic device used to implement the hot topic event push method of the embodiments of this disclosure. Detailed Implementation

[0019] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0020] Figure 1 An exemplary system architecture 100 is shown, in which embodiments of the hotspot event push method and hotspot event push device of this disclosure can be applied.

[0021] like Figure 1 As shown, system architecture 100 may include mobile devices 101, 102, and 103, a network 104, and a server 105. Network 104 serves as the medium for providing communication links between mobile devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0022] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104, and can also interact with each other through server 105. Various client applications can be installed on terminal devices 101, 102, and 103, such as online news media applications, social media applications, and video media applications.

[0023] Terminal devices 101, 102, and 103 can be either hardware or software. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices, including but not limited to smartphones, tablets, laptops, and desktop computers. When terminal devices 101, 102, and 103 are software, they can be installed in the aforementioned electronic devices. They can be implemented as multiple software programs or software modules, or as a single software program or software module. No specific limitations are made here.

[0024] Server 105 can provide various interactive services. For example, server 105 analyzes and processes the user's historical hot event viewing list obtained from terminal devices 101, 102, and 103 to obtain a hot event preference vector; or it obtains the user's corresponding hot event preference vector from other servers or terminal devices; and then, based on the hot event preference vector, it selects the target hot event from multiple candidate hot events and pushes it to the user.

[0025] It should be noted that server 105 can be either hardware or software. When server 105 is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When server 105 is software, it can be implemented as multiple software programs or software modules (e.g., used to provide distributed services), or as a single software program or software module. No specific limitations are made here.

[0026] It should be noted that the hot event push method provided in this embodiment is generally executed by server 105, and correspondingly, the hot event push device is generally set in server 105.

[0027] It should be understood that Figure 1 The number of mobile devices, networks, and servers shown is merely illustrative. Any number of mobile devices, networks, and servers can be included depending on implementation needs.

[0028] Continue to refer to Figure 2 The diagram illustrates a flow 200 of an embodiment of a hot topic event push method according to this disclosure. The push method includes the following steps:

[0029] Step 201: Determine the target hot event based on the hot event preference vector and candidate hot events.

[0030] In this embodiment, the execution body of the hot topic event push method (e.g.) Figure 1 The server 105 shown can first obtain the hot event preference vector corresponding to the user, and then select the target hot event to be pushed from the current candidate hot events according to the hot event preference vector.

[0031] The aforementioned hot topics refer to events or topics that, within a specific period, attract widespread public attention and discussion due to social, economic, and cultural factors. These events are often time-sensitive, sudden, and have a broad social impact, quickly becoming the focus of public opinion.

[0032] The aforementioned hot topic event preference vector is used to characterize a user's viewing tendency for hot topics events. It is obtained based on the user's historical hot topic event viewing list. The user's historical hot topic event viewing list can be a list of hot topics events viewed by the user within a preset historical period, a list of hot topics events whose viewing time is closest to the current time, or a list of hot topics events whose viewing time is determined to be within a preset historical period and whose viewing time is closest to the current time. The preset historical period can be the past few days, the past month, the past six months, etc.

[0033] As an example, the aforementioned trending event preference vector can be obtained by performing cross-feature learning on trending events in a user's historical trending event viewing list. For instance, it can identify the themes of each trending event in the historical trending event viewing list and then perform cross-feature learning on the theme features of each trending event to obtain the trending event preference vector. Another example is selecting key trending events from the historical trending event viewing list, where key trending events refer to those where the user has engaged in interactive actions such as forwarding, commenting, or liking; and then performing cross-feature learning on each key trending event to obtain the trending event preference vector.

[0034] Step 202: Push target trending events to users.

[0035] In this embodiment, the aforementioned execution entity determines the target hot event and pushes the target hot event to the user's client application.

[0036] In this embodiment, the hot topic event push method provided by this disclosure determines the target hot topic event based on a hot topic event preference vector and candidate hot topic events. The hot topic event preference vector is obtained by performing cross-feature learning between hot topic events in the user's historical hot topic event viewing list. The target hot topic event is then pushed to the user. Personalized hot topic event pushes are performed based on the user's hot topic event preferences, improving push accuracy and enhancing the user's viewing experience of hot topic events.

[0037] Continue to refer to Figure 3 This illustrates a flow 300 of yet another embodiment of the hot topic push method according to this disclosure. The push method includes the following steps:

[0038] Step 301: Based on the historical hot events viewing list, determine the target events to follow.

[0039] In this embodiment, steps 301 to 304 are different from Figure 2 The execution entity of the hot event push method shown (e.g.) Figure 1 The server (105) shown is executed by the server or mobile terminal. In special cases, steps 301 to 304 can also be performed by... Figure 2 The execution entity of the hot event push method shown (e.g.) Figure 1 The process is executed by server 105. In this step, after obtaining the user's historical hot events viewing list from terminal devices 101, 102, and 103, the executing entity selects the target event from the historical hot events viewing list.

[0040] It should be noted that while users are watching trending events, they may repeatedly watch the same trending event pushed by different pushers. Therefore, the number of items in a user's historical trending event viewing list is not the same as the number of trending events included in the list.

[0041] As an example, the above method of determining target events based on a historical hot event viewing list includes: determining the viewing duration of each hot event in the historical hot event viewing list; and selecting hot events with viewing durations exceeding a preset viewing duration threshold or the second preset number of hot events with the highest viewing duration as target events.

[0042] As another example, the above-mentioned determination of target events based on the historical hot event viewing list includes: determining the average viewing time of each hot event in the historical hot event viewing list; and identifying hot events with an average viewing time greater than the preset viewing time as target events.

[0043] The preset viewing time mentioned above is the average viewing time of all trending events in the historical trending events viewing list.

[0044] As another example, the above method of determining target events based on a historical hot event viewing list includes: determining the number of times each hot event in the historical hot event viewing list has been viewed, and identifying hot events with a viewing count greater than a threshold as target events.

[0045] Step 302: Determine the sequence of hot events corresponding to the target events of interest.

[0046] After identifying the target events of interest, the entity executing this step generates a sequence of hot events corresponding to those events.

[0047] As an example, the above process determines whether the number of target events of interest is less than a third preset number, which represents a predetermined sequence length. If the number of target events of interest is less than the third preset number, the insufficient portion is filled with blank events, and then a hotspot event sequence is generated. If the number of target events of interest is not less than the third preset number, a third preset number of hotspot events are selected from the target events of interest to generate a hotspot event sequence.

[0048] Step 303: Determine the hot event feature coding sequence corresponding to the hot event sequence.

[0049] After determining the hot event sequence, the entity executing this step determines the hot event feature encoding sequence corresponding to the hot event sequence.

[0050] As an example, the above-mentioned determination of the hot event feature coding sequence corresponding to the hot event sequence includes: determining the event name coding sequence corresponding to the hot event sequence; and determining the hot event feature coding sequence based on the event name coding sequence.

[0051] As one possible approach, the above method of determining the hot event feature coding sequence based on the event name coding sequence includes: determining the event name coding sequence as the hot event feature coding sequence.

[0052] Among them, the event name is a textual representation that can reflect the core content of a hot event. Therefore, the event name of a hot event can be used as a feature of the hot event to achieve accurate representation of the hot event.

[0053] As another feasible approach, the above-mentioned method of determining the hot event feature coding sequence based on the event name coding sequence includes: determining the popularity value coding sequence corresponding to the hot event sequence; and adding the event name coding sequence and the popularity value coding sequence bit by bit to obtain the hot event feature coding sequence.

[0054] Among them, the popularity value is an important indicator of the spread of trending events. Introducing the popularity value into the characteristics of trending events can more accurately represent them. The above-mentioned implementation method adopts a similar approach to constructing positional encoding for the input sequence of a transformer model, incorporating the popularity value encoding features of each trending event into the corresponding event name encoding features.

[0055] In this embodiment, determining the event name encoding sequence or popularity value encoding sequence corresponding to the hot event sequence can be achieved by using a pre-trained Large Language Model (LLM) to embed and encode the event names or popularity values ​​of each hot event in the hot event sequence, thus obtaining the event name encoding sequence or popularity value encoding sequence corresponding to the hot event sequence. The aforementioned Large Language Model (LLM) is a class of basic models that, after training on a large amount of data, can understand and generate natural language and other types of content to perform various tasks. It includes models such as the Bidirectional Encoder Representations from Transformers (BERT) model and the Chinese-English Semantic Vector Model (BAAI General Embedding, BGE model).

[0056] Step 304: Perform inter-event feature cross-learning on the hot event feature encoding sequence to obtain the hot event preference vector.

[0057] After determining the feature encoding sequence of hot events, the execution entity in this step performs feature cross-learning between events on the feature encoding sequence of hot events to obtain the hot event preference vector.

[0058] As an example, the above-mentioned method of performing inter-event feature cross-learning on the feature encoding sequence of hot events to obtain a hot event preference vector includes: inputting the feature encoding sequence of hot events into a first feature cross-learning model to obtain the output hot event preference vector; wherein, the first feature cross-learning model is a converter model used for inter-event feature cross-learning.

[0059] The aforementioned converter model refers to the Transformer model. In this step, the main body uses the hot event feature encoding sequence as the input representation of the Transformer model. In the self-attention layer of each encoder, the self-attention mechanism is used to perform cross-learning of features between events. The output of the learned Transformer model is used as the hot event preference vector.

[0060] As another example, the above-mentioned hot event feature cross-learning of the hot event feature encoding sequence to obtain the hot event preference vector includes: determining the user attribute feature encoding vector corresponding to the user; inputting the hot event feature encoding sequence and the user attribute feature encoding vector into the second feature cross-learning model to obtain the output hot event preference vector; wherein, the second feature cross-learning model is a converter model used for inter-event feature cross-learning and user attribute feature and event feature cross-learning.

[0061] Among them, the user attribute feature encoding vector corresponding to the user is obtained by encoding the user attribute information. The user attribute information refers to the basic information of the user obtained with the user's authorization, including but not limited to age, gender, zodiac sign, etc.

[0062] The aforementioned converter model refers to the Transformer model. In this step, the main body uses the hot event feature encoding sequence as the input representation of the Transformer model and inputs the user attribute feature encoding vector into the feedforward neural network of the Transformer model. In the self-attention layer of each encoder, the self-attention mechanism is used to perform cross-learning of event features, and the user attribute features and event features are cross-learned in the feedforward neural network layer. The output of the learned Transformer model is used as the hot event preference vector.

[0063] The aforementioned second feature cross-learning model introduces user-side features to complete the feature cross-learning between user features and time features, thereby improving the performance of hot topic event preference learning.

[0064] The training process of the first feature cross-learning model and the second feature cross-learning model is roughly the same. The first feature cross-learning model will be used as an example for explanation.

[0065] See Figure 4 A schematic diagram of the training process of the publicly disclosed first feature cross-learning model, which includes:

[0066] Step 1: Obtain sample data. Each sample includes a sequence of user hot events. In particular, the sample data for the second feature cross-learning model includes the sequence of user hot events and the user attribute feature encoding vector.

[0067] Step 2: Perform data augmentation on the sample data.

[0068] Step 3: Use a pre-trained Large Language Model (LLM model) to perform event name embedding encoding (or event name embedding encoding and popularity value embedding encoding) on ​​the hot event sequences in the enhanced sample data to obtain the hot event feature encoding sequence.

[0069] Step 4: Perform comparative learning on the hotspot event feature encoding sequence. Specifically, a comparative learning structure is introduced into the encoder part of the Transformer model. This structure consists of a first branch and a second branch that is completely identical to the first branch. The first branch is formed by connecting the Transformer model's encoder to a fully connected layer, and the second branch updates parameters using momentum update. The hotspot event feature encoding sequence is input into the first branch and the second branch respectively to obtain the corresponding first output vector and second output vector.

[0070] Step 5: Input the first output vector into a fully connected layer to update the spatial dimension of the first output vector to be consistent with that of the second output vector.

[0071] Step 6: Use the spatial L2 distance between the updated first and second output vectors as the contrastive loss to optimize the model parameters.

[0072] The training process described above employed a comparative learning approach, which improved the accuracy of model training.

[0073] Step 305: Determine the target hot event based on the hot event preference vector and candidate hot events.

[0074] Step 306: Push target trending events to users.

[0075] Steps 305 to 306 are by Figure 2 The execution entity of the hot event push method shown (e.g.) Figure 1 The server 105 shown is executing the steps 305 and 306, which are detailed in the documentation. Figure 2 Steps 201 and 202 in the illustrated embodiments are described in detail and will not be repeated here.

[0076] In this embodiment, the hot topic event push method provided by this disclosure selects hot topics with high user attention from the historical hot topic event viewing list to generate a hot topic event sequence; based on the event name and popularity value of the hot topic events, a hot topic event feature encoding sequence corresponding to the hot topic event sequence is determined; the hot topic event feature encoding sequence is subjected to inter-event feature cross-learning to obtain a hot topic event preference vector. The method uses machine learning to accurately learn the user's hot topic event preferences, and then pushes personalized hot topic events to the user based on the user's hot topic event preferences, thereby improving the accuracy of the push and enhancing the user's viewing experience of hot topic events.

[0077] Continue to refer to Figure 5 This illustrates a flow 500 of yet another embodiment of the hot topic push method according to this disclosure. The push method includes the following steps:

[0078] Step 501: Based on the historical hot events viewing list, determine the target events to watch. The target events to watch are hot events with an average viewing time longer than the preset viewing time.

[0079] In this embodiment, steps 501 to 504 are different from Figure 2 The execution entity of the hot event push method shown (e.g.) Figure 1 The server (105) shown is executed by the server or mobile terminal. In special cases, steps 501 to 504 can also be performed by... Figure 2 The execution entity of the hot event push method shown (e.g.) Figure 1 The server 105 shown is executed.

[0080] The specific operations of steps 501 to 504 have been described in detail. Figure 3 Steps 301 to 304 in the illustrated embodiments are described in detail and will not be repeated here.

[0081] Step 502: Determine the sequence of hot events corresponding to the target events of interest.

[0082] Step 503: Determine the hot event feature coding sequence corresponding to the hot event sequence.

[0083] Step 504: Perform inter-event feature cross-learning on the hot event feature encoding sequence to obtain the hot event preference vector.

[0084] Step 505: Based on the hot topic event preference vector and candidate hot topic events, determine the predicted viewing duration of the candidate hot topic events.

[0085] Steps 505 and 507 are by Figure 2 The execution entity of the hot event push method shown (e.g.) Figure 1 The execution is performed on server 105 shown. After determining the hot event preference vector, the execution entity in this step can use the hot event preference vector to select the target hot event to be pushed from many candidate hot events.

[0086] As an example, the above method for determining the predicted viewing duration of candidate hot events based on hot event preference vectors and candidate hot events includes: determining the feature encoding vector of candidate hot events; concatenating the feature encoding vector and the hot event preference vector to obtain a concatenated vector; and inputting the concatenated vector into a preset viewing duration prediction model to obtain the output predicted viewing duration of candidate hot events.

[0087] The method for determining the feature encoding vector of the candidate hot event is consistent with the method for determining the feature encoding sequence of the hot event corresponding to the above hot event sequence.

[0088] The training process of the above-mentioned viewing duration prediction model includes:

[0089] Step A: Obtain training samples, which include hot events and the actual predicted duration of hot events;

[0090] Step B: Input the features (event name or event name and popularity value) of the hot events in the training samples into the Large Language Model (LLM model), and determine the feature encoding vector of the hot events based on the output results;

[0091] Step C: Concatenate the feature encoding vector of hot events with the hot event preference vector in a fully connected layer to obtain the concatenated vector; the role of the fully connected layer is to update the spatial dimension of the feature encoding vector of hot events to be consistent with the hot event preference vector.

[0092] Step D: Input the concatenated vector into the initial prediction network to determine the predicted viewing duration of hot events;

[0093] Step E: Update the parameters of the initial prediction network using the cross-entropy loss between the predicted and actual viewing times of trending events as the loss function;

[0094] Step F: Repeat the above steps until the preset iteration termination condition is reached, and use the final prediction network as the above viewing duration prediction model.

[0095] Step 506: Based on the predicted viewing time, identify target trending events.

[0096] After determining the predicted viewing time of candidate trending events, the entity executing this step identifies the target trending event based on the predicted viewing time.

[0097] As an example, the above method of determining target hot events based on predicted viewing time includes: sorting the predicted viewing time in descending order, and selecting the candidate hot events corresponding to the fourth preset number of predicted viewing times at the top of the sequence as target hot events.

[0098] As another example, the above method of determining target hot events based on predicted viewing time includes: identifying candidate hot events whose predicted viewing time is greater than a preset viewing time threshold as target hot events.

[0099] Step 507: Push target trending events to users.

[0100] In this embodiment, the hot topic event push method provided by this disclosure feeds the user's hot topic event preference vector and the feature vector of the candidate hot topic event into a pre-trained viewing duration prediction model to predict the viewing duration of the candidate hot topic event, and then selects and pushes the target hot topic event based on the predicted viewing duration, thereby improving the push accuracy.

[0101] Continue to refer to Figure 6 , Figure 6 A schematic diagram 600 illustrates one embodiment of the hot topic push device of this disclosure; this device embodiment is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0102] like Figure 6 As shown, the hot topic event push device in this embodiment may include a determining module 601 and a push module 602. The determining module 601 is configured to determine a target hot topic event based on a hot topic event preference vector and candidate hot topic events. The hot topic event preference vector is obtained by performing cross-feature learning between hot topic events in the user's historical hot topic event viewing list. The push module 602 is configured to push the target hot topic event to the user.

[0103] In this embodiment, the specific processing of the determining module 601 and the pushing module 602 and the resulting technical effects can be referred to respectively. Figure 2 The relevant descriptions of steps 201-202 in the corresponding embodiments will not be repeated here.

[0104] In some optional implementations of this embodiment, the device further includes a hot topic event preference vector generation module 603; the hot topic event preference vector generation module 603 includes: a first determining submodule configured to determine a target attention event based on a historical hot topic event viewing list, wherein the target attention event is a hot topic event with an average viewing time longer than a preset viewing time; a second determining submodule configured to determine a hot topic event sequence corresponding to the target attention event; a third determining submodule configured to determine a hot topic event feature encoding sequence corresponding to the hot topic event sequence; and a cross-learning submodule configured to perform inter-event feature cross-learning on the hot topic event feature encoding sequence to obtain a hot topic event preference vector.

[0105] In some optional implementations of this embodiment, the third determining submodule includes: a first determining unit configured to determine an event name encoding sequence corresponding to a hotspot event sequence; and a second determining unit configured to determine a hotspot event feature encoding sequence based on the event name encoding sequence.

[0106] In some optional implementations of this embodiment, the second determining unit includes: a first determining subunit configured to determine a heat value encoding sequence corresponding to a hot event sequence; and an encoding fusion subunit configured to add the event name encoding sequence and the heat value encoding sequence bit by bit to obtain a hot event feature encoding sequence.

[0107] In some optional implementations of this embodiment, the cross-learning submodule includes: a first cross-learning unit configured to input a hotspot event feature encoding sequence into a first feature cross-learning model to obtain an output hotspot event preference vector; wherein, the first feature cross-learning model is a converter model for cross-learning of features between events.

[0108] In some optional implementations of this embodiment, the cross-learning submodule further includes: a third determining unit configured to determine a user attribute feature encoding vector corresponding to the user; and a second cross-learning unit configured to input the hot event feature encoding sequence and the user attribute feature encoding vector into a second feature cross-learning model to obtain an output hot event preference vector; wherein the second feature cross-learning model is a converter model used for cross-learning of features between events and cross-learning of user attribute features and event features.

[0109] In some optional implementations of this embodiment, the preset viewing duration is the average viewing duration of all hot events in the historical hot event viewing list.

[0110] In some optional implementations of this embodiment, the determining module includes: a fourth determining submodule configured to determine the predicted viewing duration of candidate hot events based on hot event preference vectors and candidate hot events; and a fifth determining submodule configured to determine the target hot event based on the predicted viewing duration.

[0111] In some optional implementations of this embodiment, the fourth determining submodule includes: a sixth determining unit configured to determine the feature encoding vector of the candidate hot event; a splicing unit configured to splice the feature encoding vector and the hot event preference vector to obtain a spliced ​​vector; and a prediction unit configured to input the spliced ​​vector into a preset viewing duration prediction model to obtain the predicted viewing duration of the candidate hot event.

[0112] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0113] Figure 7A schematic block diagram of an example electronic device 700 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0114] like Figure 7 As shown, device 700 includes a computing unit 701, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 702 or a computer program loaded from storage unit 708 into random access memory (RAM) 703. RAM 703 may also store various programs and data required for the operation of device 700. The computing unit 701, ROM 702, and RAM 703 are interconnected via bus 704. Input / output (I / O) interface 705 is also connected to bus 704.

[0115] Multiple components in device 700 are connected to I / O interface 705, including: input unit 706, such as keyboard, mouse, etc.; output unit 707, such as various types of monitors, speakers, etc.; storage unit 708, such as disk, optical disk, etc.; and communication unit 709, such as network card, modem, wireless transceiver, etc. Communication unit 709 allows device 700 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0116] The computing unit 701 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 701 performs the various methods and processes described above, such as the hot event push method. For example, in some embodiments, the hot event push method can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 708. In some embodiments, part or all of the computer program can be loaded and / or installed on device 700 via ROM 702 and / or communication unit 709. When the computer program is loaded into RAM 703 and executed by the computing unit 701, one or more steps of the hot event push method described above can be performed. Alternatively, in other embodiments, the computing unit 701 can be configured to perform the hot event push method by any other suitable means (e.g., by means of firmware).

[0117] Various embodiments 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-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0118] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0119] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0120] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, 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 sound input, voice input, or tactile input).

[0121] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0122] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0123] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0124] The acquisition, storage, and application of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0125] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for pushing trending events, comprising: Based on the historical hot events viewing list, target events for attention are determined. The target events for attention are hot events with an average viewing time longer than a preset viewing time, and the hot events are time-sensitive. Identify the sequence of hot events corresponding to the target events of interest; Determine the hotspot event feature encoding sequence corresponding to the hotspot event sequence; The hot topic event preference vector is obtained by performing inter-event feature cross-learning on the hot topic event feature encoding sequence, including: determining the user attribute feature encoding vector corresponding to the user; inputting the hot topic event feature encoding sequence and the user attribute feature encoding vector into a second feature cross-learning model to obtain the output hot topic event preference vector; wherein, the second feature cross-learning model is a converter model for inter-event feature cross-learning and user attribute feature and event feature cross-learning, and the second feature cross-learning model introduces user-side features to complete the feature cross-learning between user features and time features; Based on the hot topic event preference vector and candidate hot topics events, the target hot topics event is determined; The target trending event is pushed to the user.

2. The method according to claim 1, wherein, The step of determining the hotspot event feature encoding sequence corresponding to the hotspot event sequence includes: Determine the event name encoding sequence corresponding to the hotspot event sequence; The hotspot event feature encoding sequence is determined based on the event name encoding sequence.

3. The method according to claim 2, wherein, The step of determining the hotspot event feature encoding sequence based on the event name encoding sequence includes: Determine the heat value encoding sequence corresponding to the hot event sequence; The event name encoding sequence is added bit by bit to the popularity value encoding sequence to obtain the hot event feature encoding sequence.

4. The method according to claim 1, wherein, The step of performing inter-event feature cross-learning on the feature encoding sequence of the hot events to obtain the hot event preference vector includes: The hot topic event feature encoding sequence is input into the first feature cross-learning model to obtain the output hot topic event preference vector; wherein, the first feature cross-learning model is a converter model for inter-event feature cross-learning.

5. The method according to claim 1, wherein, The preset viewing duration is the average viewing duration of all hot events in the historical hot event viewing list.

6. The method according to any one of claims 1 to 5, wherein, The process of determining the target hot event based on the hot event preference vector and candidate hot events includes: Based on the hot topic event preference vector and the candidate hot topic events, the predicted viewing duration of the candidate hot topic events is determined; Based on the predicted viewing duration, the target trending event is determined.

7. The method according to claim 6, wherein, The step of determining the predicted viewing duration of the candidate hot events based on the hot event preference vector and the candidate hot events includes: Determine the feature encoding vector of the candidate hotspot event; The feature encoding vector and the hotspot event preference vector are concatenated to obtain the concatenated vector. The concatenated vector is input into a preset viewing duration prediction model to obtain the predicted viewing duration of the candidate hot events.

8. A hot topic event push device, comprising: The first determining submodule is configured to determine target events of interest based on a historical list of trending events viewed. The target events of interest are trending events with an average viewing time longer than a preset viewing time, and the trending events are time-sensitive. The second determining submodule is configured to determine a sequence of hot events corresponding to the target event of interest. The third determining submodule is configured to determine the hotspot event feature encoding sequence corresponding to the hotspot event sequence; A cross-learning submodule is configured to perform inter-event feature cross-learning on the hotspot event feature encoding sequence to obtain the hotspot event preference vector; wherein, the cross-learning submodule includes: The third determining unit is configured to determine the user attribute feature encoding vector corresponding to the user; The second cross-learning unit is configured to input the hot topic event feature encoding sequence and the user attribute feature encoding vector into the second feature cross-learning model to obtain the output hot topic event preference vector; wherein, the second feature cross-learning model is a converter model for inter-event feature cross-learning and user attribute feature and event feature cross-learning, and the second feature cross-learning model introduces user-side features to complete the feature cross-learning between user features and time features; the determination module is configured to determine the target hot topic event based on the hot topic event preference vector and candidate hot topic events, wherein, the hot topic event preference vector is obtained by performing inter-event feature cross-learning on hot topic events in the user's historical hot topic event viewing list; The push module is configured to push the target hot topic event to the user.

9. The apparatus according to claim 8, wherein, The third determining submodule includes: The first determining unit is configured to determine the event name encoding sequence corresponding to the hotspot event sequence; The second determining unit is configured to determine the hotspot event feature encoding sequence based on the event name encoding sequence.

10. The apparatus according to claim 9, wherein, The second determining unit includes: The first determining subunit is configured to determine a heat value encoding sequence corresponding to the hot event sequence; The encoding fusion subunit is configured to add the event name encoding sequence and the popularity value encoding sequence bit by bit to obtain the hot event feature encoding sequence.

11. The apparatus according to claim 9, wherein, The cross-learning submodule includes: The first cross-learning unit is configured to input the hot event feature encoding sequence into the first feature cross-learning model to obtain the output hot event preference vector; wherein, the first feature cross-learning model is a converter model for inter-event feature cross-learning.

12. The apparatus according to claim 9, wherein, The preset viewing duration is the average viewing duration of all hot events in the historical hot event viewing list.

13. The apparatus according to any one of claims 8 to 12, wherein, The determining module includes: The fourth determining submodule is configured to determine the predicted viewing duration of the candidate hot events based on the hot event preference vector and the candidate hot events; The fifth determination submodule is configured to determine the target hot topic event based on the predicted viewing duration.

14. The apparatus according to claim 13, wherein, The fourth determining submodule includes: The sixth determining unit is configured to determine the feature encoding vector of the candidate hotspot event; The splicing unit is configured to splice the feature encoding vector and the hotspot event preference vector to obtain a spliced ​​vector; The prediction unit is configured to input the concatenated vector into a preset viewing duration prediction model to obtain the predicted viewing duration of the candidate hot events.

15. An electronic device comprising: At least one processor; as well as The memory communicatively connected to the at least one processor, wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.

16. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method described in any one of claims 1-7.

17. A computer program product comprising a computer program that, when executed by a processor, implements the method of any one of claims 1-7.

Images