A data processing method, a computer device and a readable storage medium

Abstract

Embodiments of the present application provide a data processing method, a computer device and a readable storage medium. The method can be applied to various scenes such as cloud technology, artificial intelligence, intelligent transportation and multimedia. The method comprises: obtaining business data associated with a target object, determining a sharing attribute feature of the target object for the business data; generating an initial business trigger feature and an initial business conversion feature according to the sharing attribute feature; performing full connection processing on the initial business trigger feature to obtain a target business trigger feature, and generating a trigger probability estimate of the target object for the business data according to the target business trigger feature; performing feature sharing and transmission processing on the target business trigger feature and the initial business conversion feature to obtain a target business conversion feature, and generating a conversion probability estimate of the target object for the business data according to the target business conversion feature. The present application can improve the accuracy of the trigger probability estimate and the conversion probability estimate of the predicted business data.

Description

Technical Field

[0001] This application relates to the field of Internet technology, and in particular to a data processing method, computer device, and readable storage medium. Background Technology

[0002] With the development of multimedia technology, multimedia data (such as text and images) has become the main medium for obtaining information and enjoying entertainment in daily life. Business data recommendation systems can distribute text and images to application clients while simultaneously delivering business data. It's understood that business data recommendation systems typically determine the business data to be delivered to application clients based on click-through rates and conversion rates.

[0003] Existing business data recommendation systems use deep learning models to train click-through rate (CTR) prediction models and conversion rate (CTR) prediction models, respectively. These models then extract click and conversion features from business data (e.g., business data G). The conversion rate prediction model is trained using sample business data that shows a click-through rate to a conversion rate, meaning all sample business data exhibits click behavior. However, the business data G used for prediction in the conversion rate prediction model does not exhibit click behavior. This inconsistency between the training and prediction data distributions leads to inaccurate conversion features extracted from business data G, thus reducing the model's overall prediction accuracy. Summary of the Invention

[0004] This application provides a data processing method, a computer device, and a readable storage medium, which can improve the accuracy of the predicted trigger probability and conversion probability of business data.

[0005] One embodiment of this application provides a data processing method, including:

[0006] Obtain business data associated with the target object and determine the shared attribute characteristics of the target object in relation to the business data;

[0007] Based on shared attribute characteristics, generate initial business trigger characteristics and initial business conversion characteristics of the target object for business data;

[0008] The initial business triggering features are processed by a full connection to obtain the target business triggering features. Based on the target business triggering features, the triggering probability estimate of the target object for the business data is generated.

[0009] The target business triggering features and initial business conversion features are processed by feature sharing and transmission to obtain the target business conversion features. Based on the target business conversion features, a conversion probability estimate of the target object for the business data is generated.

[0010] One embodiment of this application provides a data processing apparatus, including:

[0011] The feature determination module is used to acquire business data associated with the target object and determine the shared attribute features of the target object in relation to the business data.

[0012] The feature generation module is used to generate initial business trigger features and initial business conversion features for the target object based on shared attribute features.

[0013] The first prediction module is used to perform full-connection processing on the initial business triggering features to obtain the target business triggering features, and generate a trigger probability prediction value of the target object for the business data based on the target business triggering features.

[0014] The second prediction module is used to perform feature sharing and transmission processing on the target business trigger features and the initial business conversion features to obtain the target business conversion features. Based on the target business conversion features, a conversion probability prediction value of the target object for the business data is generated.

[0015] The feature determination module includes:

[0016] The attribute input unit is used to obtain a set of business data associated with the target object and to retrieve business data from the set of business data.

[0017] The attribute input unit is used to input the object attributes of the target object, the business attributes of the business data, and the context attributes associated with the target object into the input network layer of the target network model;

[0018] The feature embedding unit is used to embed the features of object attributes through the input network layer to obtain the object features corresponding to the object attributes.

[0019] The feature embedding unit is used to embed features of business attributes through the input network layer to obtain the business features corresponding to the business attributes.

[0020] The feature embedding unit is used to embed the context attributes into the input network layer to obtain the context features corresponding to the context attributes.

[0021] The feature concatenation unit is used to concatenate object features, business features, and context features to obtain the shared attribute features of the target object for business data.

[0022] Among them, object attributes include object identifier, basic attributes, and behavioral attributes;

[0023] The feature embedding unit is specifically used to perform a hash operation on the object identifier in the input network layer to obtain the hash object identifier corresponding to the object identifier;

[0024] The feature embedding unit is specifically used to obtain the object identifier lookup table associated with the object identifier, and to search for the object identifier feature corresponding to the object identifier in the object identifier lookup table based on the hash object identifier;

[0025] The feature embedding unit is specifically used to obtain the basic attribute lookup table associated with the basic attribute, and to look up the basic attribute feature corresponding to the basic attribute in the basic attribute lookup table.

[0026] The feature embedding unit is specifically used to obtain a behavior attribute lookup table associated with the behavior attribute, and to look up the behavior attribute feature corresponding to the behavior attribute in the behavior attribute lookup table.

[0027] The feature embedding unit is specifically used to perform feature fusion on object identification features, basic attribute features and behavioral attribute features to obtain object features corresponding to object attributes.

[0028] The target network model also includes a parameter-sharing network layer; the parameter-sharing network layer includes a shared fully connected layer, a first triggering fully connected layer and a transformation fully connected layer, which are respectively connected to the shared fully connected layer;

[0029] The feature generation module includes:

[0030] The shared processing unit is used to input the shared attribute features into the shared fully connected layer in the parameter sharing network layer, and to perform fully connected processing on the shared attribute features through the shared fully connected layer to obtain fully connected shared features.

[0031] The first processing unit is used to input the fully connected shared features into the first triggering fully connected layer, and to perform full-connection processing on the fully connected shared features through the first triggering fully connected layer to generate the initial business triggering features of the target object for the business data.

[0032] The second processing unit is used to input the fully connected shared features into the transformation fully connected layer, and to perform full-connection processing on the fully connected shared features through the transformation fully connected layer to generate the initial business transformation features of the target object for the business data.

[0033] The transformation full-connection layer includes a first shallow transformation full-connection layer and a first deep transformation full-connection layer; the initial business transformation features include initial shallow transformation features and initial deep transformation features.

[0034] The second processing unit is specifically used to input the fully connected shared features into the first shallow transformation fully connected layer, and to perform full connection processing on the fully connected shared features through the first shallow transformation fully connected layer to generate the initial shallow transformation features of the target object for the business data.

[0035] The second processing unit is specifically used to input the fully connected shared features into the first deep transformation fully connected layer, and to perform full connection processing on the fully connected shared features through the first deep transformation fully connected layer to generate the initial deep transformation features of the target object for the business data.

[0036] The target network model also includes a multi-layer perceptron layer; the multi-layer perceptron layer includes a second triggered fully connected layer.

[0037] The first prediction module is specifically used to input the initial business trigger features into the second trigger fully connected layer, and to perform fully connected processing on the initial business trigger features through the second trigger fully connected layer to obtain the target business trigger features.

[0038] Among them, the initial business conversion characteristics include initial shallow conversion characteristics and initial deep conversion characteristics; the target business conversion characteristics include target shallow conversion characteristics and target deep conversion characteristics; the conversion probability estimate includes shallow conversion probability estimate and deep conversion probability estimate;

[0039] The second prediction module includes:

[0040] The first splicing unit is used to splice the target business trigger features and the initial shallow conversion features to obtain shared shallow splicing features. The shared shallow splicing features are then fully connected to obtain the target shallow conversion features.

[0041] The first generation unit is used to generate a shallow conversion probability estimate of the target object for business data based on the shallow conversion characteristics of the target.

[0042] The second splicing unit is used to splice the target shallow transformation features and the initial deep transformation features to obtain shared deep splicing features. The shared deep splicing features are then fully connected to obtain the target deep transformation features.

[0043] The second generation unit is used to generate a deep conversion probability estimate of the target object for business data based on the deep conversion characteristics of the target.

[0044] The multilayer perceptron layer also includes a second shallow transformation fully connected layer and a shallow connection layer;

[0045] The first splicing unit is specifically used to input the target business triggering features and the initial shallow transformation features into the shallow connection layer, and to splice the target business triggering features and the initial shallow transformation features through the shallow connection layer to obtain shared shallow splicing features;

[0046] The first splicing unit is specifically used to input the shared shallow splicing features into the second shallow transformation fully connected layer, and to perform fully connected processing on the shared shallow splicing features through the second shallow transformation fully connected layer to obtain the target shallow transformation features.

[0047] The multilayer perceptron layer also includes a second deep transformation fully connected layer and a deep connection layer;

[0048] The second splicing unit is specifically used to input the target shallow transformation features and the initial deep transformation features into the deep connection layer, and to splice the target shallow transformation features and the initial deep transformation features through the deep connection layer to obtain shared deep splicing features;

[0049] The second splicing unit is specifically used to input the shared deep splicing features into the second deep transformation fully connected layer, and to perform fully connected processing on the shared deep splicing features through the second deep transformation fully connected layer to obtain the target deep transformation features.

[0050] The device also includes:

[0051] The calculation and processing module is used to perform calculations on the trigger probability estimate, conversion probability estimate, and estimated resources corresponding to the business data to obtain the business data quality of the target object for the business data.

[0052] The sorting module is used to perform a first sorting process on the business data based on the quality of the business data, so as to obtain the initial sorting information corresponding to the business data.

[0053] The sorting processing module is used to perform a second sorting process on the business data based on the sorting strategy and the initial sorting information to obtain the target sorting information corresponding to the business data.

[0054] One embodiment of this application provides a data processing method, including:

[0055] Obtain sample business data associated with the sample object, and determine the sample-shared attribute characteristics of the sample object for the sample business data through the initial network model;

[0056] In the initial network model, based on the shared attribute characteristics of the samples, the initial sample service triggering characteristics and initial sample service conversion characteristics of the sample objects for the sample service data are generated;

[0057] The initial sample business triggering features are processed by a full connection to obtain the target sample business triggering features. Based on the target sample business triggering features, a sample triggering probability prediction value for the sample object relative to the sample business data is generated.

[0058] The target sample business triggering features and the initial sample business conversion features are processed by feature sharing and transmission to obtain the target sample business conversion features. Based on the target sample business conversion features, a sample conversion probability prediction value for the sample object relative to the sample business data is generated.

[0059] Based on the sample label information, sample trigger probability prediction, and sample conversion probability prediction of the sample object for the sample business data, the parameters of the initial network model are adjusted, and the parameter-adjusted initial network model is used as the target network model; the target network model is used to predict the trigger probability prediction and conversion probability prediction of the target object for the business data.

[0060] One embodiment of this application provides a data processing apparatus, including:

[0061] The sample feature determination module is used to acquire sample business data associated with the sample object and determine the sample-shared attribute features of the sample object in relation to the sample business data through the initial network model.

[0062] The sample feature generation module is used to generate initial sample service trigger features and initial sample service transformation features for sample objects based on sample shared attribute features in the initial network model.

[0063] The trigger probability prediction module is used to perform fully connected processing on the initial sample business trigger features to obtain the target sample business trigger features, and generate a sample trigger probability prediction value for the sample object relative to the sample business data based on the target sample business trigger features.

[0064] The conversion probability prediction module is used to perform feature sharing and transmission processing on the target sample business trigger features and the initial sample business conversion features to obtain the target sample business conversion features. Based on the target sample business conversion features, a sample conversion probability prediction value for the sample object relative to the sample business data is generated.

[0065] The parameter adjustment module is used to adjust the parameters of the initial network model based on the sample label information, sample trigger probability prediction, and sample conversion probability prediction of the sample object for the sample business data. The parameter-adjusted initial network model is used as the target network model. The target network model is used to predict the trigger probability prediction and conversion probability prediction of the target object for the business data.

[0066] The sample tag information includes business trigger tag information and business conversion tag information;

[0067] The parameter adjustment module includes:

[0068] The loss determination unit is used to determine the trigger loss value of the initial network model based on the business trigger label information and the sample trigger probability prediction of the sample business data of the sample object.

[0069] The loss determination unit is used to determine the conversion loss value of the initial network model based on the business conversion label information of the sample object for the sample business data and the sample conversion probability prediction value.

[0070] The loss determination unit is used to determine the model loss value of the initial network model based on the trigger loss value and the transformation loss value.

[0071] The parameter adjustment unit is used to adjust the parameters of the initial network model based on the model loss value. When the parameter-adjusted initial network model meets the model convergence condition, the parameter-adjusted initial network model is used as the target network model.

[0072] One embodiment of this application provides a computer device, including: a processor and a memory;

[0073] The processor is connected to a memory, which stores a computer program. When the computer program is executed by the processor, it causes the computer device to perform the method provided in the embodiments of this application.

[0074] One aspect of this application provides a computer-readable storage medium storing a computer program adapted to be loaded and executed by a processor, so that a computer device having the processor performs the method provided in this application.

[0075] One embodiment of this application provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method provided in this application embodiment.

[0076] In this embodiment, the computer device can acquire business data associated with a target object, determine the shared attribute features of the target object regarding the business data, and then generate initial business triggering features and initial business conversion features of the target object regarding the business data based on the shared attribute features. Further, the computer device can perform full connection processing on the initial business triggering features to obtain target business triggering features, and then generate a trigger probability estimate of the target object regarding the business data based on the target business triggering features. Further, the computer device can perform feature sharing and transfer processing on the target business triggering features and initial business conversion features to obtain target business conversion features, and then generate a conversion probability estimate of the target object regarding the business data based on the target business conversion features. Therefore, this embodiment can extract the initial business triggering features and initial business conversion features of the target object regarding the business data based on the shared attribute features of the target object regarding the business data, and then generate the target business triggering features corresponding to the initial business triggering features. These target business triggering features can be used to generate a trigger probability estimate of the target object regarding the business data. It is understandable that by performing feature sharing and transfer processing on the target business triggering features and the initial business conversion features, the target business triggering features corresponding to the business triggering task can be transferred to the initial business conversion features corresponding to the business conversion task. This allows the business triggering task and the business conversion task to share underlying parameters, which is equivalent to sharing the underlying parameters corresponding to the business triggering task with the business conversion task. This generates target business conversion features with higher accuracy, which can be used to generate a conversion probability prediction value for the target object based on business data. Therefore, the embodiments of this application can simultaneously determine the trigger probability prediction value and the conversion probability prediction value, and based on the target business triggering features and the target business conversion features, improve the accuracy of the trigger probability prediction value and the conversion probability prediction value for predicting business data. Attached Figure Description

[0077] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0078] Figure 1 This is a schematic diagram of a network architecture provided in an embodiment of this application;

[0079] Figure 2 This is a schematic diagram of a data interaction scenario provided in an embodiment of this application;

[0080] Figure 3This is a flowchart illustrating a data processing method provided in an embodiment of this application;

[0081] Figure 4 This is a flowchart illustrating a data processing method provided in an embodiment of this application;

[0082] Figure 5 This is a flowchart illustrating a data processing method provided in an embodiment of this application;

[0083] Figure 6 This is a schematic diagram of the structure of a network model provided in an embodiment of this application;

[0084] Figure 7 This is a schematic diagram of a business data recommendation process provided in an embodiment of this application;

[0085] Figure 8 This is a flowchart illustrating a data processing method provided in an embodiment of this application;

[0086] Figure 9 This is a schematic diagram of the structure of a data processing device provided in an embodiment of this application;

[0087] Figure 10 This is a schematic diagram of the structure of a data processing device provided in an embodiment of this application;

[0088] Figure 11 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0089] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0090] It should be understood that Artificial Intelligence (AI) is the theory, methods, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess the functions of perception, reasoning, and decision-making.

[0091] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, as well as machine learning / deep learning, autonomous driving, and intelligent transportation.

[0092] The solutions provided in this application primarily relate to Machine Learning (ML) technology in artificial intelligence. Machine learning is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory, and many other disciplines. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and instructional learning.

[0093] For details, please see Figure 1 , Figure 1 This is a schematic diagram of a network architecture provided in an embodiment of this application. Figure 1 As shown, this network architecture may include a service server 2000 and a terminal device cluster. The terminal device cluster may specifically include one or more terminal devices; the number of terminal devices in the cluster is not limited here. Figure 1 As shown, the multiple terminal devices may specifically include terminal device 3000a, terminal device 3000b, terminal device 3000c, ..., terminal device 3000n; terminal device 3000a, terminal device 3000b, terminal device 3000c, ..., terminal device 3000n can be directly or indirectly connected to the business server 2000 via wired or wireless communication, so that each terminal device can interact with the business server 2000 through the network connection.

[0094] Each terminal device in the terminal device cluster can include: smart TVs, smartphones, tablets, laptops, desktop computers, smart home devices, wearable devices, in-vehicle terminals, and other smart terminals with data processing capabilities. It should be understood that, for example... Figure 1 Each terminal device in the terminal device cluster shown can be integrated with an application client. When the application client runs on each terminal device, it can interact with the aforementioned... Figure 1The business servers 2000 shown interact with each other. The application client can be a standalone client or an embedded sub-client integrated into another client; this application does not impose any limitations on this.

[0095] Specifically, the application clients can include browsers, in-vehicle clients, smart home clients, entertainment clients, multimedia clients (e.g., video clients), social clients, and information clients (e.g., news clients) with data processing capabilities. The in-vehicle terminal can be a smart terminal in a smart transportation scenario, and the application clients on this in-vehicle terminal can be the aforementioned in-vehicle clients.

[0096] Among them, the business server 2000 can be the server corresponding to the application client. The business server 2000 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms.

[0097] For ease of understanding, the embodiments of this application may be described in detail below. Figure 1 From the multiple terminal devices shown, one terminal device is selected as the target terminal device. For example, in the embodiments of this application, a terminal device can be selected as the target terminal device. Figure 1 The terminal device 3000b shown serves as the target terminal device, which may integrate an application client with data processing capabilities. In this case, the target terminal device can interact with the business server 2000 through this application client.

[0098] For ease of understanding, in this application embodiment, the user corresponding to the target terminal device can be collectively referred to as the target object; in this application embodiment, the business data recommended by the business server 2000 to the target object can be collectively referred to as recommended business data; and in this application embodiment, the business data sent by the advertiser (i.e., a legal person, other economic organization, or individual who designs, produces, or publishes advertisements themselves or entrusts others to do so for the purpose of promoting goods or providing services) to the business server 2000 can be collectively referred to as business data to be recommended. In this application embodiment, the business data can be advertising data (i.e., advertisements).

[0099] Understandably, the target object can send a business data retrieval request to the business server 2000 through the application client in the target terminal device. Upon receiving the request, the business server 2000 can retrieve a set of business data associated with the target object, extract recommended business data from this set, and then send this recommended business data to the target object's corresponding application client. Therefore, after receiving the recommended business data from the business server 2000, the application client can display the recommended business data, allowing the target object to determine whether to perform a trigger operation based on the recommended business data.

[0100] In this context, ad impressions represent the products (i.e., recommendation data) that a target audience observes in the application client. The triggering actions performed by the target audience in response to this recommendation data can be categorized into three types: The first type is ad clicks, which represent the target audience's click behavior on the recommendation data; the second type is shallow ad conversions, which represent the shallow conversion behavior of the target audience after clicking on the recommendation data, such as downloading, activating, or registering; and the third type is deep ad conversions, which represent the deep conversion behavior of the target audience after clicking on the recommendation data, such as paying for services or maintaining user retention the following day.

[0101] Among them, shallow conversion behavior is based on click behavior, that is, click behavior comes first and then shallow conversion behavior; deep conversion behavior is based on shallow conversion behavior, that is, shallow conversion behavior comes first and then deep conversion behavior.

[0102] It should be understood that the embodiments of this application can measure the behaviors corresponding to the above three types of triggering operations using different metrics. CTR (Click Through Rate) represents the ratio from ad impression to ad click, i.e., ad clicks divided by impressions; CVR (Conversion Rate) represents the ratio from ad clicks to ad conversions, i.e., ad conversions divided by clicks. CVR can be divided into CVR1 (shallow conversion rate) and CVR2 (deep conversion rate). CVR1 represents the ratio from ad clicks to shallow ad conversions, i.e., shallow conversions divided by clicks; CVR2 represents the ratio from ad clicks to deep ad conversions, i.e., deep conversions divided by clicks.

[0103] It should be understood that when advertisers send data for recommended services, they need to set the corresponding business objectives. These business objectives can be divided into shallow-target ads and deep-target ads. Shallow-target ads refer to ads where the bidding objective includes a shallow-target bid, i.e., a single bid for shallow metrics. This shallow-target bid can represent a bid for shallow conversion behaviors (i.e., shallow metrics, such as downloads, activations, and registrations). Shallow-target ads require estimated click-through rates and shallow conversion rates. Deep-target ads refer to ads where the bidding objective includes both shallow-target bids and deep-target bids, i.e., a bid that involves bidding again for deep metrics based on a single bid. This deep-target bid can represent a bid for deep conversion behaviors (i.e., deep metrics, such as payments and next-day retention). Deep-target ads require estimated click-through rates, shallow conversion rates, and deep conversion rates. Furthermore, in this application's embodiments, shallow-target ads can be referred to as single-target ads, and deep-target ads as multi-target ads.

[0104] In this application, the Bayesian network is a probabilistic graphical model, also known as a directed acyclic graph model. A Bayesian network is a graphical network model used to describe uncertain causal relationships between variables. It consists of nodes, directed connections, and a node probability table, where the directed connections represent causal dependencies between nodes.

[0105] It is understood that the aforementioned network framework can be applied to the field of advertising recommendation. Specific business scenarios in this field may include: advertising distribution scenarios, advertising search scenarios, and advertising viewing scenarios, etc., which will not be listed here. It should be understood that this application does not limit the format in which business data is displayed.

[0106] For example, in an advertising distribution scenario, a computer device (e.g., the aforementioned business server 2000) can analyze the user profile of the target audience to determine recommended business data (e.g., business data G1) that the target audience may be interested in. Then, when the target audience opens the application client, the distributed content and business data G1 are pushed to the application client. Here, the distributed content can be video data, text and image data, or audio data; there are no specific limitations.

[0107] For example, in an advertising search scenario, a computer device (e.g., the aforementioned business server 2000) can analyze the search text entered by the target audience in the application client, determine recommended business data (e.g., business data G2) that the target audience may be interested in, and then push the search content corresponding to the search text and business data G2 to the application client. Here, the search content can be video data, image and text data, or audio data; there are no specific limitations.

[0108] For example, in an advertising viewing scenario, a computer device (e.g., the aforementioned business server 2000) can analyze the contextual content viewed by the target object in the application client, determine the recommended business data (e.g., business data G3) that the target object may be interested in, and then push the business data G3 to the application client.

[0109] It is understood that in the specific implementation of this application, data related to object attributes, context attributes, etc. are involved. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0110] For better understanding, please refer to [link / reference]. Figure 2 , Figure 2 This is a schematic diagram illustrating a data interaction scenario provided in an embodiment of this application. For example... Figure 2 The server 20a shown can be the one described above. Figure 1 The corresponding business server 2000 in the embodiment, such as Figure 2 The terminal device 20b shown can be the one described above. Figure 1 The target terminal device in the corresponding embodiment. The terminal device 20b has an application client installed, which can be used to display recommended business data associated with the object 20c corresponding to the terminal device 20b.

[0111] Understandably, object 20c can send a business data retrieval request to server 20a through an application client in terminal device 20b. This allows it to view multimedia data (e.g., video data, text and image data) tailored to its interests, along with recommended business data matching its preferences, within the application client. For example, when object 20c performs a search operation on the application client, the application client can send a business data retrieval request to server 20a, so that server 20a returns recommended business data associated with the search results along with the search results. Similarly, when object 20c performs an open operation on the application client, the application client can send a business data retrieval request to server 20a, so that server 20a returns recommended business data associated with the distributed content along with the distributed content. Likewise, when object 20c performs a news feed viewing operation on the application client, the application client can send a business data retrieval request to server 20a, so that server 20a returns recommended business data associated with the news feed content along with the news feed content.

[0112] For ease of understanding, this application embodiment uses the example of a business data acquisition request being triggered by a search operation. The specific processes of opening operations, information flow viewing operations, etc., can be found in the description of the search operation, and will not be repeated here.

[0113] like Figure 2 As shown, server 20a can receive business data retrieval requests sent by application clients, and then retrieve a set of business data associated with object 20c (e.g., business data set 21b) from business database 21a. Business database 21a can include multiple databases, specifically databases 22a, 22b, ..., 22n. Databases 22a, 22b, ..., 22n can be used to store different types of business data. For example, database 22a can be used to store business data related to animals, database 22b can be used to store business data related to plants, ..., database 22n can be used to store business data related to space.

[0114] When the search text for object 20c is "cat," server 20a can retrieve a set of business data 21b associated with object 20c from database 22a. This means server 20a can construct an object vector for object 20c and a business vector for the business data in database 22a, and then use methods such as cosine similarity or Euclidean distance to determine the vector similarity between the object vector and the business vector. This allows server 20a to obtain business data whose vector similarity meets a threshold, and the business data retrieved from database 22a constitutes the business data set 21b. The business data set 21b can include multiple sets of business data, specifically business data 23a, business data 23b, ..., business data 23m.

[0115] Optionally, it should be understood that server 20a can also simultaneously obtain business data from databases 22a, 22b, ..., 22n to form business data set 21b. Here, we will take the example of business data 23a, business data 23b, ..., business data 23m being obtained from database 22a for illustration.

[0116] like Figure 2As shown, after obtaining the business data set 21b, server 20a can sequentially retrieve business data from the business data set 21b. Here, we will use business data 23a as an example. Server 20a can determine the shared attribute characteristics of object 20c for business data 23a, and then generate initial business trigger characteristics and initial business conversion characteristics for object 20c for business data 23a based on these shared attribute characteristics. The initial business trigger characteristics correspond to the characteristics of object 20c for click behavior, and the initial business conversion characteristics correspond to the characteristics of object 20c for conversion behavior.

[0117] Furthermore, server 20a can perform full-connection processing on the initial business trigger features to obtain target business trigger features, and then generate a trigger probability estimate for object 20c for business data 23a based on the target business trigger features. Simultaneously, the server can perform feature sharing and transfer processing on the target business trigger features and the initial business conversion features to obtain target business conversion features, and then generate a conversion probability estimate for object 20c for business data 23a based on the target business conversion features. Here, the trigger probability estimate corresponds to the estimate for object 20c regarding click behavior, and the conversion probability estimate corresponds to the estimate for object 20c regarding conversion behavior.

[0118] like Figure 2 As shown, after determining the estimated trigger probability and the estimated conversion probability for each business data in the business data set 21b, server 20a sorts each business data in the business data set 21b based on the estimated trigger probability and the estimated conversion probability for each business data, obtains recommended business data from each sorted business data, uses the recommended business data as the business data that matches object 20c, and returns the recommended business data to the application client.

[0119] Therefore, this embodiment of the application can integrate the intermediate features of the business triggering task (i.e., the target business triggering feature obtained after performing full connection processing on the initial business triggering feature) into the business conversion task. This involves performing feature sharing and transfer processing on the target business triggering feature and the initial business conversion feature to obtain the target business conversion feature. The initial business triggering feature and the initial business conversion feature are generated from shared attribute features. It can be understood that the feature sharing and transfer processing is equivalent to passing the underlying parameters of the business triggering task to the business conversion task. The target business triggering feature generated in this way can improve the accuracy of the predicted trigger probability estimate of the business data, and the target business conversion feature can improve the accuracy of the predicted conversion probability estimate of the business data.

[0120] Further, please see Figure 3 , Figure 3 This is a flowchart illustrating a data processing method provided in an embodiment of this application. The method can be executed by a server, by an application client, or by both a server and an application client. The server can be one of the aforementioned... Figure 2 The server 20a in the corresponding embodiment can be the application client described above. Figure 2 The corresponding application client in the embodiments. For ease of understanding, this application embodiment uses the method executed by the server as an example for illustration. The data processing method may include the following steps S101-S104:

[0121] Step S101: Obtain business data associated with the target object and determine the shared attribute characteristics of the target object for the business data;

[0122] Specifically, the server can acquire a set of business data associated with the target object and extract business data from this set. Further, the server can input the object attributes of the target object, the business attributes of the business data, and the context attributes associated with the target object into the input network layer of the target network model. Further, the server can perform feature embedding on the object attributes through the input network layer to obtain object features (i.e., user-side features) corresponding to the object attributes. Further, the server can perform feature embedding on the business attributes through the input network layer to obtain business features (i.e., advertising-side features, item features) corresponding to the business attributes. Further, the server can perform feature embedding on the context attributes through the input network layer to obtain context features corresponding to the context attributes. Further, the server can concatenate the object features, business features, and context features to obtain the shared attribute features of the target object for each piece of business data in the business data set. The server can determine the shared attribute features of the target object for each piece of business data in the business data set.

[0123] It is understandable that object attributes can be used to represent target objects, business attributes can be used to represent business data, and context attributes can be used to represent the terminal device corresponding to the target object. Specifically, object attributes may include, but are not limited to, the target object's object identifier, basic attributes, and behavioral attributes (i.e., behavioral interests); business attributes may include, but are not limited to, the business identifier of business data (i.e., advertisement identifier), the business data object identifier (i.e., advertiser identifier), the business data category (i.e., advertisement category, e.g., application client advertisement), the business data text (i.e., semantic features), and the business data image (i.e., image features); and context attributes may include, but are not limited to, the terminal device's device content (i.e., context content in the application client) and device type (e.g., smartphone).

[0124] Among these, the object identifier of the target object is unique, with one target object corresponding to one object identifier; the business identifier of the business data is unique, with one piece of business data corresponding to one business identifier. It should be understood that the target object does not necessarily have all object attributes, and the business data does not necessarily have all business attributes.

[0125] It can be understood that the input network layer can include a feature embedding layer and a feature concatenation layer. The feature embedding layer can be used to embed features of object attributes, business attributes and context attributes, and the feature concatenation layer can be used to concatenate features of object features, business features and context features.

[0126] The target network model is obtained by iteratively training the initial network model. The specific process of the server iteratively training the initial network model to obtain the target network model can be found below. Figure 8 The description of steps S301-S305 in the corresponding embodiment.

[0127] It should be understood that the target network model also includes a parameter-sharing network layer, which can be used to perform the following step S102. The parameter-sharing network layer includes a shared fully connected layer, a first triggering fully connected layer, and a transformation fully connected layer, each connected to the shared fully connected layer.

[0128] Step S102: Based on the shared attribute characteristics, generate the initial business trigger characteristics and initial business conversion characteristics of the target object for the business data;

[0129] Specifically, the server can input shared attribute features into the shared fully connected layer in the parameter sharing network layer. The shared fully connected layer performs fully connected processing on the shared attribute features to obtain fully connected shared features. Further, the server can input the fully connected shared features into the first triggering fully connected layer. The first triggering fully connected layer performs fully connected processing on the fully connected shared features to generate initial business trigger features for the target object based on the business data. Further still, the server can input the fully connected shared features into the transformation fully connected layer. The transformation fully connected layer performs fully connected processing on the fully connected shared features to generate initial business transformation features for the target object based on the business data.

[0130] It should be understood that the target network model also includes a multilayer perceptron layer (i.e., a Bayesian network layer), wherein the multilayer perceptron layer includes a second triggered fully connected layer, which can be used to perform the following step S103.

[0131] Step S103: Perform full connection processing on the initial business triggering features to obtain the target business triggering features. Based on the target business triggering features, generate the triggering probability estimate of the target object for the business data.

[0132] Specifically, the server can input the initial business triggering features into the second triggering fully connected layer, which then performs full-connection processing on the initial business triggering features to obtain the target business triggering features. Furthermore, the server can generate a predicted trigger probability value for the target object based on the target business triggering features.

[0133] The target business triggering feature is composed of a first triggering feature value and a second triggering feature value. The first triggering feature value represents the probability that the business data will trigger a click, and the second triggering feature value represents the probability that the business data will not trigger a click. Therefore, the server can use the first triggering feature value as an estimated value of the triggering probability of the target object for the business data.

[0134] Step S104: Perform feature sharing and transmission processing on the target business trigger features and the initial business conversion features to obtain the target business conversion features. Based on the target business conversion features, generate a conversion probability estimate of the target object for the business data.

[0135] Specifically, the server can concatenate the target business trigger features and the initial business conversion features to obtain shared concatenated features. Then, it can perform a full-connection process on these shared concatenated features to obtain the target business conversion features. Furthermore, the server can generate a conversion probability estimate for the target object based on the business data, using these target business conversion features.

[0136] The target business conversion characteristic is composed of a first conversion characteristic value and a second conversion characteristic value. The first conversion characteristic value represents the probability that the business data will lead to a conversion behavior, and the second conversion characteristic value represents the probability that the business data will not lead to a conversion behavior. Therefore, the server can use the first conversion characteristic value as an estimated value of the conversion probability of the target object for the business data.

[0137] Therefore, this embodiment of the application can extract the initial business triggering features and initial business conversion features of the target object for business data based on the shared attribute features of the target object for business data, and then generate the target business triggering features corresponding to the initial business triggering features. These target business triggering features can be used to generate the trigger probability prediction value of the target object for business data. It is understood that by performing feature sharing and transfer processing on the target business triggering features and initial business conversion features, the target business triggering features corresponding to the business triggering task can be transferred to the initial business conversion features corresponding to the business conversion task, allowing the business triggering task and the business conversion task to share underlying parameters. This is equivalent to sharing the underlying parameters corresponding to the business triggering task with the business conversion task, thereby generating a target business conversion feature with higher accuracy. These target business conversion features can be used to generate the conversion probability prediction value of the target object for business data. Therefore, this embodiment of the application can simultaneously determine the trigger probability prediction value and the conversion probability prediction value, and based on the target business triggering features and the target business conversion features, improve the accuracy of the predicted trigger probability prediction value and the predicted conversion probability prediction value of business data.

[0138] Further, please see Figure 4 , Figure 4 This is a flowchart illustrating a data processing method provided in an embodiment of this application. The data processing method may include the following steps S1011-S1016, and steps S1011-S1016 are... Figure 3 A specific embodiment of step S101 in the corresponding embodiment.

[0139] Step S1011: Obtain the business data set associated with the target object, and retrieve business data from the business data set;

[0140] Understandably, the server can obtain a set of business data associated with the target object through a recall model. This recall model can find as much business data as possible that is highly relevant to the target object from massive amounts of business data, and then perform the following steps S1012-S1016 on the business data that is highly relevant to the target object.

[0141] The recall model can comprehensively consider the object attributes of the target object, the business attributes of the business data, the contextual attributes associated with the target object, and the search terms associated with the target object. The search terms can represent the search text entered by the target object when performing a search operation in the application client.

[0142] Recall models can be broadly categorized into two types. The first type uses keyword-based hard matching, such as gender (male, female) in business data targeting, which divides business data into two categories: one for males and the other for females. Another example is search terms in search text (e.g., "cat"), which retrieves business data associated with cats. The second type uses vector-based soft matching, constructing object vectors for the target object and business vectors for the business data, and then using methods such as cosine similarity or Euclidean distance to determine the vector similarity between the object vector and the business vector.

[0143] Step S1012: Input the object attributes of the target object, the business attributes of the business data, and the context attributes associated with the target object into the input network layer of the target network model;

[0144] Among them, object attributes may include, but are not limited to, object identifier, basic attributes, and behavioral attributes; business attributes may include, but are not limited to, business identifier, business data object identifier, business data category, business data text, and business data image; context attributes may include, but are not limited to, device content and device type.

[0145] Step S1013: The object attributes are embedded by inputting the network layer to obtain the object features corresponding to the object attributes;

[0146] Specifically, the server can perform a hash operation on the object identifier at the input network layer to obtain the hash object identifier corresponding to the object identifier. Further, the server can obtain an object identifier lookup table associated with the object identifier, and search for the object identifier feature corresponding to the object identifier in the object identifier lookup table based on the hash object identifier. Further, the server can obtain a basic attribute lookup table associated with basic attributes, and search for the basic attribute feature corresponding to the basic attribute in the basic attribute lookup table. Further, the server can obtain a behavior attribute lookup table associated with behavior attributes, and search for the behavior attribute feature corresponding to the behavior attribute in the behavior attribute lookup table. Further, the server can perform feature fusion on the object identifier feature, basic attribute feature, and behavior attribute feature to obtain the object feature corresponding to the object attribute.

[0147] It is understandable that the object identifier lookup table, basic attribute lookup table, and behavioral attribute lookup table can each include multiple target features (i.e., object identifier features, basic attribute features, and behavioral attribute features). The multiple target features in the object identifier lookup table are obtained by training the initial features in the initial object identifier lookup table; the multiple target features in the basic attribute lookup table are obtained by training the initial features in the initial basic attribute lookup table; and the multiple target features in the behavioral attribute lookup table are obtained by training the initial features in the initial behavioral attribute lookup table. Here, the initial object identifier lookup table, initial basic attribute lookup table, and initial behavioral attribute lookup table are the lookup tables in the initial input network layer of the initial network model, where the initial input network layer corresponds to the input network layer.

[0148] The basic attributes can include, but are not limited to, the gender and age of the target object. Therefore, the basic attribute lookup table can include an age lookup table corresponding to age and a gender lookup table corresponding to gender. In this way, the server can look up the age feature corresponding to age in the age lookup table and the gender feature corresponding to gender in the gender lookup table, and then perform feature fusion on the age feature and gender feature to obtain the basic attribute feature corresponding to the basic attribute.

[0149] The number of behavioral attributes can be one or more, where "multiple" can mean at least two. Understandably, when there is only one behavioral attribute (e.g., basketball), the server can look up the behavioral attribute feature corresponding to that attribute in the behavioral attribute lookup table (i.e., look up the behavioral attribute feature corresponding to basketball). Optionally, when there are multiple behavioral attributes (e.g., volleyball, vegetables), the server can look up the behavioral attribute features corresponding to each attribute in the behavioral attribute lookup table, and then perform feature fusion on these multiple behavioral attribute features (i.e., fuse the behavioral attribute features corresponding to volleyball and the behavioral attribute features corresponding to vegetables) to obtain the behavioral attribute feature corresponding to each attribute.

[0150] It is understood that the dimensions of object identification features, basic attribute features, and behavioral attribute features are the same. The server can perform feature fusion on object identification features, basic attribute features, and behavioral attribute features in the following ways: feature concatenation (e.g., concatenating three 128-dimensional features to obtain a 384-dimensional object feature), feature addition (e.g., adding the corresponding dimensions of three 128-dimensional features to obtain a 128-dimensional object feature), or weighted averaging (e.g., obtaining a 128-dimensional object feature based on the three 128-dimensional features and their respective weights). It should be understood that the embodiments of this application do not limit the dimensions of object identification features, basic attribute features, and behavioral attribute features, nor do they limit the specific method of feature fusion.

[0151] Optionally, the server can obtain the object attribute text of the object attributes, perform vector transformation on the object attribute text to obtain the object attribute vector corresponding to the object attribute text, and then input the object attribute vector into the feature embedding layer in the input network layer. The feature embedding layer encodes the object attribute vector to obtain the object vector corresponding to the object attribute text. Further, the server can determine the object features corresponding to the object attributes based on the object vector. Specifically, the server can extract the object attribute text from the object attributes. For example, when the object attribute is a behavioral attribute, the behavioral attribute can be "volleyball," then the object attribute text can be "volleyball" and "ball." It should be understood that the feature embedding layer can be the encoder structure in the Transformer model; this embodiment does not limit the structure of the feature embedding layer.

[0152] Step S1014: The business attributes are embedded by inputting the network layer to obtain the business features corresponding to the business attributes.

[0153] Specifically, the server can perform a hash operation on the service identifier at the input network layer to obtain the hashed service identifier corresponding to the service identifier. Further, the server can obtain a service identifier lookup table associated with the service identifier, and search for the corresponding service identifier feature based on the hashed service identifier. Further, the server can obtain a service data object identifier lookup table associated with the service data object identifier, and search for the corresponding service data object identifier feature. Further, the server can obtain a service data category lookup table associated with the service data category, and search for the corresponding service data category feature. Further, the server can obtain a service data text lookup table associated with the service data text, and search for the corresponding service data text feature. Further, the server can obtain a service data image lookup table associated with the service data image, and search for the corresponding service data image feature. Further, the server can perform feature fusion on the service identifier feature, service data object identifier feature, service data category feature, service data text feature, and service data image feature to obtain the service feature corresponding to the service attribute.

[0154] It is understood that the business identifier lookup table, business data object identifier lookup table, business data category lookup table, business data text lookup table, and business data image lookup table can each include multiple target features (i.e., business identifier features, business data object identifier features, business data category features, business data text features, and business data image features), and these multiple target features are obtained after training the initial features. It should be understood that the specific process by which the server trains the initial features to obtain the target features can be found in the description of step S1013 above, and will not be repeated here.

[0155] It should be understood that the specific process of the server searching the business data category lookup table can be found in the description of searching the behavior attribute lookup table above, and will not be repeated here. Similarly, it should be understood that the specific process of the server searching the business data object identifier lookup table can be found in the description of searching the basic attribute lookup table above, and will not be repeated here.

[0156] Specifically, the server can perform text analysis on business data text to obtain the corresponding business data text key, and then look up the corresponding business data text feature in a business data text lookup table based on the business data text key. Similarly, the server can perform image analysis on business data images to obtain the corresponding business data image key, and then look up the corresponding business data image feature in a business data image lookup table based on the business data image key.

[0157] It should be understood that the specific process of the server performing feature fusion on business identification features, business data object identification features, business data category features, business data text features, and business data image features can be found in the above description of feature fusion on object identification features, basic attribute features, and behavioral attribute features, and will not be repeated here.

[0158] Optionally, the server can obtain the business attribute text of the business attribute, perform vector transformation on the business attribute text to obtain the corresponding business attribute vector, and then input the business attribute vector into the feature embedding layer in the input network layer. The feature embedding layer encodes the business attribute vector to obtain the business vector corresponding to the business attribute text. Further, the server can determine the business features corresponding to the business attribute based on the business vector. Wherein, when the business attribute is a business data image, the business data text can be the text data in the business data image. The server can extract the business attribute text from the business attribute. For example, when the business attribute is business data text, the business attribute text can be "Please download our latest application client!", then the business attribute text can be "Please", "You", "Download", "Our", "Latest", "Version", "Application", "Client", and "!".

[0159] Step S1015: Embed the context attributes into features by inputting the network layer to obtain the context features corresponding to the context attributes;

[0160] Specifically, the server can obtain a device content lookup table associated with device content at the input network layer, and look up the device content features corresponding to the device content in the device content lookup table. Furthermore, the server can obtain a device type lookup table associated with device type, and look up the device type features corresponding to the device type in the device type lookup table. Further, the server can perform feature fusion on the device content features and device type features to obtain the context features corresponding to the context attributes.

[0161] It is understood that the device content lookup table and the device type lookup table can each include multiple target features (i.e., device content features and device type features), which are obtained after training the initial features. It should be understood that the specific process by which the server trains the initial features to obtain the target features can be found in the description of step S1013 above, and will not be repeated here.

[0162] It should be understood that the specific process of the server looking up the device content lookup table can be found in the description of looking up the business data text lookup table above, and will not be repeated here. Similarly, it should be understood that the specific process of the server looking up the device content type lookup table can be found in the description of looking up the basic attribute lookup table above, and will not be repeated here.

[0163] Optionally, the server can obtain the context attribute text of the context attribute, perform vector transformation on the context attribute text to obtain the context attribute vector corresponding to the context attribute text, and then input the context attribute vector into the feature embedding layer in the input network layer. The feature embedding layer encodes the context attribute vector to obtain the context vector corresponding to the context attribute text. Further, the server can determine the context features corresponding to the context attribute based on the context vector. Specifically, the server can extract the context attribute text from the context attribute; for example, when the context attribute is device content, the business attribute text could be "Sports can improve physical fitness."

[0164] Step S1016: Perform feature concatenation on object features, business features, and context features to obtain the shared attribute features of the target object for business data.

[0165] The dimensions of object features, business features, and context features can be N, where N can be a positive integer. Therefore, the dimension of shared attribute features is 3N (i.e., 3*N). It should be understood that this application does not limit the value of N.

[0166] It should be understood that before training the initial network model, the server can initialize the initial features in the lookup table (i.e., the embedding table). Then, during the training of the initial network model, the server can update the initial features in the lookup table based on the backpropagation gradient to obtain a lookup table containing the target features. Furthermore, for high-dimensional features (e.g., business identifier features), the server can map the high-dimensional features to low-dimensional features before storing the low-dimensional features in the lookup table.

[0167] It should be understood that shallow conversion rates are generally higher than deep conversion rates, and shallow conversion data can be returned on the same day, while deep conversion data has a long delay (usually more than a week). Therefore, deep conversion data is characterized by high sparsity and high latency. Thus, in the embodiments of this application, the target business triggering features, target shallow conversion features, and target deep conversion features can share underlying parameters (i.e., object attributes, business attributes, and context attributes), thereby improving the accuracy of the estimated trigger probability, shallow conversion probability, and deep conversion probability, and thus solving the conversion rate prediction problem in high-sparse and high-latency scenarios.

[0168] Further, please see Figure 5 , Figure 5 This is a flowchart illustrating a data processing method provided in an embodiment of this application. The method can be executed by a server, by an application client, or by both a server and an application client. The server can be one of the aforementioned... Figure 2The server 20a in the corresponding embodiment can be the application client described above. Figure 2 The corresponding application client in the embodiments. For ease of understanding, this application embodiment uses the method executed by the server as an example for illustration. The data processing method may include the following steps S201-S211:

[0169] Step S201: Obtain business data associated with the target object and determine the shared attribute characteristics of the target object for the business data;

[0170] The specific process by which the server determines the shared attribute characteristics of the target object for business data can be found in the above. Figure 3 The description of step S101 in the corresponding embodiments will not be repeated here.

[0171] Step S202: Perform fully connected processing on the shared attribute features to obtain fully connected shared features;

[0172] Specifically, the server can input the shared attribute features into the shared fully connected layer in the parameter sharing network layer, and then perform fully connected processing on the shared attribute features through the shared fully connected layer to obtain the fully connected shared features.

[0173] Step S203: Perform full-connection processing on the fully connected shared features to generate the initial business trigger features of the target object for the business data;

[0174] Specifically, the server can input the fully connected shared features into the first triggered fully connected layer in the parameter sharing network layer. The first triggered fully connected layer performs fully connected processing on the fully connected shared features to generate the initial business trigger features of the target object for the business data.

[0175] Step S204: Perform full-connection processing on the fully connected shared features to generate the initial shallow transformation features of the target object for the business data;

[0176] Specifically, the server can input the fully connected shared features into the first shallow transformation fully connected layer in the parameter sharing network layer, and perform full connection processing on the fully connected shared features through the first shallow transformation fully connected layer to generate the initial shallow transformation features of the target object for the business data.

[0177] Step S205: Perform full-connection processing on the fully connected shared features to generate the initial deep transformation features of the target object for the business data;

[0178] Specifically, the server can input the fully connected shared features into the first deep transformation fully connected layer in the parameter sharing network layer. The first deep transformation fully connected layer performs full connection processing on the fully connected shared features to generate the initial deep transformation features of the target object for the business data.

[0179] It should be understood that the first shallow conversion fully connected layer and the first deep conversion fully connected layer can be collectively referred to as the conversion fully connected layer; the initial shallow conversion features and the initial deep conversion features can be collectively referred to as the initial business conversion features. In other words, the conversion fully connected layer includes the first shallow conversion fully connected layer and the first deep conversion fully connected layer; the initial business conversion features include the initial shallow conversion features and the initial deep conversion features.

[0180] Step S206: Perform full-connection processing on the initial service triggering features to obtain the target service triggering features;

[0181] Specifically, the server can input the initial service trigger features into the second fully connected trigger layer in the multilayer perceptual network layer, and then perform fully connected processing on the initial service trigger features through the second fully connected trigger layer to obtain the target service trigger features.

[0182] Step S207: Based on the target business triggering characteristics, generate a trigger probability estimate of the target object for the business data;

[0183] The specific process for the server to generate the predicted trigger probability can be found in the above. Figure 3 The description of step S103 in the corresponding embodiments will not be repeated here.

[0184] Step S208: Perform feature concatenation on the target business trigger feature and the initial shallow conversion feature to obtain the shared shallow concatenation feature, and perform full connection processing on the shared shallow concatenation feature to obtain the target shallow conversion feature;

[0185] Specifically, the server can input the target service trigger feature and the initial shallow transformation feature into the shallow connection layer. The shallow connection layer then concatenates these features to obtain a shared shallow concatenated feature. The multilayer perceptron layer also includes a second shallow transformation fully connected layer and a shallow connection layer. Further, the server can input the shared shallow concatenated feature into the second shallow transformation fully connected layer, which then performs fully connected processing on the shared shallow concatenated feature to obtain the target shallow transformation feature.

[0186] Step S209: Based on the shallow conversion characteristics of the target, generate a shallow conversion probability estimate of the target object for the business data;

[0187] The target shallow conversion feature is composed of a first shallow conversion feature value and a second shallow conversion feature value. The first shallow conversion feature value represents the probability that the business data will exhibit shallow conversion behavior, and the second shallow conversion feature value represents the probability that the business data will not exhibit shallow conversion behavior. Therefore, the server can use the first shallow conversion feature value as an estimated value of the shallow conversion probability of the target object for the business data.

[0188] Step S210: Perform feature concatenation on the target shallow transformation features and the initial deep transformation features to obtain shared deep concatenation features. Perform full connection processing on the shared deep concatenation features to obtain the target deep transformation features.

[0189] Specifically, the server can input the target shallow transformation features and the initial deep transformation features into a deep connected layer. The deep connected layer then concatenates these features to obtain shared deep concatenated features. The multilayer perceptron layer also includes a second fully connected deep transformation layer and a deep connected layer. Furthermore, the server can input the shared deep concatenated features into the second fully connected deep transformation layer, which then performs fully connected processing on these features to obtain the target deep transformation features.

[0190] Step S211: Based on the target's deep conversion characteristics, generate a predicted value of the target object's deep conversion probability for business data.

[0191] The target deep conversion feature is composed of a first deep conversion feature value and a second deep conversion feature value. The first deep conversion feature value represents the probability that the business data will exhibit deep conversion behavior, and the second deep conversion feature value represents the probability that the business data will not exhibit deep conversion behavior. Therefore, the server can use the first deep conversion feature value as an estimated value of the probability of deep conversion of the target object for the business data.

[0192] It should be understood that the target shallow conversion characteristics and the target deep conversion characteristics can be collectively referred to as target business conversion characteristics; the shallow conversion probability estimate and the deep conversion probability estimate can be collectively referred to as conversion probability estimate. In other words, target business conversion characteristics include target shallow conversion characteristics and target deep conversion characteristics; conversion probability estimate includes shallow conversion probability estimate and deep conversion probability estimate.

[0193] It should be understood that the shared fully connected layer, the first triggered fully connected layer, the second triggered fully connected layer, the first shallow transformation fully connected layer, the second shallow transformation fully connected layer, the first deep transformation fully connected layer, and the second deep transformation fully connected layer can be a multi-layer perceptron (MLP). A multi-layer perceptron is a deep learning architecture that can non-linearly combine extracted features to obtain an output. It should be understood that the embodiments of this application do not limit the model structure of the shared fully connected layer, the first triggered fully connected layer, the second triggered fully connected layer, the first shallow transformation fully connected layer, the second shallow transformation fully connected layer, the first deep transformation fully connected layer, and the second deep transformation fully connected layer.

[0194] Understandably, the specific process by which the server determines the predicted trigger probability, shallow conversion probability, and deep conversion probability for the target object based on business data can be found in [reference needed]. Figure 6 , Figure 6 This is a schematic diagram of the structure of a network model provided in an embodiment of this application. Figure 6 The network structure 40a shown may include three network layers, specifically network layer 40b, network layer 40c, and network layer 40d. Figure 6 When the network structure 40a shown corresponds to the model prediction process, the network structure 40a can be the target network model, which is obtained by training the initial network model.

[0195] Among them, Figure 6 When the network structure 40a shown is the target network model, network layer 40b can be called the input network layer, network layer 40c can be called the parameter sharing network layer, and network layer 40d can be called the multilayer perceptron network layer.

[0196] like Figure 6 As shown, the server can input object attributes, business attributes, and context attributes into the input network layer 40b. The feature embedding layer in the input network layer 40b embeds features of the object attributes, business attributes, and context attributes to obtain object features corresponding to the object attributes, business features corresponding to the business attributes, and context features corresponding to the context attributes. Furthermore, the server can use the feature concatenation layer in the input network layer 40b (not shown in the diagram; for ease of understanding, the diagram shows the feature embedding layer and feature concatenation layer as the same layer) to concatenate the object features, business features, and context features to obtain the shared attribute features of the target object for the business data.

[0197] like Figure 6As shown, the server can input the shared attribute features output from the input network layer 40b to the parameter sharing network layer 40c, and the shared fully connected layer in the parameter sharing network layer 40c outputs the fully connected shared features. Further, the server can input the fully connected shared features to the first triggering fully connected layer, the first shallow transformation fully connected layer, and the first deep transformation fully connected layer in the shared network layer 40c, respectively. The first triggering fully connected layer outputs the initial service triggering features, the first shallow transformation fully connected layer outputs the initial shallow transformation features, and the first deep transformation fully connected layer outputs the initial deep transformation features. Wherein, Figure 6 The first shallow transformation fully connected layer and the first deep transformation fully connected layer shown can be collectively referred to as the transformation fully connected layer.

[0198] like Figure 6 As shown, the server can input the initial service trigger feature into the multilayer perceptron layer 40d, and the second fully connected trigger layer in the multilayer perceptron layer 40d outputs the target service trigger feature. Further, the server can perform feature concatenation between the target service trigger feature and the initial shallow transformation feature through the shallow connection layer in the multilayer perceptron layer 40d to obtain a shared shallow concatenation feature, which is then output by the second fully connected shallow transformation layer in the multilayer perceptron layer 40d. Further still, the server can perform feature concatenation between the target shallow transformation feature and the initial deep transformation feature through the deep connection layer in the multilayer perceptron layer 40d to obtain a shared deep concatenation feature, which is then output by the second fully connected deep transformation layer in the multilayer perceptron layer 40d.

[0199] like Figure 6 As shown, the server can generate a trigger probability estimate based on the target business trigger characteristics, a shallow conversion probability estimate based on the target shallow conversion characteristics, and a deep conversion probability estimate based on the target deep conversion characteristics.

[0200] It should be understood that the server can process the estimated trigger probability, estimated conversion probability, and estimated resources corresponding to the business data to obtain the business data quality of the target object for the business data. Here, estimated resources can represent the advertiser's shallow target bid and / or deep target bid for the business data, and business data quality can represent the ECPM (Effective Cost Per Mille) metric, which represents the advertising revenue that can be obtained per thousand ad impressions. Furthermore, the server can perform a first sorting process on the business data based on the business data quality to obtain the initial sorting information corresponding to the business data. Further, the server can perform a second sorting process on the business data using the sorting strategy and the initial sorting information to obtain the target sorting information corresponding to the business data.

[0201] Among them, the server can perform calculations on the estimated trigger probability, estimated conversion probability, and estimated resources using multiplication. That is, the server can use the product of the estimated trigger probability, estimated conversion probability, and estimated resources as the business data quality.

[0202] It is understandable that when the server receives a business data acquisition request sent by the target object, it can determine the target object's trigger probability estimate for the business data (i.e., the click-through rate estimate ectr(ExpectClick ThroughRate)), the shallow conversion probability estimate (i.e., the shallow conversion rate estimate ecvr1(ExpectConversion Rate 1)), and the deep conversion probability estimate (i.e., the deep conversion rate estimate ecvr2(ExpectConversion Rate 2)). The click-through rate estimate (hereinafter referred to as click-through rate) can be represented by the following formula (1), the shallow conversion rate estimate (hereinafter referred to as shallow conversion rate) can be represented by the following formula (2), and the deep conversion rate estimate (hereinafter referred to as deep conversion rate) can be represented by the following formula (3):

[0203] ectr=P(ectr|x,H) (1)

[0204] ecvr1=P(ecvr1|ectr,x,H) (2)

[0205] ecvr2=P(ecvr2|ectr,ecvr1,x,H) (3)

[0206] Here, x can represent all input features (i.e., object attributes, business attributes, and context attributes), and H can represent the parameters of the target network model (i.e., the parameters of the shared fully connected layer, the first triggered fully connected layer, the second triggered fully connected layer, the first shallow conversion fully connected layer, the second shallow conversion fully connected layer, the first deep conversion fully connected layer, and the second deep conversion fully connected layer). Therefore, P(ectr|x,H) can represent the predicted click-through rate output by the target network model, P(ecvr1|ectr,x,H) can represent the predicted shallow conversion rate output by the target network model, and P(ecvr2|ectr,ecvr1,x,H) can represent the predicted deep conversion rate output by the target network model.

[0207] Understandably, when the business data belongs to shallow target advertising, the server can determine the quality of the business data by referring to the following formula (4):

[0208] ecpm=ectr*ecvr1*bid1 (4)

[0209] Here, ectr can represent the estimated click probability (i.e., the estimated click-through rate) of a single-target ad, ecvr1 can represent the estimated shallow conversion probability (i.e., the estimated shallow conversion rate) of a single-target ad, and bid1 can represent the shallow target bid of a single-target ad.

[0210] Optionally, when the business data belongs to deep target ads, if the shallow conversion rate estimate is greater than the deep conversion rate estimate, the server can determine the quality of the business data by referring to the following formula (5). If the shallow conversion rate estimate is not greater than (i.e. less than or equal to) the deep conversion rate estimate, the server can determine the quality of the business data by referring to the following formula (6).

[0211] ecpm=ectr*ecvr1*bid1 (5)

[0212] ecpm=ectr*ecvr2*bid2 (6)

[0213] Here, ectr can represent the estimated click probability (i.e., the estimated click-through rate) of a multi-objective ad, ecvr1 can represent the estimated shallow conversion probability (i.e., the estimated shallow conversion rate) of a multi-objective ad, ecvr2 can represent the estimated deep conversion probability (i.e., the estimated deep conversion rate) of a multi-objective ad, bid1 can represent the shallow target bid of a multi-objective ad, and bid2 can represent the deep target bid of a multi-objective ad.

[0214] For easier understanding, please refer to Figure 7 , Figure 7 This is a schematic diagram of a business data recommendation process provided in an embodiment of this application. For example... Figure 7 As shown, the application client can initiate a request (i.e., a business data retrieval request) to the server through step S71. Then, the server can retrieve and recall advertisements from the advertisement database through step S72. This advertisement database can be the one described above. Figure 2 In the corresponding embodiment, the business database 21a can be business data (i.e., advertisements) in the business data set obtained through the recall model.

[0215] like Figure 7As shown, the server can execute step S73, in which the click-through rate (CTR) of the advertisement is estimated using the target network model, and then execute step S74, in which the business type of the business data is determined. This business type can include single-target advertisements and multi-target advertisements. It can be understood that when the business data is a single-target advertisement, the server can execute step S75, in which the shallow conversion rate of the advertisement is estimated using the target network model, and then execute step S76, calculating the quality (ECPM) of the single-target advertisement based on the CTR and shallow conversion rate. Optionally, it can be understood that when the business data is a multi-target advertisement, the server can execute step S77, in which the shallow and deep conversion rates of the advertisement are estimated using the target network model, and then execute step S78, calculating the quality of the multi-target advertisement based on the CTR, shallow conversion rate, and deep conversion rate.

[0216] Therefore, after calculating the quality of a single-target advertisement, the server can sort the single-target advertisements based on their quality to obtain initial sorting information; similarly, after calculating the quality of a multi-target advertisement, the server can sort the multi-target advertisements based on their quality to obtain initial sorting information.

[0217] Furthermore, such as Figure 7 As shown, in step S79, the server can reorder the advertisements according to ranking strategies such as ad diversity, frequency control, category control, and specific result weighting to obtain target ranking information for single-target or multi-target advertisements. Then, step S80 is executed, whereby the server retrieves the final top-ranked advertisement from the target ranking information and exposes this advertisement (i.e., recommendation business data) to the user (i.e., the target audience) on media (i.e., multimedia data displayed in the application client). Ad diversity, frequency control, and category control can be collectively referred to as ranking strategies. Optionally, the server can also retrieve the final top-ranked advertisement from the initial ranking information without considering ranking strategies and expose this advertisement (i.e., recommendation business data) to the target audience on media.

[0218] Therefore, the target network model in this embodiment can include three tasks: a click-through rate (CTR) prediction task, a shallow conversion rate (SCRR) prediction task, and a deep conversion rate (DCPR) prediction task. Each task can share the same shared fully connected layer. This sharing mechanism allows the shallow and deep DCPR prediction tasks to share the underlying parameters of the CTR prediction task. Specifically, the target business triggering features of the CTR prediction task are shared with the DCPR prediction task (i.e., the shallow and deep DCPR prediction models). More specifically, the target business triggering features of the CTR prediction task are shared with the shallow DCPR prediction task, and the target shallow conversion features of the shallow DCPR prediction task are shared with the deep DCPR prediction task. This improves the accuracy of the predicted trigger probability, shallow conversion probability, and deep conversion probability. Therefore, when using the aforementioned highly accurate trigger probability prediction and conversion probability prediction to sort business data, relatively accurate sorting information can be obtained. Consequently, when using this sorting information to recommend business data to target objects, the accuracy of business data recommendations can be improved, similar ads can be avoided, and the user experience of the advertising recommendation system (i.e., the business data recommendation system) can be enhanced.

[0219] Further, please see Figure 8 , Figure 8 This is a flowchart illustrating a data processing method provided in an embodiment of this application. The method can be executed by a server, by an application client, or by both a server and an application client. The server can be one of the aforementioned... Figure 2 The server 20a in the corresponding embodiment can be the application client described above. Figure 2 The corresponding application client in the embodiments. For ease of understanding, this application embodiment uses the method executed by the server as an example for explanation. The data processing method may include the following steps S301-S305:

[0220] Step S301: Obtain sample business data associated with the sample object, and determine the sample-shared attribute characteristics of the sample object for the sample business data through the initial network model;

[0221] Among them, the sample objects have click behavior, shallow conversion behavior or deep conversion behavior for the sample business data, and the number of sample business data is one or more.

[0222] The initial network model may include an initial input network layer. This initial input network layer determines the sample-shared attribute characteristics of the sample objects relative to the sample business data. This initial input network layer corresponds to the input network layer in the target network model. It should be understood that the specific process by which the server determines the sample-shared attribute characteristics through the initial input network layer can be found in the description above regarding determining the target object's shared attribute characteristics for business data through the input network layer; this will not be repeated here.

[0223] Step S302: In the initial network model, based on the sample shared attribute characteristics, generate the initial sample service triggering characteristics and initial sample service conversion characteristics of the sample object for the sample service data;

[0224] The initial network model may include an initial parameter-sharing network layer. This layer generates initial sample service triggering features and initial sample service transformation features for the sample objects based on the sample service data. This initial parameter-sharing network layer corresponds to the parameter-sharing network layer in the target network model. It should be understood that the specific process by which the server generates the initial sample service triggering features and initial sample service transformation features through the initial parameter-sharing network layer can be found in the description above, and will not be repeated here.

[0225] The initial sample business conversion characteristics include the initial sample shallow conversion characteristics and the initial sample deep conversion characteristics.

[0226] Step S303: Perform full-connection processing on the initial sample business triggering features to obtain the target sample business triggering features. Based on the target sample business triggering features, generate a sample triggering probability prediction value for the sample object against the sample business data.

[0227] The initial network model may include an initial multilayer perceptron layer. This initial multilayer perceptron layer generates a sample trigger probability prediction for the sample object based on the sample business data. This initial multilayer perceptron layer corresponds to the multilayer perceptron layer in the target network model. It should be understood that the specific process by which the server generates the sample trigger probability prediction through the initial multilayer perceptron layer can be found in the description above regarding the generation of the trigger probability prediction through the multilayer perceptron layer; it will not be repeated here.

[0228] Step S304: Perform feature sharing and transmission processing on the target sample business triggering features and the initial sample business conversion features to obtain the target sample business conversion features. Based on the target sample business conversion features, generate a sample conversion probability prediction value for the sample object against the sample business data.

[0229] Specifically, the initial multilayer perceptron layer can generate a sample conversion probability estimate for the sample object based on the sample business data. It should be understood that the specific process by which the server generates the sample conversion probability estimate through the initial multilayer perceptron layer is described above. Figure 3 The description of generating the conversion probability prediction through the multilayer perceptron layer in step S104 of the corresponding embodiment, or refer to the above. Figure 5 The description of generating shallow conversion probability estimates and deep conversion probability estimates (i.e., conversion probability estimates can include both shallow and deep conversion probability estimates) through the multilayer perceptual network layer in steps S208-S211 of the corresponding embodiment will not be repeated here.

[0230] Among them, the sample conversion probability prediction includes the sample shallow conversion probability prediction and the sample deep conversion probability prediction, and the target sample business conversion characteristics include the target sample shallow conversion characteristics and the target sample deep conversion characteristics.

[0231] Step S305: Based on the sample label information, sample trigger probability prediction and sample conversion probability prediction of the sample object for the sample business data, adjust the parameters of the initial network model, and use the parameter-adjusted initial network model as the target network model.

[0232] Specifically, the server can determine the trigger loss value of the initial network model based on the business trigger tag information and the sample trigger probability prediction of the sample object for the sample business data. The sample tag information includes business trigger tag information and business conversion tag information. Further, the server can determine the conversion loss value of the initial network model based on the business conversion tag information and the sample conversion probability prediction of the sample object for the sample business data. Further, the server can determine the model loss value of the initial network model based on the trigger loss value and the conversion loss value. Further, the server can adjust the parameters of the initial network model based on the model loss value. When the parameter-adjusted initial network model meets the model convergence condition, it is used as the target network model. The target network model can be used to predict the trigger probability prediction and conversion probability prediction of the target object for the business data.

[0233] The business conversion tag information includes shallow conversion tag information and deep conversion tag information. It can be understood that the server can determine the shallow conversion loss value of the initial network model based on the shallow conversion tag information of the sample object for the sample business data and the sample shallow conversion probability prediction value. Furthermore, the server can determine the deep conversion loss value of the initial network model based on the deep conversion tag information of the sample object for the sample business data and the sample deep conversion probability prediction value. Furthermore, the server can determine the conversion loss value of the initial network model based on both the shallow and deep conversion loss values.

[0234] It should be understood that the specific process of the server adjusting the parameters of the initial network model based on the model loss value (i.e., iterative training) can be described as follows: When the model loss value of the initial network model does not meet the model convergence condition, the server can adjust the model parameters of the initial network model based on the model loss value that does not meet the model convergence condition. Furthermore, the server can determine the initial network model after adjusting the model parameters as a transitional network model, and iteratively train the transitional network model until the model loss value of the iteratively trained transitional network model meets the model convergence condition. The transitional network model that meets the model convergence condition is then used as the target network model.

[0235] It should be understood that the initial network model and the target network model can be collectively referred to as the generalized network model. The initial network model and the target network model are names of the generalized network model at different times. During the training phase, the generalized network model can be called the initial network model, and during the prediction phase, the generalized network model can be called the target network model.

[0236] It is understandable that when the server iteratively trains the initial network model, it can output multiple target probability values ​​P(ectr,ecvr1,ecvr2|x,H) (i.e., sample conversion probability prediction, shallow sample conversion probability prediction, and deep sample conversion probability prediction) from the initial network model. The server can decompose the joint distribution into smaller individual distributions and obtain the probability values ​​output by the initial network model based on the assumptions of the Bayesian probability model. The probability values ​​output by the initial network model can be found in the following formula (7):

[0237]

[0238] Here, x can represent all input features (i.e., sample object attributes, sample business attributes, and sample context attributes), and H can represent the parameters of the initial network model. Therefore, P(ectr|x,H) can represent the sample conversion probability prediction value output by the initial network model, P(ecvr1|ectr,x,H) can represent the sample shallow conversion probability prediction value output by the target network model, and P(ecvr2|ectr,ecvr1,x,H) can represent the sample deep conversion probability prediction value output by the target network model.

[0239] It is understandable that taking the negative log-likelihood loss function (i.e., negative log-likelihood) of formula (7) yields the loss function L(x,H) shown in formula (8) below:

[0240]

[0241] It is understandable that by adding weight parameters (or simply weights) to different objectives in formula (8), we can obtain the loss function L(x,H) containing the weight parameters as shown in formula (9) below:

[0242]

[0243] Here, W1, W2, and W3 can represent the weights of the three loss functions respectively. W1 represents the weight corresponding to the loss function P(ectr|x,H), W2 can represent the weight corresponding to P(ecvr1|ectr,x,H), and W3 can represent the weight corresponding to P(ecvr2|ectr,ecvr1,x,H).

[0244] Understandably, the weight parameters can represent the click behavior, shallow conversion behavior, or deep conversion behavior of the sample object in response to the sample business data. For example, when the sample object has click behavior but no shallow or deep conversion behavior, W1 can be equal to 1, and W2 and W3 can be equal to 0; or, when the sample object has click behavior and shallow conversion behavior but no deep conversion behavior, W1 and W2 can be equal to 1, and W3 can be equal to 0; or, when the sample object has click behavior, shallow conversion behavior, and deep conversion behavior, W1, W2, and W3 can be equal to 1.

[0245] It should be understood that the initial network model can be derived from... Figure 6 It consists of three parts, as shown. Figure 6The diagram shows a portion that can be used to fit ectr, another portion to fit ecvr1, and yet another portion to fit ecvr2. Therefore, the initial network model has three subtasks, which can be used to output ectr, ecvr1, and ecvr2 respectively. Since this embodiment can predict multiple targets—ectr, ecvr1, and ecvr2—based on a Bayesian network, the goal of the server training the initial network model is to optimize P(ectr, ecvr1, ecvr2|x, H).

[0246] Please see again. Figure 6 ,exist Figure 6 The network structure 40a shown corresponds to the model training process, and network structure 40a can be the initial network model. Specifically, in... Figure 6 When the network structure 40a shown is the initial network model, network layer 40b can be called the initial input network layer, network layer 40c can be called the initial parameter sharing network layer, and network layer 40d can be called the initial multilayer perceptron network layer.

[0247] like Figure 6 As shown, the server can input the sample object attributes of the sample object, the sample business attributes of the sample business data, and the sample context attributes associated with the sample object into the initial input network layer 40b. The initial input network layer 40b then outputs the sample-shared attribute features of the sample object for the sample business data. At this time, the initial input network layer 40b can include the pre-training sub-network layers corresponding to the feature embedding layer and the feature concatenation layer, i.e. Figure 6 The feature embedding layer shown can represent a sub-network layer in the initial network model.

[0248] like Figure 6 As shown, the server can input the sample shared attribute features output by the initial input network layer 40b into the initial parameter sharing network layer 40c, and output the initial sample business trigger features, initial sample shallow transformation features, and initial sample deep transformation features through the initial parameter sharing network layer 40c. At this time, the initial parameter sharing network layer 40c can include the pre-training sub-network layers corresponding to the shared fully connected layer, the first trigger fully connected layer, the first shallow transformation fully connected layer, and the first deep transformation fully connected layer, i.e. Figure 6 The shared fully connected layer, the first triggered fully connected layer, the first shallow transformation fully connected layer, and the first deep transformation fully connected layer shown can represent sub-network layers in the initial network model.

[0249] like Figure 6As shown, the server can input the initial sample service trigger features, initial sample shallow transformation features, and initial sample deep transformation features into the initial multilayer perceptron layer 40d, and output the target sample service trigger features, target sample shallow transformation features, and target sample deep transformation features through the initial multilayer perceptron layer 40d. At this time, the initial multilayer perceptron layer 40d can include the sub-network layers before training corresponding to the shallow connection layer, deep connection layer, second trigger fully connected layer, second shallow transformation fully connected layer, and second deep transformation fully connected layer, i.e. Figure 6 The shallow connection layer, deep connection layer, second triggered fully connected layer, second shallow transformation fully connected layer, and second deep transformation fully connected layer shown can represent sub-network layers in the initial network model.

[0250] like Figure 6 As shown, the server can generate a sample trigger probability estimate based on the target sample's business trigger characteristics, a sample shallow conversion probability estimate based on the target sample's shallow conversion characteristics, and a sample deep conversion probability estimate based on the target sample's deep conversion characteristics. At this point, Figure 6 The predicted trigger probability shown represents the predicted trigger probability for the sample. Figure 6 The shallow conversion probability estimate shown represents the shallow conversion probability estimate of the sample. Figure 6 The deep conversion probability estimate shown represents the deep conversion probability estimate of the sample.

[0251] Therefore, this application proposes a multi-task model based on Bayesian networks, integrating click-through rate (CTR), shallow conversion rate (SCR), and deep conversion rate (DCR) into a single model. Intermediate training results for CTR are incorporated into the training of the shallow conversion rate objective, and vice versa. By simultaneously using CTR, SCR, and DCR as three objectives, the initial network model can fully utilize underlying parameters (i.e., sample object attributes, sample business attributes, and sample context attributes), alleviating the problem of sparse training data. This is equivalent to sharing the click prediction training samples of the CTR prediction model with the conversion rate prediction model, and sharing the shallow conversion samples of the shallow conversion rate prediction model with the deep conversion rate prediction model (in fact, the CTR prediction model, conversion rate prediction model, shallow conversion rate prediction model, and deep conversion rate prediction model all belong to the initial network model). This improves the accuracy of CTR, shallow conversion rate, and deep conversion rate predictions, thereby enhancing the accuracy of business data recommendations.

[0252] Further, please see Figure 9 , Figure 9This is a schematic diagram of the structure of a data processing device provided in an embodiment of this application. The data processing device 1 may include: a feature determination module 11, a feature generation module 12, a first estimation module 13, and a second estimation module 14; further, the data processing device 1 may also include: a calculation processing module 15 and a sorting processing module 16;

[0253] Feature determination module 11 is used to acquire business data associated with the target object and determine the shared attribute features of the target object in relation to the business data;

[0254] The feature determination module 11 includes: an attribute input unit 111, a feature embedding unit 112, and a feature splicing unit 113.

[0255] The attribute input unit 111 is used to obtain a set of business data associated with the target object and to obtain business data from the set of business data.

[0256] The attribute input unit 111 is used to input the object attributes of the target object, the business attributes of the business data, and the context attributes associated with the target object into the input network layer of the target network model;

[0257] The feature embedding unit 112 is used to embed the object attributes into the input network layer to obtain the object features corresponding to the object attributes.

[0258] Among them, object attributes include object identifier, basic attributes, and behavioral attributes;

[0259] The feature embedding unit 112 is specifically used to perform a hash operation on the object identifier in the input network layer to obtain the hash object identifier corresponding to the object identifier;

[0260] The feature embedding unit 112 is specifically used to obtain an object identifier lookup table associated with the object identifier, and to search for the object identifier feature corresponding to the object identifier in the object identifier lookup table based on the hash object identifier;

[0261] The feature embedding unit 112 is specifically used to obtain the basic attribute lookup table associated with the basic attribute, and to look up the basic attribute feature corresponding to the basic attribute in the basic attribute lookup table.

[0262] The feature embedding unit 112 is specifically used to obtain a behavior attribute lookup table associated with the behavior attribute, and to look up the behavior attribute feature corresponding to the behavior attribute in the behavior attribute lookup table.

[0263] The feature embedding unit 112 is specifically used to perform feature fusion on object identification features, basic attribute features and behavioral attribute features to obtain object features corresponding to object attributes.

[0264] The feature embedding unit 112 is used to embed features of business attributes through the input network layer to obtain business features corresponding to the business attributes.

[0265] The feature embedding unit 112 is used to embed the context attributes into the input network layer to obtain the context features corresponding to the context attributes.

[0266] The feature splicing unit 113 is used to splice object features, business features and context features to obtain the shared attribute features of the target object for business data.

[0267] The specific implementation methods of the attribute input unit 111, the feature embedding unit 112, and the feature concatenation unit 113 can be found in the above description. Figure 3 In the corresponding embodiment, steps S101 and Figure 4 The descriptions of steps S1011-S1016 in the corresponding embodiments will not be repeated here.

[0268] Feature generation module 12 is used to generate initial business trigger features and initial business conversion features of the target object for business data based on shared attribute features;

[0269] The target network model also includes a parameter-sharing network layer; the parameter-sharing network layer includes a shared fully connected layer, a first triggering fully connected layer and a transformation fully connected layer, which are respectively connected to the shared fully connected layer;

[0270] The feature generation module 12 includes: a shared processing unit 121, a first processing unit 122, and a second processing unit 123;

[0271] The shared processing unit 121 is used to input the shared attribute features into the shared fully connected layer in the parameter sharing network layer, and to perform fully connected processing on the shared attribute features through the shared fully connected layer to obtain fully connected shared features.

[0272] The first processing unit 122 is used to input the fully connected shared features into the first triggering fully connected layer, and to perform full-connection processing on the fully connected shared features through the first triggering fully connected layer to generate the initial business triggering features of the target object for the business data.

[0273] The second processing unit 123 is used to input the fully connected shared features into the transformation fully connected layer, and to perform full-connection processing on the fully connected shared features through the transformation fully connected layer to generate the initial business transformation features of the target object for the business data.

[0274] The transformation full-connection layer includes a first shallow transformation full-connection layer and a first deep transformation full-connection layer; the initial business transformation features include initial shallow transformation features and initial deep transformation features.

[0275] The second processing unit 123 is specifically used to input the fully connected shared features into the first shallow transformation fully connected layer, and to perform full connection processing on the fully connected shared features through the first shallow transformation fully connected layer to generate the initial shallow transformation features of the target object for the business data.

[0276] The second processing unit 123 is specifically used to input the fully connected shared features into the first deep transformation fully connected layer, and to perform full connection processing on the fully connected shared features through the first deep transformation fully connected layer to generate the initial deep transformation features of the target object for the business data.

[0277] The specific implementation methods of the shared processing unit 121, the first processing unit 122, and the second processing unit 123 can be found in the above description. Figure 3 In the corresponding embodiment, steps S102 and Figure 5 The descriptions of steps S202-S205 in the corresponding embodiments will not be repeated here.

[0278] The first prediction module 13 is used to perform full-connection processing on the initial business triggering features to obtain the target business triggering features, and generate a trigger probability prediction value of the target object for the business data based on the target business triggering features.

[0279] The target network model also includes a multi-layer perceptron layer; the multi-layer perceptron layer includes a second triggered fully connected layer.

[0280] The first prediction module 13 is specifically used to input the initial service trigger features into the second trigger fully connected layer, and to perform full connection processing on the initial service trigger features through the second trigger fully connected layer to obtain the target service trigger features.

[0281] The second prediction module 14 is used to perform feature sharing and transmission processing on the target business trigger features and the initial business conversion features to obtain the target business conversion features, and generate a conversion probability prediction value of the target object for the business data based on the target business conversion features.

[0282] Among them, the initial business conversion characteristics include initial shallow conversion characteristics and initial deep conversion characteristics; the target business conversion characteristics include target shallow conversion characteristics and target deep conversion characteristics; the conversion probability estimate includes shallow conversion probability estimate and deep conversion probability estimate;

[0283] The second prediction module 14 includes: a first splicing unit 141, a first generation unit 142, a second splicing unit 143, and a second generation unit 144;

[0284] The first splicing unit 141 is used to splice the target business trigger feature and the initial shallow conversion feature to obtain the shared shallow splicing feature, and to perform full connection processing on the shared shallow splicing feature to obtain the target shallow conversion feature.

[0285] The multilayer perceptron layer also includes a second shallow transformation fully connected layer and a shallow connection layer;

[0286] The first splicing unit 141 is specifically used to input the target business triggering feature and the initial shallow transformation feature into the shallow connection layer, and to splice the target business triggering feature and the initial shallow transformation feature through the shallow connection layer to obtain the shared shallow splicing feature.

[0287] The first splicing unit 141 is specifically used to input the shared shallow splicing features into the second shallow transformation fully connected layer, and to perform fully connected processing on the shared shallow splicing features through the second shallow transformation fully connected layer to obtain the target shallow transformation features.

[0288] The first generation unit 142 is used to generate a shallow conversion probability estimate of the target object for business data based on the shallow conversion characteristics of the target.

[0289] The second splicing unit 143 is used to splice the target shallow transformation features and the initial deep transformation features to obtain shared deep splicing features, and to perform fully connected processing on the shared deep splicing features to obtain the target deep transformation features.

[0290] The multilayer perceptron layer also includes a second deep transformation fully connected layer and a deep connection layer;

[0291] The second splicing unit 143 is specifically used to input the target shallow transformation features and the initial deep transformation features into the deep connection layer, and to splice the target shallow transformation features and the initial deep transformation features through the deep connection layer to obtain shared deep splicing features;

[0292] The second splicing unit 143 is specifically used to input the shared deep splicing features into the second deep transformation fully connected layer, and to perform fully connected processing on the shared deep splicing features through the second deep transformation fully connected layer to obtain the target deep transformation features.

[0293] The second generation unit 144 is used to generate a deep conversion probability estimate of the target object for business data based on the deep conversion characteristics of the target.

[0294] The specific implementation methods of the first splicing unit 141, the first generation unit 142, the second splicing unit 143, and the second generation unit 144 can be found in the above description. Figure 3 In the corresponding embodiment, steps S104 and Figure 5The descriptions of steps S208-S211 in the corresponding embodiments will not be repeated here.

[0295] Optionally, the calculation and processing module 15 is used to perform calculation and processing on the trigger probability estimate, the conversion probability estimate, and the estimated resources corresponding to the business data to obtain the business data quality of the target object for the business data.

[0296] The sorting processing module 16 is used to perform a first sorting process on the business data based on the quality of the business data, so as to obtain the initial sorting information corresponding to the business data.

[0297] The sorting processing module 16 is used to perform a second sorting process on the business data through a sorting strategy and initial sorting information to obtain the target sorting information corresponding to the business data.

[0298] The specific implementation methods of the feature determination module 11, feature generation module 12, first prediction module 13, second prediction module 14, calculation processing module 15, and sorting processing module 16 can be found in the above description. Figure 3 In the corresponding embodiment, steps S101-S104, Figure 4 In the corresponding embodiment, steps S1011-S1016 and Figure 5 The descriptions of steps S201-S211 in the corresponding embodiments will not be repeated here. Furthermore, the beneficial effects of using the same method will also not be repeated.

[0299] Further, please see Figure 10 , Figure 10 This is a schematic diagram of a data processing device provided in an embodiment of this application. The data processing device 2 may include: a sample feature determination module 21, a sample feature generation module 22, a trigger probability estimation module 23, a conversion probability estimation module 24, and a parameter adjustment module 25;

[0300] The sample feature determination module 21 is used to acquire sample business data associated with the sample object and determine the sample shared attribute features of the sample object for the sample business data through the initial network model.

[0301] The sample feature generation module 22 is used to generate initial sample service trigger features and initial sample service conversion features for sample objects based on sample shared attribute features in the initial network model.

[0302] The trigger probability prediction module 23 is used to perform full-connection processing on the initial sample business trigger features to obtain the target sample business trigger features, and generate a sample trigger probability prediction value of the sample object for the sample business data based on the target sample business trigger features.

[0303] The conversion probability prediction module 24 is used to perform feature sharing and transmission processing on the target sample business trigger features and the initial sample business conversion features to obtain the target sample business conversion features, and generate a sample conversion probability prediction value of the sample object for the sample business data based on the target sample business conversion features.

[0304] The parameter adjustment module 25 is used to adjust the parameters of the initial network model based on the sample label information, sample trigger probability prediction and sample conversion probability prediction of the sample object for the sample business data, and use the parameter-adjusted initial network model as the target network model; the target network model is used to predict the trigger probability prediction and conversion probability prediction of the target object for the business data.

[0305] The sample tag information includes business trigger tag information and business conversion tag information;

[0306] The parameter adjustment module 25 includes: a loss determination unit 251 and a parameter adjustment unit 252;

[0307] The loss determination unit 251 is used to determine the trigger loss value of the initial network model based on the business trigger tag information of the sample object for the sample business data and the sample trigger probability prediction value.

[0308] The loss determination unit 251 is used to determine the conversion loss value of the initial network model based on the business conversion label information of the sample object for the sample business data and the sample conversion probability prediction value.

[0309] The loss determination unit 251 is used to determine the model loss value of the initial network model based on the trigger loss value and the transformation loss value.

[0310] The parameter adjustment unit 252 is used to adjust the parameters of the initial network model according to the model loss value. When the parameter-adjusted initial network model meets the model convergence condition, the parameter-adjusted initial network model is used as the target network model.

[0311] The specific implementation methods of the loss determination unit 251 and the parameter adjustment unit 252 can be found in the above description. Figure 8 The description of step S305 in the corresponding embodiment will not be repeated here.

[0312] The specific implementation methods of the sample feature determination module 21, sample feature generation module 22, trigger probability prediction module 23, conversion probability prediction module 24, and parameter adjustment module 25 can be found in the above description. Figure 8 The descriptions of steps S301-S305 in the corresponding embodiments will not be repeated here. Furthermore, the beneficial effects of using the same method will also not be repeated.

[0313] Further, please see Figure 11 , Figure 11 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Figure 11 As shown, the computer device 1000 may include a processor 1001, a network interface 1004, and a memory 1005. Furthermore, the computer device 1000 may also include a user interface 1003 and at least one communication bus 1002. The communication bus 1002 is used to enable communication between these components. In some embodiments, the user interface 1003 may include a display screen and a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. Optionally, the network interface 1004 may include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be high-speed RAM or non-volatile memory, such as at least one disk storage device. Optionally, the memory 1005 may also be at least one storage device located remotely from the processor 1001. Figure 11 As shown, the memory 1005, which is a computer-readable storage medium, may include an operating system, a network communication module, a user interface module, and a device control application.

[0314] In such Figure 11 In the computer device 1000 shown, the network interface 1004 provides network communication functions; the user interface 1003 is mainly used to provide an input interface for users; and the processor 1001 can be used to call the device control application stored in the memory 1005.

[0315] It should be understood that the computer device 1000 described in the embodiments of this application can execute the foregoing text. Figure 3 , Figure 4 , Figure 5 and Figure 8 The description of the data processing method in the corresponding embodiments can also be performed as described above. Figure 9 In the corresponding embodiments, data processing device 1 and Figure 10 The description of the data processing device 2 in the corresponding embodiments will not be repeated here. Furthermore, the beneficial effects of using the same method will also not be repeated.

[0316] Furthermore, it should be noted that this application embodiment also provides a computer-readable storage medium, which stores a computer program executed by the aforementioned data processing apparatus 1 and data processing apparatus 2. The computer program includes program instructions, and when the processor executes the program instructions, it can execute the aforementioned... Figure 3 , Figure 4 , Figure 5 and Figure 8 The description of the data processing method in the corresponding embodiments is already provided and will not be repeated here. Furthermore, the beneficial effects of using the same method will also not be repeated. For technical details not disclosed in the computer-readable storage medium embodiments related to this application, please refer to the description of the method embodiments of this application.

[0317] Furthermore, it should be noted that this application also provides a computer program product or computer program, which may include computer instructions, which may be stored in a computer-readable storage medium. The processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor may execute the computer instructions, causing the computer device to perform the aforementioned actions. Figure 3 , Figure 4 , Figure 5 and Figure 8 The description of the data processing method in the corresponding embodiments is already provided and will not be repeated here. Furthermore, the beneficial effects of using the same method will also not be repeated. For technical details not disclosed in the computer program products or computer program embodiments related to this application, please refer to the description of the method embodiments of this application.

[0318] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0319] The above-disclosed embodiments are merely preferred embodiments of this application and should not be construed as limiting the scope of this application. Therefore, any equivalent variations made in accordance with the claims of this application shall still fall within the scope of this application.

Claims

1. A data processing method, characterized in that, include: Obtain business data associated with the target object, and determine the shared attribute characteristics of the target object in relation to the business data; Based on the shared attribute characteristics, the initial business trigger characteristics and initial business conversion characteristics of the target object for the business data are generated; The initial business triggering features are processed by a full connection to obtain the target business triggering features. Based on the target business triggering features, a trigger probability estimate of the target object for the business data is generated. The target service triggering feature and the initial service conversion feature are subjected to feature sharing and transmission processing to obtain the target service conversion feature. The feature sharing and transmission processing includes: concatenating the target service triggering feature and the initial service conversion feature to obtain a shared concatenated feature. The shared splicing features are processed by a full connection to obtain the target business conversion features; Based on the target business conversion characteristics, generate a conversion probability estimate for the target object in relation to the business data.

2. The method according to claim 1, characterized in that, The step of acquiring business data associated with the target object and determining the shared attribute characteristics of the target object for the business data includes: Obtain a set of business data associated with the target object, and retrieve business data from the set of business data; The object attributes of the target object, the business attributes of the business data, and the context attributes associated with the target object are input into the input network layer of the target network model; The object attributes are embedded by the input network layer to obtain the object features corresponding to the object attributes; The business attributes are embedded by the input network layer to obtain the business features corresponding to the business attributes; The context attributes are embedded using the input network layer to obtain the context features corresponding to the context attributes; The object features, business features, and context features are concatenated to obtain the shared attribute features of the target object for the business data.

3. The method according to claim 2, characterized in that, The object attributes include object identifier, basic attributes, and behavioral attributes; The step of embedding features of the object attributes through the input network layer to obtain object features corresponding to the object attributes includes: In the input network layer, a hash operation is performed on the object identifier to obtain the hash object identifier corresponding to the object identifier; Obtain the object identifier lookup table associated with the object identifier, and based on the hash object identifier, search for the object identifier feature corresponding to the object identifier in the object identifier lookup table; Obtain the basic attribute lookup table associated with the basic attribute, and search for the basic attribute feature corresponding to the basic attribute in the basic attribute lookup table; Obtain the behavior attribute lookup table associated with the behavior attribute, and search for the behavior attribute feature corresponding to the behavior attribute in the behavior attribute lookup table; The object identification feature, the basic attribute feature, and the behavioral attribute feature are fused to obtain the object feature corresponding to the object attribute.

4. The method according to claim 2, characterized in that, The target network model further includes a parameter-sharing network layer; the parameter-sharing network layer includes a shared fully connected layer, and a first triggering fully connected layer and a transformation fully connected layer respectively connected to the shared fully connected layer; The step of generating the initial business triggering features and initial business conversion features of the target object for the business data based on the shared attribute features includes: The shared attribute features are input into the shared fully connected layer in the parameter sharing network layer, and the shared attribute features are processed by the shared fully connected layer to obtain fully connected shared features; The fully connected shared feature is input to the first triggered fully connected layer, and the fully connected shared feature is processed by the first triggered fully connected layer to generate the initial business trigger feature of the target object for the business data; The fully connected shared features are input into the transformation fully connected layer, and the transformation fully connected layer performs full-connection processing on the fully connected shared features to generate the initial business transformation features of the target object for the business data.

5. The method according to claim 4, characterized in that, The transformation fully connected layer includes a first shallow transformation fully connected layer and a first deep transformation fully connected layer; the initial business transformation features include initial shallow transformation features and initial deep transformation features; The step of inputting the fully connected shared features into the transformation fully connected layer, and performing full-connection processing on the fully connected shared features through the transformation fully connected layer to generate the initial business transformation features of the target object for the business data includes: The fully connected shared features are input into the first shallow transformation fully connected layer, and the fully connected shared features are processed by the first shallow transformation fully connected layer to generate the initial shallow transformation features of the target object for the business data. The fully connected shared features are input into the first deep transformation fully connected layer, and the fully connected shared features are processed by the first deep transformation fully connected layer to generate the initial deep transformation features of the target object for the business data.

6. The method according to claim 2, characterized in that, The target network model further includes a multi-layer perceptron layer; the multi-layer perceptron layer includes a second triggered fully connected layer; The step of performing full-connection processing on the initial service triggering features to obtain the target service triggering features includes: The initial service trigger feature is input into the second trigger fully connected layer, and the initial service trigger feature is processed by the second trigger fully connected layer to obtain the target service trigger feature.

7. The method according to claim 6, characterized in that, The initial business conversion characteristics include initial shallow conversion characteristics and initial deep conversion characteristics; the target business conversion characteristics include target shallow conversion characteristics and target deep conversion characteristics; the conversion probability estimate includes shallow conversion probability estimate and deep conversion probability estimate; The step of performing feature sharing and transmission processing on the target business triggering features and the initial business conversion features to obtain target business conversion features, and generating a conversion probability estimate of the target object for the business data based on the target business conversion features, includes: The target service trigger feature and the initial shallow conversion feature are concatenated to obtain a shared shallow concatenated feature. The shared shallow concatenated feature is then subjected to a fully connected processing to obtain the target shallow conversion feature. Based on the target shallow conversion characteristics, generate a shallow conversion probability estimate of the target object for the business data; The target shallow transformation feature and the initial deep transformation feature are concatenated to obtain a shared deep concatenated feature. The shared deep concatenated feature is then fully connected to obtain the target deep transformation feature. Based on the target deep conversion characteristics, generate the target object's deep conversion probability estimate for the business data.

8. The method according to claim 7, characterized in that, The multilayer sensing network layer also includes a second shallow transformation fully connected layer and a shallow connection layer; The step of concatenating the target service trigger feature and the initial shallow conversion feature to obtain a shared shallow concatenated feature, and then performing a fully connected operation on the shared shallow concatenated feature to obtain the target shallow conversion feature, includes: The target service trigger feature and the initial shallow conversion feature are input into the shallow connection layer. The shallow connection layer performs feature concatenation on the target service trigger feature and the initial shallow conversion feature to obtain a shared shallow concatenation feature. The shared shallow splicing features are input into the second shallow transformation fully connected layer, and the shared shallow splicing features are fully connected through the second shallow transformation fully connected layer to obtain the target shallow transformation features.

9. The method according to claim 7, characterized in that, The multilayer sensing network layer also includes a second deep transformation fully connected layer and a deep connection layer; The step of concatenating the target shallow transformation feature and the initial deep transformation feature to obtain a shared deep concatenated feature, and then performing a fully connected operation on the shared deep concatenated feature to obtain the target deep transformation feature, includes: The target shallow transformation feature and the initial deep transformation feature are input into the deep connection layer. The deep connection layer performs feature concatenation on the target shallow transformation feature and the initial deep transformation feature to obtain a shared deep concatenation feature. The shared deep splicing features are input into the second deep transformation fully connected layer, and the shared deep splicing features are fully connected through the second deep transformation fully connected layer to obtain the target deep transformation features.

10. The method according to claim 1, characterized in that, The method further includes: The estimated trigger probability, the estimated conversion probability, and the estimated resources corresponding to the business data are processed to obtain the business data quality of the target object for the business data. Based on the quality of the business data, the business data is subjected to a first sorting process to obtain the initial sorting information corresponding to the business data; The business data is subjected to a second sorting process using the sorting strategy and the initial sorting information to obtain the target sorting information corresponding to the business data.

11. A data processing method, characterized in that, include: Obtain sample business data associated with the sample object, and determine the sample-shared attribute features of the sample object in relation to the sample business data through an initial network model; In the initial network model, based on the sample shared attribute characteristics, initial sample service triggering characteristics and initial sample service conversion characteristics of the sample object for the sample service data are generated; The initial sample service triggering features are processed by a full connection to obtain the target sample service triggering features. Based on the target sample service triggering features, a sample triggering probability estimate of the sample object for the sample service data is generated. The target sample service triggering feature and the initial sample service conversion feature are subjected to feature sharing and transfer processing to obtain the target sample service conversion feature. The feature sharing and transfer processing includes: concatenating the target sample service triggering feature and the initial sample service conversion feature to obtain the sample shared concatenated feature; and performing full connection processing on the sample shared concatenated feature to obtain the target sample service conversion feature. Based on the target sample business conversion characteristics, generate a sample conversion probability estimate for the sample object relative to the sample business data; Based on the sample label information of the sample object for the sample business data, the sample trigger probability prediction, and the sample conversion probability prediction, the parameters of the initial network model are adjusted, and the parameter-adjusted initial network model is used as the target network model; the target network model is used to predict the target object's trigger probability prediction and conversion probability prediction for the business data.

12. The method according to claim 11, characterized in that, The sample tag information includes business trigger tag information and business conversion tag information; The step of adjusting the parameters of the initial network model based on the sample label information of the sample object for the sample business data, the sample trigger probability prediction, and the sample conversion probability prediction, and using the parameter-adjusted initial network model as the target network model, includes: Based on the service triggering label information of the sample object for the sample service data and the sample triggering probability prediction value, the triggering loss value of the initial network model is determined; Based on the business conversion tag information of the sample object for the sample business data and the sample conversion probability prediction value, the conversion loss value of the initial network model is determined; The model loss value of the initial network model is determined based on the trigger loss value and the conversion loss value. The parameters of the initial network model are adjusted based on the model loss value. When the initial network model with adjusted parameters meets the model convergence condition, the initial network model with adjusted parameters is used as the target network model.

13. A computer device, characterized in that, include: Processor and memory; The processor is connected to the memory, wherein the memory is used to store a computer program, and the processor is used to invoke the computer program to cause the computer device to perform the method according to any one of claims 1-12.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded and executed by a processor to cause a computer device having the processor to perform the method of any one of claims 1-12.

15. A computer program product, characterized in that, The computer program product includes computer instructions stored in a computer-readable storage medium and adapted to be read and executed by a processor to cause a computer device having the processor to perform the method of any one of claims 1-12.

Images