Transformation event processing method and device, storage medium and electronic equipment

Abstract

The invention discloses a conversion event processing method and device, a storage medium and electronic equipment. The method comprises the steps of obtaining target object data generated by a first account in a target application, generating first sample object data and second sample object data through a sample account, training an initial prediction model by the first sample object data and the second sample object data together to obtain a target prediction model, and inputting the target object data into a pre-trained target prediction model, determining a target account tag corresponding to the first account, and indicating that the first account triggers the target conversion event by the target account tag. The technical problem that in a part of media information publishing scenes, due to the fact that data synchronization of a media information publisher and a media information publishing platform is slow, the media information is difficult to publish accurately, and the publishing efficiency of the media information is low is solved.

Description

Technical Field

[0001] The present application relates to the field of computers, and in particular, to a method and device for processing a conversion event, a storage medium, and an electronic device. Background Art

[0002] At present, in the relevant technology, in the process of publishing media information on the application, the media information publisher usually transmits the custom conversion event back to the media information publishing platform to optimize the publishing effect of the media information and realize the accurate delivery of the media information. The feedback time of the custom conversion event needs to be as early as possible so that the media information publishing platform can adjust the exposure method of the relevant media information in time. In the relevant technology, some account types require the long-term object data of the account to be accumulated to a certain extent before accurate prediction can be achieved, which leads to low prediction efficiency of the account type. Due to the slow data synchronization between the media information publisher and the media information publishing platform, the adaptation accuracy of the media information published by the media information publishing platform and the corresponding account is poor, resulting in a technical problem of low media information publishing efficiency.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The embodiments of the present application provide a method and device for processing a conversion event, a storage medium, and an electronic device, so as to at least solve the technical problem that in some media information publishing scenarios, due to slow data synchronization between the media information publisher and the media information publishing platform, it is difficult to accurately publish the media information, resulting in low efficiency in publishing the media information.

[0005] According to one aspect of an embodiment of the present application, a method for processing a conversion event is provided, comprising: obtaining target object data generated by a first account in a target application, wherein the target object data is used to represent object data generated by the first account within a first historical duration; inputting the target object data into a pre-trained target prediction model, and determining a target account label corresponding to the first account, wherein the target account label is used to represent an account type predicted for the first account after a target future duration, the time length of the first historical duration is less than the time length of the target future duration, the target prediction model is a model obtained by training an initial prediction model using a sample account, the sample account generates first sample object data within the first historical duration, and generates second sample object data within a second historical duration, the time length of the second historical duration is the same as the time length of the target future duration, and the first sample object data and the second sample object data are used together to train the initial prediction model; when the target account label indicates that the account type of the first account is a preset account type, determining that the first account has triggered a target conversion event, wherein the triggering of the target conversion event indicates that the first account is a converted account of the target application.

[0006] According to another aspect of an embodiment of the present application, a conversion event processing device is further provided, comprising: an acquisition module, configured to acquire target object data generated by a first account in a target application, wherein the target object data is used to represent object data generated by the first account within a first historical duration; a prediction module, configured to input the target object data into a pre-trained target prediction model, and determine a target account label corresponding to the first account, wherein the target account label is used to represent an account type predicted for the first account after a target future duration, the time length of the first historical duration is less than the time length of the target future duration, the target prediction model is a model obtained by training an initial prediction model using a sample account, the sample account generates first sample object data within the first historical duration, and generates second sample object data within a second historical duration, the time length of the second historical duration is the same as the time length of the target future duration, and the first sample object data and the second sample object data are used together to train the initial prediction model; a determination module, configured to determine that the first account has triggered a target conversion event when the target account label indicates that the account type of the first account is a preset account type, wherein the triggering of the target conversion event indicates that the first account is a converted account of the target application.

[0007] Optionally, the device is also used to: input the target object data into a pre-trained target prediction model, and before determining the target account label corresponding to the first account, obtain the first sample object data, wherein the first sample object data is used to represent the object data generated by the sample account in the first historical period in the target application; input the first sample object data into the initial prediction model to obtain a prediction label, wherein the prediction label is used to represent the account type predicted for the sample account after the second historical period; obtain the second sample object data generated by the sample account within the second historical period, wherein the second sample object data is used to represent the object data generated by the sample account in the second historical period in the target application; determine a sample label based on the second sample object data, wherein the sample label is used to represent the actual account type of the sample account after the second historical period; and train the initial prediction model based on the sample label and the prediction label.

[0008] Optionally, the device is used to determine a sample label based on the second sample object data in the following manner, including at least one of the following: when the value of a first parameter in the second sample object data satisfies a first preset value condition, labeling a first sample label for the sample account, wherein the first parameter represents the social ability of the sample object, and the sample label includes the first sample label; when the value of a second parameter in the second sample object data satisfies a second preset value condition, labeling a second sample label for the sample account, wherein the second parameter represents the business ability of the sample object, and the sample label includes the second sample label; when the value of a third parameter in the second sample object data satisfies a third preset value condition, labeling a third sample label for the sample account, wherein the third parameter represents the consumption ability of the sample object, and the sample label includes the third sample label.

[0009] Optionally, the device is used to label the sample account with a first sample label when the value of the first parameter in the second sample object data satisfies a first preset value condition in the following manner: obtaining a first sub-parameter set from the second sample object data, wherein the first sub-parameter set includes at least one of the following: the number of accounts invited by the sample account, the length of time that the sample account and the friend account use the target application together, and the number of friend accounts of the sample account; performing a weighted operation on at least two sub-parameters in the first sub-parameter set to determine a first weighted result, determining the first weighted result as the first parameter, and labeling the sample account with a first sample label when the value of the first weighted result satisfies the first preset value condition.

[0010] Optionally, the device is used to perform a weighted operation on at least two sub-parameters in the first sub-parameter set to determine a first weighted result in the following manner: performing regression modeling based on the second sample object data to determine a weight coefficient for each sub-parameter in the first sub-parameter set, wherein the target of the regression modeling is the sum of the consumption capabilities or the sum of the activity capabilities of the sample account and the friend account; and using the weight coefficient to perform a weighted operation on at least two sub-parameters in the first sub-parameter set to determine the first weighted result.

[0011] Optionally, the device is used to label the sample account with a second sample label when the value of the second parameter in the second sample object data satisfies a second preset value condition in the following manner: obtaining a second sub-parameter set from the second sample object data, wherein the second sub-parameter set includes at least one of the following: the number of times the sample account is the account with the highest business score in a business of the target application, the number of accounts defeated by the sample account in the target application, and the number of tasks completed by the sample account in the target application; performing a weighted operation on at least two sub-parameters in the second sub-parameter set to determine a second weighted result, determining the second weighted result as the second parameter, and labeling the sample account with a second sample label when the value of the second weighted result satisfies the second preset value condition.

[0012] Optionally, the device is also used to: when the sample label corresponding to the first sample account is determined, determine the second sample account according to the second sample object data corresponding to the first sample account, wherein the similarity between the second sample object data corresponding to the first sample account and the second sample object data corresponding to the second sample account satisfies a preset similarity condition; and label the second sample account with the same sample label as the first sample account.

[0013] Optionally, the device is used to input the target object data into a pre-trained target prediction model to determine the target account label corresponding to the first account in the following manner: perform a feature extraction operation on the target object data to obtain initial feature data; select target feature data from the initial feature data based on the first historical duration and the target future duration; input the target feature data into the target prediction model to obtain a target classification label or a target evaluation label, wherein the target classification label indicates whether the first account is the preset account type, and the target evaluation label is used to evaluate the probability of whether the first account will perform a preset behavior after the target future duration. When the target evaluation label meets the preset evaluation parameter conditions, the account type of the first account is an account type that will perform the preset behavior.

[0014] Optionally, the device is also used for: when the target account tag indicates that the account type of the first account is a preset account type, after determining that the first account has triggered a target conversion event, sending the account data of the first account to a media information publishing platform to determine a second account, wherein the second account is a similar account found by the media information publishing platform based on the account data of the first account; determining target media information based on the preset account type; and publishing the target media information to the first account and the second account through the media information publishing platform.

[0015] According to another aspect of the embodiments of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned method for processing conversion events when running.

[0016] According to another aspect of the embodiment of the present application, a computer program product or a computer program is provided, the computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the above conversion event processing method.

[0017] According to another aspect of the embodiments of the present application, there is also provided an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the above-mentioned method for processing conversion events through the computer program.

[0018] In an embodiment of the present application, target object data generated by a first account in a target application is obtained, wherein the target object data is used to represent object data generated by the first account within a first historical duration; the target object data is input into a pre-trained target prediction model to determine a target account label corresponding to the first account, wherein the target account label is used to represent the account type predicted for the first account after a target future duration, the time length of the first historical duration is less than the time length of the target future duration, the target prediction model is a model obtained by training an initial prediction model using a sample account, the sample account generates first sample object data within the first historical duration, and generates second sample object data within the second historical duration, the time length of the second historical duration is the same as the time length of the target future duration, and the first sample object data and the second sample object data are used together to train the initial prediction model training; when the target account label indicates that the account type of the first account is a preset account type, determining that the first account has triggered a target conversion event, wherein the triggering of the target conversion event indicates that the first account is a converted account of the target application, by obtaining the target object data generated by the relevant account in the target application, inputting the target object data into the pre-trained target prediction model, determining the target account label corresponding to the relevant account, and the account type of the relevant account indicated by the target account label, when the account type is the preset account type, that is, the relevant account has triggered the target conversion event, thereby achieving the purpose of predicting the long-term type of the relevant account, thereby solving the technical problem of low efficiency in media information publishing due to slow data synchronization between media information publishers and media information publishing platforms in some media information publishing scenarios.

[0019] On the other hand, the acquired target object data is input into the pre-trained target prediction model, and the pre-trained target prediction model is used to determine the target account label corresponding to the relevant account, thereby improving the efficiency of the target application in analyzing the object data of the relevant account. According to different account types, more personalized and accurate services can be provided in a targeted manner.

[0020] In addition, different target account tags are used to divide the account types of related accounts. When the account type is the preset account type, the related account is considered to have triggered a target conversion event, which increases the types of target conversion events in the target application and further improves the efficiency of media information release. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0022] Figure 1is a schematic diagram of an application environment of an optional method for processing a conversion event according to an embodiment of the present application;

[0023] Figure 2 is a flow chart of an optional method for processing a conversion event according to an embodiment of the present application;

[0024] Figure 3 is a schematic diagram of optional object data according to an embodiment of the present application;

[0025] Figure 4 is a schematic diagram of an optional process of object behavior prediction according to an embodiment of the present application;

[0026] Figure 5 is a schematic diagram of an optional process of predicting an object label according to an embodiment of the present application;

[0027] Figure 6 is a schematic diagram of a system architecture of an optional conversion event according to an embodiment of the present application;

[0028] Figure 7 is a schematic diagram of an optional method for predicting account type according to an embodiment of the present application;

[0029] Figure 8 is a schematic diagram of an optional method for determining a scoring coefficient according to an embodiment of the present application;

[0030] Fig. 9 is a schematic diagram of different optional account types according to an embodiment of the present application;

[0031] Fig.10 is a schematic structural diagram of an optional conversion event processing device according to an embodiment of the present application;

[0032] Fig.11 is a schematic structural diagram of an optional conversion event processing product according to an embodiment of the present application;

[0033] Fig.12 It is a schematic diagram of the structure of an optional electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0034] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0035] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0036] First, some nouns or terms that appear in the description of the embodiments of the present application are subject to the following interpretation:

[0037] Standard conversion actions: Generally directly connected to the software development kit of mobile standalone applications or website applications. Events are based on object data related to a single account, for example, registration completed, added to shopping cart, purchase, etc.

[0038] Shallow conversion events: shallow conversions are short-term behaviors after an account enters the application, usually occurring within 24 hours.

[0039] Deep conversion events: Deep conversions are behaviors that occur over a long period of time after an account enters the application, usually after 2 days.

[0040] Custom conversion events: Specific behaviors or goals defined by the publisher in the media information release activities, such as account clicks on ads, account registration, product purchases, etc., are transmitted to the media information release platform through code or API interfaces. The publisher can set different custom conversion events according to its own business needs and advertising goals, and associate the custom conversion events with the media information release activities released by the media information release platform. By returning the custom conversion event data, the publisher can understand the effect and conversion rate of the media information release activities, optimize the media information delivery strategy, and improve the media information return on investment.

[0041] The present application is described below in conjunction with embodiments:

[0042] According to one aspect of an embodiment of the present application, a method for processing a conversion event is provided. Optionally, in this embodiment, the method for processing a conversion event can be applied to: Figure 1 In the hardware environment composed of the server 101 and the terminal device 103 shown in FIG. Figure 1 As shown, the server 101 is connected to the terminal device 103 via a network, and can be used to provide services for the terminal device or an application installed on the terminal device. The application can be a video application, an instant messaging application, a browser application, an educational application, a game application, etc. A database 105 may be set up on the server or independently of the server to provide data storage services for the server 101, for example, a game data storage server. The above-mentioned network may include, but is not limited to, a wired network and a wireless network, wherein the wired network includes, a local area network, a metropolitan area network and a wide area network; the wireless network includes, Bluetooth, WIFI and other networks that implement wireless communication; the terminal device 103 may be a terminal configured with an application, and may include, but is not limited to, at least one of the following: a mobile phone (such as an Android phone, an iOS phone, etc.), a laptop computer, a tablet computer, a PDA, a MID (Mobile Internet Devices), a PAD, a desktop computer, a smart TV, an intelligent voice interaction device, a smart home appliance, a vehicle-mounted terminal, an aircraft, a virtual reality (VR) terminal, an augmented reality (AR) terminal, a mixed reality (MR) terminal and other computer devices; the above-mentioned server may be a single server, a server cluster consisting of multiple servers, or a cloud server.

[0043] Optionally, in this embodiment, the above-mentioned method for processing the conversion event can also be implemented by a server, for example, Figure 1 It is implemented in the server 101 shown; or it is implemented by the terminal device and the server together.

[0044] The above is only an example and is not specifically limited in this embodiment.

[0045] Optionally, as an optional implementation, as Figure 2 As shown, the processing method of the above conversion event includes:

[0046] S202, obtaining target object data generated by the first account in the target application, wherein the target object data is used to represent object data generated by the first account within a first historical period;

[0047] Optionally, in this embodiment, the above-mentioned first account may include but is not limited to a personal account, a business account, an organizational account, a media account, a brand account, etc., and the above-mentioned target application may include but is not limited to various types of applications such as social media applications, shopping applications, game applications, music applications, video applications, etc.

[0048] Exemplarily, the above-mentioned target application can be used for account login on the following devices, including but not limited to mobile phones, computers, intelligent voice interaction devices, smart home appliances, vehicle-mounted terminals, aircraft, etc., and can be applied to various scenarios, including but not limited to cloud technology, artificial intelligence, smart transportation, assisted driving, etc. The above-mentioned target application can be applied to intelligent transportation systems or intelligent vehicle-road collaborative systems.

[0049] Intelligent Traffic System (ITS), also known as Intelligent Transportation System, effectively and comprehensively applies advanced science and technology (information technology, computer technology, data communication technology, sensor technology, electronic control technology, automatic control theory, operations research, artificial intelligence, etc.) to transportation, service control and vehicle manufacturing, strengthens the connection between vehicles, roads and users, and thus forms a comprehensive transportation system that ensures safety, improves efficiency, improves the environment and saves energy.

[0050] Intelligent Vehicle Infrastructure Cooperative Systems (IVICS), referred to as IVICS, is a development direction of Intelligent Transportation Systems (ITS). IVICS uses advanced wireless communications and new generation Internet technologies to implement all-round dynamic real-time information interaction between vehicles and roads, and conducts active vehicle safety control and road cooperative management based on the collection and integration of dynamic traffic information in all time and space, fully realizing the effective coordination of people, vehicles and roads, ensuring traffic safety, and improving traffic efficiency, thus forming a safe, efficient and environmentally friendly road traffic system.

[0051] Exemplarily, the target object data corresponding to the first account is used to represent the object data generated by the first account in the target application within the first historical period, wherein the first historical period can be flexibly set, and this application does not limit the specific time range. The object data may include but is not limited to transaction type object data obtained after authorization by the account, such as payment, etc., and social type object data obtained after authorization by the account, such as completing a team with friends, etc.

[0052] It should be noted that the above target object data may include but is not limited to account name, account industry information, account avatar, account order history, etc. Figure 3 is a schematic diagram of an optional object data according to an embodiment of the present application, such as Figure 3As shown, the target object data 302 includes a part of the target object data on the media information publishing platform side and a small part of the lost target object data on the object data side generated by the target application. Specifically, the account sends the account payment message to the media information publishing platform side, which can be achieved through the following steps, including but not limited to:

[0053] S1, the media information publishing platform publishes relevant media information of the target application 304;

[0054] S2, the account (corresponding to the aforementioned first account) enters the download interface 306 of the target application, including but not limited to clicking a button, voice wake-up, touch screen sliding, fingerprint verification, etc.;

[0055] S3, the account downloads and installs the target application 308;

[0056] S4, after the account enters the target application, a new account is registered 310;

[0057] S5, 7 days later, the account logs into the target application again 312, and a payment behavior 314 occurs.

[0058] In an exemplary embodiment, Figure 3 As shown, on December 12, the first account made a payment in the target application, and the target object data obtained by the media information publisher includes the historical records of the payment made by the first account on December 12.

[0059] S204, inputting the target object data into a pre-trained target prediction model, and determining a target account label corresponding to the first account, wherein the target account label is used to indicate the account type predicted for the first account after the target future duration, the time length of the first historical duration is less than the time length of the target future duration, the target prediction model is a model obtained by training an initial prediction model using a sample account, the sample account generates first sample object data within the first historical duration, and generates second sample object data within the second historical duration, the time length of the second historical duration is the same as the time length of the target future duration, and the first sample object data and the second sample object data are used together to train the initial prediction model;

[0060] Optionally, in this embodiment, the target prediction model may include but is not limited to regression models, classification models, time series models, clustering models, reinforcement learning models, neural network models, combination models, etc., and the target account labels may include but are not limited to social relationship labels, technical labels, geographic location labels, demographic labels, personality characteristic labels, device labels, consumption level labels, etc., among which social relationship labels may include but are not limited to social expert labels, technical labels may include but are not limited to technical expert labels, etc.

[0061] It should be noted that the above-mentioned regression models are usually used to predict models of continuous target variables, such as linear regression, polynomial regression, support vector regression, etc.; the above-mentioned classification models are usually used to predict models of discrete target variables, such as logistic regression, decision tree, random forest, naive Bayes, support vector machine, etc.; the above-mentioned time series models are usually used to predict models of time series data, such as ARIMA model, exponential smoothing method, neural network, etc.; the above-mentioned clustering models are usually used to perform cluster analysis on data, unsupervised learning models, such as K-means clustering, hierarchical clustering, etc.; the above-mentioned reinforcement learning models are usually used to learn optimal strategies through interaction with the environment, such as deep reinforcement learning; the above-mentioned neural network models usually use artificial neural networks to simulate the connection and information transmission between neurons in the human brain, such as deep neural networks, convolutional neural networks, recurrent neural networks, etc.; the above-mentioned combined model refers to a model that makes predictions by combining multiple basic models.

[0062] Exemplarily, the first sample object data may be used to represent the object data generated by the sample account in the target application within the first historical period, and the object data may include but is not limited to making payments, sending orders, adding to shopping carts, adding friends, etc. The first sample object data may include but is not limited to the account name of the sample account, object data, account industry information, account avatar, account order history, etc.

[0063] In addition, the above-mentioned second sample object data can be used to represent the object data generated by the above-mentioned sample account in the above-mentioned target application within the second historical period. The above-mentioned object data may include but is not limited to making payments, sending orders, adding to shopping carts, adding friends, etc. The above-mentioned second sample object data may include but is not limited to account name, object data, account industry information, account avatar, account order history, etc.

[0064] Furthermore, by obtaining the target object data of the first account in the short term and inputting the target object data into the pre-trained target prediction model, the long-term target account label of the first account can be predicted. In other words, different target object data of different accounts in the short term can be used to predict different account types corresponding to different accounts in the long term. Finally, accurate release of media information can be achieved based on different account types.

[0065] For example, by obtaining the game data of account A using the target game application on the first day and inputting this part of the game data into the above-mentioned target prediction model, the account type of account A on the 21st day can be predicted, and then accurate media information can be recommended after the first day based on the account type to improve the click-through rate or conversion rate of the media information.

[0066] Optionally, in this embodiment, the time length between the above target future duration and the above second historical duration is greater than the above first historical duration, and the initial prediction model is trained through sample accounts to obtain a target prediction model, which predicts the account type of the above first account according to the above target account label. It should be noted that the above sample accounts may include but are not limited to personal accounts, corporate accounts, organizational accounts, media accounts, brand accounts, etc.

[0067] In an exemplary embodiment, Figure 4 is a schematic diagram of an optional object behavior prediction process according to an embodiment of the present application, such as Figure 4 As shown:

[0068] S402, after the account browses a promotional video released by a role-playing game application on the media information publishing platform, the account clicks a game application download button on the interface of the media information publishing platform to jump to the game application download interface for downloading and installing;

[0069] S404, entering the role-playing game application to register an account;

[0070] S406, during the use of the role-playing game application, the account obtains virtual prop resources through the payment function, that is, payment object data is generated;

[0071] S408, if the account does not log into the game within one month after registering the role-playing game account, it is determined that the account is not interested in the game;

[0072] S410, after the account is registered in the role-playing game application, if the account actively adds other accounts as friends 25 times, the account is identified as a social expert type (social object data);

[0073] S412, the role-playing game application feeds back the above-mentioned payment object data and social object data to the media information publishing platform proposed in S402, and the media information publishing platform adjusts the delivery strategy of the game promotion media information delivery;

[0074] S414, in the above S404, when an account enters the role-playing game for the first time to register an account, a preliminary account type prediction will be performed through the target prediction model, for example, predicting that a certain account is the social expert type proposed in S410.

[0075] It should be noted that Figure 5 is a schematic diagram of an optional process of predicting an object label according to an embodiment of the present application. The above S414 predicts the account type through the target prediction model and determines the target account label corresponding to the account, such as Figure 5 As shown, this can be achieved through the following steps, including but not limited to:

[0076] S1, the account is registered in the target game application 502, and prediction is performed based on the object data generated on the first day in the target game application;

[0077] S2, predict the account type 7 days later based on the object data of the relevant account on the first day;

[0078] S3, the behavior of the relevant account on the first day includes 25 behaviors of actively adding other target game application accounts as friends, that is, the preliminary prediction is that after 7 days, the account will be a social expert type 504;

[0079] S4, the actions of the relevant account on the first day also included 30 virtual battles, 27 of which were victorious, which means that the account is initially predicted to be a technical expert type 506 after 7 days.

[0080] Furthermore, a supervised machine learning (ML) model can be used to obtain target account labels for the above account types. This application does not limit the algorithm used by the supervised model. That is, logistic regression, XGBoost, and deep neural network can be used to divide the account labels for the above account types, including but not limited to binary classification of 0 or 1 for label prediction, and scoring prediction can also be performed through the above object data.

[0081] Exemplarily, the initial prediction model is trained using the first sample object data and the second sample object data sample accounts to obtain the target prediction model, wherein the first sample object data is generated within a first historical time period, and the second sample object data is generated within a second historical time period. The present application does not limit the specific value of the second historical time period, that is, the second time period can be 24 hours or 1 month. In addition, the length of the second historical time period is the same as the length of the target future time period. It can be understood that the second sample object data used for initial prediction model training is generated within a historical time period that is the same as the length of the target future time period.

[0082] In an exemplary embodiment, Figure 6 is a schematic diagram of a system architecture of an optional conversion event according to an embodiment of the present application, such as Figure 6 As shown:

[0083] S602, obtaining object data of the first account and performing basic data format conversion processing;

[0084] S604, manually or automatically extracting features from the object data of the first account;

[0085] S606-1, shallow event mining is performed on the extracted features, that is, based on the strategy rules, a custom conversion event is selected. For example, an account that has performed three virtual battles in the game is a conversion event, and the first account can be considered to be a master-type game account;

[0086] S606-2, generating a long-term account type based on the extracted features, for example, taking 7 days as a cycle, recording the object data generated by the first account within 7 days to determine the long-term type of the account and generate a long-term type label;

[0087] S606-3, predicting the long-term behavior events or types of the first account based on the extracted features, and using the long-term type label generated in S606-2 as a prediction target to predict the long-term behavior events or types of the first account;

[0088] S608, scoring the newly registered account using the target prediction model to generate a high-value conversion event;

[0089] S610, the account-related data is transmitted back to the media information publishing platform, and the media information publishing platform API interface can be directly connected;

[0090] S612, before the target prediction model proposed in S606-3 is launched, an offline evaluation is performed, and the accounts in the target application are regularly checked, the information is synchronized to the media information publishing platform proposed in S610, and the effect of the media information published on the media information publishing platform is evaluated.

[0091] Optionally, in this embodiment, Figure 7 is a schematic diagram of an optional method for predicting account type according to an embodiment of the present application, such as Figure 7 As shown, this can be achieved through the following steps, including but not limited to:

[0092] S1, the account enters the target program 702, and the account performs actions in the target program within 24 hours, showing social behavior characteristics;

[0093] S2, using the active time of the relevant account for 7 to 14 days plus the active time of the account's corresponding friends as the model training target, and using the account type 706 determined after 7 days as the prediction target of the 24-hour scoring model in S1 above, training the scoring model, and obtaining the account type 706 and the account label 710;

[0094] S3 uses supervised learning to learn the account label 710 using the 24-hour features proposed in S1, and then predicts the long-term behavioral events or types 712 of the account.

[0095] It should be noted that the scoring method of the scoring model in S2 above can be understood as follows: for example, when an account invites at least one friend to join the game within 7 days after entering a certain game application, and the game time with the friend reaches 4 hours, the social score of the relevant account is determined as a weighted calculation of the number of new accounts invited by the relevant account and the game time with the friend, wherein the weight of the number of new accounts invited by the relevant account is set to 0.3, and the weight of the game time with the friend is set to 0.7, resulting in a social score of 3.1 for the relevant account. The preset social type account identification condition is that the social score of the relevant account is not less than 3 points, that is, the above-mentioned relevant account is identified as a social expert type account;

[0096] Specifically, if there are multiple features to define a social type account, for example, a social score is defined for each relevant account, using three features: the number of invited friends, the length of time spent playing games with friend accounts, and the number of friend accounts. That is, the social score can be determined by summing the product of parameter a and the number of invited friends, parameter b and the length of time spent playing games with friends, and parameter c and the number of friends, where parameter a is the weight of the number of friends invited by the account, parameter b is the weight of the length of time spent playing games with friends, and parameter c is the weight of the number of friends of the account. Parameters a, b, and c can be determined in the following ways, including but not limited to:

[0097] S1, determined by the gaming experience of the account of one of the above-mentioned gaming applications;

[0098] S2, use unsupervised learning to cluster accounts and generate account types; the classic r clustering algorithm k-means can be used. Figure 8 is a schematic diagram of an optional method for determining a scoring coefficient according to an embodiment of the present application, such as Figure 8 As shown, the cluster center of cluster 801 is (1,1) which is a social account type, and the types of related accounts represented by ellipse 802 can all be labeled as social account types;

[0099] S3, use supervised learning of linear regression or logistic regression to train the target prediction model, for example, to make the sum of the activity levels of the above-mentioned related accounts and friend accounts the highest, or the sum of their consumption capacities the highest. For example, the active duration of the related accounts from 7 to 14 days can be used to make a regression model of the 7-day social behavior, and the coefficient of each feature generated can be used as the definition formula of the social score of the related account.

[0100] In addition, when an account logs in to a certain game application and is rated as the best related account for game performance 5 times and completes 10 game tasks within 7 days, the technical score of the related account is determined by weighting the number of times the related account is rated as the best related account for game performance and the total number of game tasks completed by the related account. Among them, the weight of the number of times the related account is rated as the best related account for game performance is set to 0.2, and the weight of the total number of game tasks completed by the related account is set to 0.8. The technical score of the related account is 9. The preset technical type account identification condition is that the technical score of the related account is not less than 5 points, that is, the above-mentioned related account is determined to be a technical expert type account.

[0101] S206, when the preset account type indicates that the account type of the first account is the preset account type, determining that the first account has triggered a target conversion event, wherein the triggering of the target conversion event indicates that the first account is a converted account of the target application.

[0102] Optionally, in this embodiment, the above-mentioned preset account types may include but are not limited to social account types, high gaming ability types, etc. The above-mentioned converted account can be understood as the above-mentioned first account downloading the above-mentioned target application, and the above-mentioned first account generated object data in the target application within the above-mentioned first historical period, that is, the account type of the above-mentioned first account after the above-mentioned target future period is predicted through the object data.

[0103] Exemplarily, if the account type prediction result of the above-mentioned first account is a preset account type, the target conversion event corresponding to the preset account type is triggered. In other words, the above-mentioned target account label is determined by the above-mentioned target object data generated by the above-mentioned first account, and then the target account label indicates the account type of the above-mentioned first account. If the above-mentioned target account label has a corresponding relationship with the above-mentioned preset account type, the above-mentioned target conversion event can be triggered.

[0104] It should be noted that the above-mentioned target conversion event may include but is not limited to adaptively publishing the media information of the above-mentioned first account in the above-mentioned target application. For example, the media information publishing platform queries other related accounts with the same account type as the above-mentioned first account, and then obtains an account set of this account type, and publishes the above-mentioned target application-related media information for the account set.

[0105] In an exemplary embodiment, Fig. 9 is a schematic diagram of different optional account types according to an embodiment of the present application, such as Fig. 9As shown, the above-mentioned first account is determined as social account 901 after a period of time, that is, the preset account type is social account 901, which triggers the above-mentioned target conversion event. The target conversion event may include but is not limited to increasing the intensity of publishing media information about the social business display of the above-mentioned target application to accounts in the account set determined as social account 901, so as to further improve the media information delivery effect of the target account.

[0106] In an exemplary embodiment, the technical solution proposed in this application can be applied in game application scenarios, such as Figure 6 As shown:

[0107] S1, obtain the object data of the account that logs into the game application, and perform basic data format conversion processing to obtain the account object data statistics table, which may include but is not limited to the behavior type, the number of behaviors that occur within a certain period of time, etc., as shown in Table 1:

[0108]

[0109] Table 1

[0110] S2, manually or automatically extracting object data features based on the original data of the first account in S1;

[0111] S3-1, the first account completes 5 single-player tasks in the game and communicates with other accounts twice, which is a target conversion event. It can be considered that the first account is a social expert type account;

[0112] S3-2, with 7 days as a cycle, if the first account has 30 combat behaviors within 7 days, the long-term type of the account is determined, and the label of the first account is determined as a skill master type label;

[0113] S3-3, the first object data feature extracted in S2 is used to predict the long-term behavior event or type of the first account, and the long-term type label generated in S3-2 is used as the prediction target to predict the long-term behavior event or type of the first account;

[0114] S4, using the target prediction model to score newly registered accounts of the game application, and generate high-value conversion events, for example, promoting new game features to high-skilled accounts on relevant media information release platforms, or providing detailed explanations of game features to low-skilled accounts;

[0115] S5, transmit the account-related data back to the media information publishing platform, and directly connect to the media information publishing platform API interface;

[0116] S6, before the target prediction model proposed in S3-3 is launched, offline evaluation is performed, and accounts in the target application are regularly monitored, information is synchronized to the media information publishing platform proposed in S5, and the effect of the media information published on the media information publishing platform is evaluated.

[0117] In another exemplary embodiment, the technical solution proposed in this application can also be applied in online car-hailing application scenarios, such as Figure 6 As shown:

[0118] S1, obtain the original click stream data of the first account that logs into the online car-hailing application, and perform basic data format conversion processing to obtain a behavior record statistics table, which may include but is not limited to the behavior type, the number of behaviors that occur within a certain period of time, etc., as shown in Table 2:

[0119]

[0120] Table 2

[0121] S2, manually or automatically extracting object data features based on the original data of the first account in S1;

[0122] S3-1, the first account initiated 4 vehicle reservation orders in the online car-hailing application and made 3 order payments, which is the target conversion event. It can be considered that the first account is a high-frequency car-using account;

[0123] S3-2, taking 7 days as a cycle, if the first account has made 30 car reservations within 7 days, the long-term type of the account is determined, and the label of the first account is determined as a high-frequency car type label;

[0124] S3-3, the first object data feature extracted in S2 is used to predict the long-term behavior event or type of the first account, and the long-term type label generated in S3-2 can be used as a prediction target to predict the long-term behavior event or type of the first account;

[0125] S4, using the target prediction model to score newly registered accounts of the online car-hailing application, and generate high-value conversion events, for example, promoting the new function of car reservation to accounts with high frequency of car use on relevant media information release platforms, or providing detailed explanations of car reservation to accounts with low frequency of car use;

[0126] S5, transmits the account-related data back to the media information publishing platform, and can directly connect to the media information publishing platform API interface;

[0127] S6, before the target prediction model proposed in S3-3 is launched, offline evaluation is performed, and accounts in the target application are regularly monitored, information is synchronized to the media information publishing platform proposed in S5, and the effect of the media information published on the media information publishing platform is evaluated.

[0128] In an embodiment of the present application, target object data generated by a first account in a target application is obtained, wherein the target object data is used to represent object data generated by the first account within a first historical duration; the target object data is input into a pre-trained target prediction model to determine a target account label corresponding to the first account, wherein the target account label is used to represent the account type predicted for the first account after a target future duration, the time length of the first historical duration is less than the time length of the target future duration, the target prediction model is a model obtained by training an initial prediction model using a sample account, the sample account generates first sample object data within the first historical duration, and generates second sample object data within the second historical duration, the time length of the second historical duration is the same as the time length of the target future duration, and the first sample object data and the second sample object data are used together for the initial prediction Model training; when the target account label indicates that the account type of the first account is a preset account type, it is determined that the first account has triggered a target conversion event, wherein the triggering of the target conversion event indicates that the first account is a converted account of the target application, by obtaining the target object data generated by the relevant account in the target application, inputting the target object data into the pre-trained target prediction model, and determining the target account label corresponding to the relevant account, wherein the account type of the relevant account is indicated by the target account label. When the account type is the preset account type, the account has triggered the target conversion event, thereby achieving the purpose of long-term type prediction of the account, thereby solving the technical problem of low efficiency in media information publishing due to slow data synchronization between media information publishers and media information publishing platforms in some media information publishing scenarios, making it difficult to accurately publish media information.

[0129] As an optional solution, before inputting the target object data into the pre-trained target prediction model and determining the target account label corresponding to the first account, the method further includes: obtaining the first sample object data, wherein the first sample object data is used to represent the object data generated by the sample account in the target application within the first historical period; inputting the first sample object data into the initial prediction model to obtain a prediction label, wherein the prediction label is used to represent the account type predicted for the sample account after the second historical period; obtaining the second sample object data generated by the sample account within the second historical period, wherein the second sample object data is used to represent the object data generated by the sample account in the target application within the second historical period; determining a sample label based on the second sample object data, wherein the sample label is used to represent the actual account type of the sample account after the second historical period; and training the initial prediction model based on the sample label and the prediction label until the training of the initial prediction model is completed.

[0130] Optionally, in this embodiment, within the first historical period, the object data generated by the sample account in the target application may include but is not limited to making payments, sending orders, adding to shopping carts, adding friends, changing avatars, initiating session requests to other accounts in the target application, etc.

[0131] It should be noted that after the above-mentioned sample account has passed the above-mentioned second historical period, the predicted account type is determined by the above-mentioned prediction label. It can be understood that after the above-mentioned first sample data is input into the above-mentioned initial prediction model, a prediction label is generated, and the prediction label can be used to indicate the account type predicted for the above-mentioned sample account after the above-mentioned second period.

[0132] Exemplarily, within the second historical period, the second sample object data generated by the sample account is obtained, and the sample label is generated from the second sample object data. That is to say, the sample label represents the real account type of the sample account after the second historical period. Finally, the training of the initial prediction model is completed according to the sample label and the prediction label to obtain the target prediction model.

[0133] Through this embodiment, the first sample object data is obtained, the first sample object data is input into the initial prediction model to obtain a prediction label, the second sample object data generated by the sample account within the second historical period is obtained, the sample label is determined according to the second sample object data, and finally the initial prediction model is trained according to the sample label and the prediction label to complete the training of the initial prediction model, which can improve the accuracy of the target prediction model, reduce the prediction deviation, and improve the generalization ability of the target prediction model, thereby achieving a better prediction effect.

[0134] As an optional scheme, the above-mentioned determination of the sample label based on the above-mentioned second sample object data includes at least one of the following: when the value of the first parameter in the above-mentioned second sample object data satisfies the first preset value condition, the above-mentioned sample account is labeled with a first sample label, wherein the above-mentioned first parameter represents the social ability of the above-mentioned sample object, and the above-mentioned sample label includes the above-mentioned first sample label; when the value of the second parameter in the above-mentioned second sample object data satisfies the second preset value condition, the above-mentioned sample account is labeled with a second sample label, wherein the above-mentioned second parameter represents the business ability of the above-mentioned sample object, and the above-mentioned sample label includes the above-mentioned second sample label; when the value of the third parameter in the above-mentioned second sample object data satisfies the third preset value condition, the above-mentioned sample account is labeled with a third sample label, wherein the above-mentioned third parameter represents the consumption ability of the above-mentioned sample object, and the above-mentioned sample label includes the above-mentioned third sample label.

[0135] Optionally, in this embodiment, the specific values ​​of the first parameter, the second parameter and the third parameter are not limited. The first parameter can be used to indicate the social ability of the sample account to socialize with other accounts in the target application, for example, the number of times the sample account invites friends to team up for games within a certain period, the number of times the sample account shares game recordings with friends within a certain period, etc. The second parameter can be used to indicate the business ability of the sample account to participate in part or all of the business of the target application. Different business capabilities can be set according to different types of target applications. Taking game applications as an example, it can include but is not limited to the number of hostile accounts defeated by the sample account in a certain period of time, the number of times the sample account wins within a certain period, etc. The third parameter can be used to indicate the consumption ability of the sample account to use virtual resource transactions in the target application. Different business capabilities can be set according to different types of target applications. Taking game applications as an example, it can include but is not limited to the virtual resource expenditure of the sample account to purchase virtual game props within a certain period, the income of the sample account from selling virtual resources within a certain period, etc.

[0136] It should be noted that the first parameter is determined according to the first preset value condition. For example, the target application is a social application. The first parameter is determined by the activity level of the relevant account and the association degree of the relevant account friends. That is, the social ability of the sample object is obtained, and the first sample label is marked for the sample account. The first sample label may include but is not limited to a high social ability account label, a low social ability account label, etc.

[0137] Exemplarily, the second parameter is determined according to the second preset value condition. For example, the target application is a racing game application. The second parameter is determined by the ranking of relevant accounts participating in the racing competition. That is, the business capability of the sample object is obtained, and the sample account is marked with a second sample label. The second sample label may include but is not limited to a high gaming ability label, a low gaming ability label, etc.

[0138] In addition, the third parameter is determined according to the third preset value condition. For example, if the target application is an online shopping application, the third parameter is determined by the number of payments initiated by the relevant account, that is, the consumption capacity of the sample object is obtained, and the sample account is marked with a third sample label. The third sample label may include but is not limited to a high-consumption account label and a low-consumption account label.

[0139] It should be noted that the above-mentioned first sample label, the above-mentioned second sample label, and the above-mentioned third sample label can all be marked on the above-mentioned same sample account, and this application does not limit the maximum number of sample labels that can be marked on the sample account.

[0140] By adopting a method in which the first sample label is labeled for the sample account when the value of the first parameter in the second sample object data satisfies the first preset value condition, the second sample label is labeled for the sample account when the value of the second parameter in the second sample object data satisfies the second preset value condition, and the third sample label is labeled for the sample account when the value of the third parameter in the second sample object data satisfies the third preset value condition, the ability to understand and analyze the sample account is improved, and a more accurate and effective basis is provided for subsequent data processing and application.

[0141] As an optional scheme, when the value of the first parameter in the above-mentioned second sample object data satisfies the first preset value condition, the above-mentioned sample account is labeled with a first sample label, including: obtaining a first sub-parameter set from the above-mentioned second sample object data, wherein the above-mentioned first sub-parameter set includes at least one of the following: the number of accounts invited by the above-mentioned sample account, the length of time that the above-mentioned sample account and the friend account use the above-mentioned target application together, and the number of friend accounts of the above-mentioned sample account; performing a weighted operation on at least two sub-parameters in the above-mentioned first sub-parameter set to determine a first weighted result, determining the above-mentioned first weighted result as the above-mentioned first parameter, and when the value of the above-mentioned first weighted result satisfies the above-mentioned first preset value condition, labeling the above-mentioned sample account with a first sample label.

[0142] Optionally, in this embodiment, the above-mentioned weighted operation can set different weight values ​​for different first sub-parameters, so as to determine a more accurate first parameter according to the different degrees of influence of different first sub-parameters in the social field, and thus, can label the sample account with a first sample label that is more in line with the social field type.

[0143] It should be noted that the calculation of the above-mentioned first weighted result can be obtained by the first sub-parameter set determined in the above-mentioned second sample object data. It can be understood that the first weighted result is obtained by weighting different first sub-parameters, so that the above-mentioned first sub-parameter set can be used together to characterize the social capabilities of the above-mentioned sample account in the target application. For example, by weighting at least two specific numbers of the number of accounts invited by the above-mentioned sample account, the length of time that the sample account and the friend account jointly use the target application, and the number of friend accounts of the sample account, the social capabilities of the above-mentioned sample account can be more effectively characterized, so as to facilitate the prediction of the account type corresponding to the sample account after the second historical length of time, thereby improving the training efficiency of the initial prediction model.

[0144] For example, in a certain target application, an account enters the target application and invites 3 new accounts to enter the target application to complete registration within 1 day, that is, the number of accounts invited by the sample account is 3, and the account has a total of 100 friends in the target application. At the same time, the total online time of these 100 friends and the online time of the account reaches 400 hours, that is, the sample account and the friend account jointly use the target application for 400 hours, and the number of friend accounts of the sample account is 100. The first weighted result of the relevant account is obtained by the following method, which may include but is not limited to:

[0145] S1, performing a weighted operation on the number of accounts invited by the sample account and the duration of time the sample account and the friend account jointly use the target application;

[0146] S2, performing a weighted operation on the number of accounts invited by the sample account and the number of friend accounts of the sample account;

[0147] S3, performing a weighted operation on the number of friend accounts of the sample account and the duration of time that the sample account and the friend account jointly use the target application;

[0148] S4, performing a weighted operation on the number of accounts invited by the sample account, the number of friend accounts of the sample account, and the duration of time the sample account and the friend accounts jointly use the target application.

[0149] In an exemplary embodiment, the value of the first weighted result is obtained. If the value of the first weighted result satisfies the first preset value condition, the first sample label is marked for the sample account. That is to say, the value of the first weighted result needs to be compared with the first preset value. When the corresponding relationship is satisfied, the sample account will be marked with the first sample label.

[0150] For example, the first weighted result is obtained by adopting the above-mentioned S3 method. The number of friend accounts of the account in the above-mentioned target application is 100, and the weight is 0.9. At the same time, the online time of these 100 friends and the total online time of the account reach 400 hours, and the weight is 0.1. That is, the first weighted result is the weighted calculation of the number of friend accounts of the above-mentioned account 100 and the time 400 that the related account and the friend account use the said target application together, and the first weighted result is 130. The above-mentioned first preset value condition is that the value of the first weighted result is not less than 100, that is, the account is identified as a social expert type account, and then it is known that the above-mentioned related account meets the first preset value condition, that is, the related account is a social expert type account and is marked with a high social label.

[0151] Through this embodiment, a first sub-parameter set is obtained from the second sample object data, a weighted operation is performed on at least two sub-parameters in the first sub-parameter set, a first weighted result is determined, and when the value of the first weighted result meets the first preset value condition, a first sample label is labeled for the sample account. Since different weight values ​​are set for different sub-parameters, the first weighted result can better reflect the social ability of the sample account, and the account type of the sample account represented by the first sample label is more accurate, thereby achieving the technical effect of quickly labeling the sample account. Since the sample label is automatically labeled for the sample account, manual labeling is avoided, and the training efficiency of the initial prediction model is further improved.

[0152] As an optional scheme, the above-mentioned weighted operation is performed on at least two sub-parameters in the above-mentioned first sub-parameter set to determine the first weighted result, including: performing regression modeling based on the above-mentioned second sample object data to determine the weight coefficient of each sub-parameter in the above-mentioned first sub-parameter set, wherein the target of the above-mentioned regression modeling is the sum of the consumption capacity or the sum of the activity capacity of the above-mentioned sample account and the above-mentioned friend account; using the above-mentioned weight coefficient to perform a weighted operation on at least two sub-parameters in the above-mentioned first sub-parameter set to determine the above-mentioned first weighted result.

[0153] Optionally, in this embodiment, there is no limitation on the specific value of the above weight coefficient.

[0154] Exemplarily, the target of the regression modeling is determined by the sum of the consumption capabilities of the sample account and the friend account, or the target of the regression modeling is determined by the sum of the activity capabilities of the sample account and the friend account.

[0155] Furthermore, the above-mentioned regression modeling can be implemented by including but not limited to linear regression or logistic regression, and can be modeled using supervised learning. In other words, this application does not impose any limitation on the training algorithm of the regression model, and can include but not limited to logistic regression, XGBoost, deep neural network. The above-mentioned first sample label (0 or 1 binary classification) can be predicted, and regression can also be used to predict social ability.

[0156] Through this embodiment, by performing regression modeling on the second sample object data, determining the weight coefficient of each sub-parameter in the first sub-parameter set, and determining the first weighted result, the accuracy of calculating the first weighted result can be improved.

[0157] As an optional scheme, when the value of the second parameter in the above-mentioned second sample object data satisfies the second preset value condition, the above-mentioned sample account is labeled with a second sample label, including: obtaining a second sub-parameter set from the above-mentioned second sample object data, wherein the above-mentioned second sub-parameter set includes at least one of the following: the number of times the above-mentioned sample account is the account with the highest business score in a business of the above-mentioned target application, the number of accounts defeated by the above-mentioned sample account in the above-mentioned target application, and the number of tasks completed by the above-mentioned sample account in the above-mentioned target application; performing a weighted operation on at least two sub-parameters in the above-mentioned second sub-parameter set to determine a second weighted result, determining the above-mentioned second weighted result as the above-mentioned second parameter, and when the value of the above-mentioned second weighted result satisfies the above-mentioned second preset value condition, labeling the above-mentioned sample account with a second sample label.

[0158] Exemplarily, the second sub-parameter set is obtained through the second sample object data, and the second weighted result is obtained from the second sub-parameter set. The second sample label is marked for the sample account according to the second weighted result. When the target application is a single-player game application and the account logs in to the single-player game application, the second sub-parameter set determined by the relevant account may include but is not limited to at least one of the number of times the account is rated as the account with the highest combat score in the game season, the number of other accounts defeated by the account in the game, and the number of tasks completed by the account in the game.

[0159] It should be noted that the second weighted result of the relevant account may be obtained by the following method, including but not limited to:

[0160] S1, performing a weighted operation on the number of times the above-mentioned related account is rated as the related account with the highest combat score in the game season and the number of other related accounts defeated by the above-mentioned related account in the game;

[0161] S2, performing a weighted operation on the number of times the above-mentioned related account is rated as the related account with the highest combat score in the game season and the number of tasks completed by the above-mentioned related account in the game;

[0162] S3, performing a weighted operation on the number of tasks completed by the above-mentioned related account in the game and the number of other related accounts defeated by the above-mentioned related account in the game;

[0163] S4, performing a weighted operation on the number of tasks completed by the above-mentioned related account in the game and the number of other related accounts defeated by the above-mentioned related account in the game.

[0164] In an exemplary embodiment, the value of the second weighted result is obtained. If the value of the second weighted result satisfies the second preset value condition, the second sample label is marked for the sample account. That is to say, the value of the second weighted result needs to be compared with the second preset value. When the corresponding relationship is satisfied, the sample account will be marked with the second sample label.

[0165] For example, the second weighted result is obtained by adopting the above-mentioned S1 method. The number of times a certain related account is rated as the related account with the highest combat score in the game season is 10, and the weight is 0.4. At the same time, the total number of other related accounts defeated by the related account in the game is 40, and the weight is 0.6. That is, the second weighted result is the weighted operation of the number of times the above-mentioned related account is rated as the related account with the highest combat score in the game season (10) and the total number of other related accounts defeated by the related account in the game (40), and the first weighted result is 28. The above-mentioned second preset value condition is that the value of the second weighted result is not less than 20, that is, the account is identified as a skilled master type account. It is further known that the above-mentioned related account meets the second preset value condition, that is, the related account is a skilled master type account and is marked with a high skill label.

[0166] Through this embodiment, a second sub-parameter set is obtained from the above-mentioned second sample object data, a weighted operation is performed on at least two sub-parameters in the above-mentioned second sub-parameter set, a second weighted result is determined, and when the value of the above-mentioned second weighted result meets the above-mentioned second preset value condition, a second sample label is marked for the above-mentioned sample account. Since different weight values ​​are set for different sub-parameters, the second weighted result can better reflect the business capability of the sample account, and the account type of the sample account represented by the second sample label is more accurate, thereby achieving the technical effect of quickly marking the sample account. Since the sample label is automatically marked for the sample account, manual marking is avoided, and the training efficiency of the initial prediction model is further improved.

[0167] As an optional scheme, when the sample label corresponding to the first sample account is determined, the second sample account is determined according to the second sample object data corresponding to the first sample account, wherein the similarity between the second sample object data corresponding to the first sample account and the second sample object data corresponding to the second sample account satisfies a preset similarity condition; and the second sample account is marked with the same sample label as the first sample account.

[0168] Optionally, in this embodiment, the above-mentioned preset similarity condition can be understood as that the object behavior characteristics in the above-mentioned second sample object data have similarity, wherein the second sample account can include but is not limited to personal accounts, corporate accounts, organizational accounts, media accounts, brand accounts, etc.

[0169] Exemplarily, in a certain social application, only accounts that meet the object behavior characteristics of having more than 10 friends will be labeled as high-social ability accounts. The first sample account in the social application has more than 10 friends, and the second sample account in the social application also has more than 10 friends. That is, the object data corresponding to the first sample account and the object data corresponding to the second sample account are similar, and the first sample account and the second sample account are both identified as social expert account types.

[0170] In an exemplary embodiment, when the first sample account and the second sample account meet preset similarity conditions, they will be respectively marked with the same sample labels, such as a high-social ability account label, a high-consumption account label, etc., that is, the account type of the first sample account and the account type of the second sample account are the same account type.

[0171] The second sample account is determined by the second sample object data corresponding to the first sample account. When the similarity between the second sample object data corresponding to the first sample account and the second sample object data corresponding to the second sample account meets the preset similarity condition, the second sample account is labeled with the same sample label as the first sample account, thereby achieving the technical effect of quickly labeling sample labels and improving sample labeling efficiency.

[0172] As an optional solution, the above-mentioned inputting the above-mentioned target object data into the pre-trained target prediction model to determine the target account label corresponding to the above-mentioned first account includes: performing a feature extraction operation on the above-mentioned target object data to obtain initial feature data; selecting target feature data from the above-mentioned initial feature data based on the above-mentioned first historical duration and the above-mentioned target future duration; inputting the above-mentioned target feature data into the above-mentioned target prediction model to obtain a target classification label or a target evaluation label, wherein the above-mentioned target classification label indicates whether the above-mentioned first account is the above-mentioned preset account type, and the above-mentioned target evaluation label is used to evaluate the probability of whether the first account will perform the preset behavior after the target future duration. When the target evaluation label meets the preset evaluation parameter conditions, the account type of the first account is the account type that will perform the preset behavior. It should be noted that the above-mentioned target evaluation label may also include but is not limited to indicating the probability of whether the first account will perform the preset behavior after the target future duration.

[0173] Optionally, in this embodiment, the above-mentioned target classification labels may include but are not limited to high social ability account labels, low social ability account labels, high consumption account labels, low consumption account labels, high gaming ability user labels, low gaming ability user labels, etc., which are used to indicate the predicted account type of the first account after the target future duration.

[0174] It should be noted that the initial feature data is obtained by extracting the features of the target object data. It can be understood that the initial feature data includes behavioral features associated with the target object, such as high social type, low social type, high consumption type, low consumption type, high skilled type, low skilled type, etc.

[0175] Exemplarily, the target feature data is selected from the initial feature data according to the first historical duration and the target future duration. It can be understood that the target feature data having correlations based on different durations is extracted from the initial feature data. For example, if the account type after 21 days needs to be predicted, the initial feature data generated by the account on the first day can be used, or the feature data of the account from days 1 to 7 can be used. If the account type after 7 days needs to be predicted, the initial feature data generated by the account on the first day can be used, or the feature data of the account from days 1 to 3 can be used. That is, feature data of different first historical durations can be obtained according to the type of feature data, or feature data of different first historical durations can be obtained according to the length of the target future duration, so as to input the pre-trained target prediction model to obtain the target classification label or target evaluation label. That is, through the target prediction model, it can be determined whether the account type of the first account is the preset account type. For example, after the target feature data is predicted by the target prediction model, the target classification label of the first account is obtained as a high social ability account label, that is, the account type of the first account is determined to be a high social ability account.

[0176] Furthermore, the target evaluation label can be used to evaluate whether the first account will perform the preset behavior after the target future time. For example, after the target feature data is predicted by the target prediction model, the evaluation label for the first account to perform social behavior is 0.98, and the value of the preset evaluation parameter condition is 0.55, which means that the first account is predicted to perform the preset behavior after the target future time. In other words, as long as the evaluation label value of the first account is greater than the preset evaluation parameter value, it is determined that the first account will perform the preset behavior after the target future time, and the account type of the first account is a high social account type.

[0177] Through this embodiment, a feature extraction operation is performed on the above-mentioned target object data to obtain initial feature data; target feature data is selected from the above-mentioned initial feature data based on the above-mentioned first historical duration and the above-mentioned target future duration; the above-mentioned target feature data is input into the above-mentioned target prediction model to obtain a target classification label or a target evaluation label. The account type of the first account can be directly determined through the target classification label, which can realize direct prediction of the account type. The target evaluation label can predict whether the first account will produce a certain behavior in the future, which can realize indirect prediction of the account type. Therefore, by using the above-mentioned two different methods to judge the prediction result of the account type of the above-mentioned first account, the technical effect of improving the prediction accuracy of the account type and optimizing the prediction efficiency of the account type can be further achieved. In addition, different first historical durations can be used to obtain feature data corresponding to different durations, and different target future durations can be used to obtain feature data corresponding to different first historical durations. Different first historical durations or target future durations can produce different prediction results, so as to achieve the purpose of more accurate prediction of account types.

[0178] As an optional solution, when the target account label indicates that the account type of the first account is a preset account type, after determining that the first account has triggered the target conversion event, the method further includes: sending the account data of the first account to the media information publishing platform, determining a second account, wherein the second account is a similar account found by the media information publishing platform based on the account data of the first account; determining the target media information based on the preset account type; and publishing the target media information to the first account and the second account through the media information publishing platform.

[0179] Optionally, in this embodiment, the above-mentioned media information publishing platform may include but is not limited to a news media platform, a video sharing platform, a blog platform, an audio sharing platform, an online live broadcast platform, etc.

[0180] Exemplarily, after the account type of the first account has been determined as the preset account type, that is, the first account has triggered the target conversion event, it can be understood that the account type of the first account is a high-social account type. In a certain social application, the account data of the first account will be sent to the media information publishing platform to determine the target media information, wherein the target media information is not limited to the promotion of the newly released functions of the social application or the update of the original functions of the social application, and then determine the second account that is similar to the first account, that is, the second account can also be determined as a high-social account type, that is, the media information publishing platform accurately delivers the target media information to the first account and the second account.

[0181] By sending the account data of the first account to the media information publishing platform, the second account can be determined. The media information publishing platform finds similar accounts based on the account data of the first account, and publishes the target media information to the first account and the second account through the platform. The expected optimal media information publishing effect can be achieved.

[0182] The present application is further explained below in conjunction with specific embodiments:

[0183] This solution can be applied to, but is not limited to, game account growth-related scenarios to mine people who meet business needs. In the scenario of acquiring new accounts, based on the mined population, a custom conversion event is generated and sent back to the media party, who uses the similar lookalike method to help the application attract more high-value accounts.

[0184] Optionally, in the embodiment of the present application, a marketing delivery system developed by the team developers is responsible for the campaign management of media information delivery activities and crowd management. The crowd mined by the algorithm will be sent back to the media party through the above marketing delivery system to detect the effect of media information delivery.

[0185] The media information distributor responsible for the target application can acquire customers overseas through overseas channels in the following ways, but not limited to. The account is exposed through media information distribution, enters the application, and finally pays to complete the conversion. The entire link spans a long time, and the media information distribution platform and the media information distributor only have part of the account data on the link, and the data is not interoperable due to privacy protection. The media information distributor is downstream of the account journey link and has the object data of the account in the application. The media information distribution platform only has the exposure, click and install behavior of the account, and does not have the post-link data of the account. In order to be able to distribute to our target account more accurately, the media information distribution platform needs the media information distributor to provide its own custom conversion events conversion actions to more accurately optimize the events we need. Conversion events may need to find different account groups under different optimization goals.

[0186] S1, account funnel, account journey, such as Figure 3 As shown, the media information distributor and the media information distribution platform have different data.

[0187] S2. Long-term behavior events can be predicted based on the short-term behavior of the account in the following ways, but not limited to. After the account enters the game and registers, it can gradually differentiate into many different types.

[0188] like Figure 5As shown in the figure, after the account is registered, some accounts are lost, and the accounts that are not lost are newbie accounts, experienced accounts, or social accounts, which start playing games with friends. After a period of time, attributes such as defeater, social expert, and achiever may be displayed.

[0189] S3, account type examples may include but are not limited to Fig. 9 The account type shown.

[0190] S4, system architecture, e.g. Figure 6 As shown, this is the entire account conversion event mining feedback system:

[0191] S4-1, Raw Log Data: The original click stream data of the account, the most original data;

[0192] S4-2, Data ETL: Convert the original account clickstream data into a usable statistical feature format, such as 24-hour online time, number of logins, number of games, etc. There are two main uses: 1. Directly use rule strategies to generate conversion events; 2. Used in machine learning models;

[0193] Exemplarily, the original click stream data of an account in the target application is an action. For example, the account logs in at time t1, performs the first action, logs in at time t2, performs the second action, logs in again at time t3, and then logs in to the target application two more times. The converted data is a statistic for a period of time. It can be understood that on the first day, the account has a first action count of 1, a second action count of 2, and a login count of 5.

[0194] S4-3, Feature Engineering: Manual or automatic feature selection. Features of different time spans can be used for strategies or models in steps 4, 5, and 6 respectively;

[0195] It should be noted that the above time span can be understood as a target prediction model using the statistical features of the account 24 hours after registration to predict the type 7 days later. The data of the account 24 hours after registration can be used for feature extraction, for example, the number of logins in the 6th hour after registration, the number of logins in the 12th hour after registration, the number of logins in the 18th hour, the number of logins in the 24th hour, and other statistics of different lengths.

[0196] S4-4, Shallow Conversion Events Mining: Shallow event mining, selects custom conversion events based on strategy rules, such as an account that has played three games is a conversion event. S4-4 is listed in parallel with S4-5 and S4-6, and does not need to be used at the same time;

[0197] S4-5, Long-term User Type mining: Account long-term type mining, for example, whether it is a social account in the past 7 days, which can be used in the machine learning model in S4-6 as a label;

[0198] S4-6, Supervised ML Model: supervised learning machine learning, using the short-term behavior of an account to predict long-term behavior events or types;

[0199] S4-7, Offline Evaluation: Offline evaluation is indispensable before going online, for example, the actual retention rate of the feedback account;

[0200] S4-8, Prediction for deep conversion events: Use the model to score new accounts and generate high-value conversion events;

[0201] S4-9, Send to Ad Platforms: Use the server-to-server method to send the account back to the media information delivery platform and directly call the API interface of the media information delivery platform;

[0202] S4-10, Monitoring & Alerts: Detection and alarm, detect the number and quality of accounts sent back every day, and alarm if the sending fails;

[0203] S4-11, Result Evaluation: Evaluation of the effectiveness of the online media information release campaign, compared with other media information release campaigns or control group media information release campaigns.

[0204] It should be noted that the long-term account type prediction can be performed in the following ways, including but not limited to:

[0205] Optionally, in the embodiment of the present application, the long-term user type mining in S4-5 above defines an "account type" label, which is used as a predicted label. Secondly, S4-6 uses the label generated in S4-5 and the "feature" generated in S4-3 to predict the account and define the account type label, that is, label the account. The long term can be 7 days, 14 days, or longer. The account can be automatically classified directly according to business rules or unsupervised learning.

[0206] Exemplarily, a social account is defined as one who has invited a friend to join the game within 7 days of entering the application (registration or return), and a master account is one who has obtained the MVP (most valuable account) of the game 5 times within 7 days of entering the application. If there are multiple features to define sociality, such as defining a social score for each account, using three features: the number of invited friends, the length of time played with friends, and the number of friends. Then the social score is the sum of the product of a preset parameter a and the number of invited friends, the product of a preset parameter b and the length of time played with friends, and the product of a preset parameter c and the number of friends. Among them, the above-mentioned preset parameter a, b, and c coefficients can be determined by the following methods, including but not limited to: determined by business experience, or using unsupervised learning to cluster accounts and generate account types; the classic r clustering algorithm k-means can be used. Figure 8 As shown:

[0207] The cluster center of cluster 801 is (1,1), which is the target account type. Then all accounts in ellipse 802 can be labeled as target account type labels, i.e., account k-mean clustering; or they can be trained using supervised learning methods such as linear regression or logistic regression to maximize the long-term retention or payment of the account and friends (when training linear regression or logistic regression, use longer-term retention or payment as labels to learn the coefficients of the features in the linear formula). For example, if you need to determine the social score of an account in 7 days, you can use the account active time between the 7th and 14th days to perform regression modeling on the 7-day social behavior, and the coefficients of each feature generated can be used as the definition formula of the social score.

[0208] S5, after obtaining the "label" of the account's long-term behavior event, the account label can be predicted by short-term behavior (such as 24 hours) in the following ways, including but not limited to using supervised machine learning (ML) model. Here, the ML model can use any classic algorithm, such as logistic regression, XGBoost, deep neural network deep learning, which can predict account labels (0 or 1 binary classification) and can also perform regression to predict scores (such as "social points"). The essence of supervised learning is to map some features to labels. Predict long-term labels through short-term account behavior, and return high-scoring accounts as conversion events.

[0209] like Figure 5 As shown in the figure, all behaviors on the first day after account registration are used to predict whether the account will be a "social expert" or "technical expert" account 7 days later.

[0210] S6, account tag definition and type prediction, such as Figure 7 As shown in the figure, the account's 7-14 days of active time plus the friends' active time are used as the target, and the account label of the account on the 7th day is learned by supervised learning, that is, the "social score" scoring formula, and the account type of the account on the 7th day is defined (S4-5). Then, supervised learning is used to learn the label of the 7th day using the 24-hour features (S4-6). The difference between this two-step learning and directly using the 24-hour features to predict the 14-day target is that only social-related features are used on the 7th day, and the generated account will be biased towards social behavior, while direct prediction will lose this behavior.

[0211] It should be noted that account-type conversion events can further improve the media information delivery effect of target accounts based on traditional events, and attract accounts that are beneficial to the application ecosystem in the long run. For example, campaigns that deliver social behavior accounts have significantly improved the social behavior of accounts compared to media information delivery campaigns that optimize registration or retention accounts. Compared with directly using 24-hour social behavior strategies to generate accounts, the long-term type is more accurate using model prediction.

[0212] Furthermore, there are many ways to generate and predict account labels, which have been specifically described in S4. Account labels can be binary classifications of 0 or 1, or they can be continuous values ​​to score accounts. Predicted account labels can also be multi-label, and conversion event generation is not limited to newly registered accounts, but also applies to conversion events of lost and returning accounts.

[0213] Exemplarily, the technical solution proposed in this application can be applied to all application programs or website systems, not limited to mobile games, such as e-commerce, trading platforms and all other systems that require media information delivery. Social scores can be used as examples, but are not limited to them, where account tags are not limited to social. This application is universally applicable, that is, the media information distributor's return of custom conversion events is a universal practice.

[0214] It is understandable that in the specific implementation of this application, related data such as user information is 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 relevant data need to comply with relevant laws, regulations and standards of relevant countries and regions.

[0215] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0216] According to another aspect of the embodiment of the present application, a device for processing a conversion event for implementing the above-mentioned method for processing a conversion event is also provided. Fig.10 As shown, the device comprises:

[0217] An acquisition module 1002 is used to acquire target object data generated by a first account in a target application, wherein the target object data is used to represent object data generated by the first account within a first historical period;

[0218] Prediction module 1004, used for inputting the target object data into a pre-trained target prediction model, and determining a target account label corresponding to the first account, wherein the target account label is used to indicate the account type predicted for the first account after a target future duration, the time length of the first historical duration is less than the time length of the target future duration, the target prediction model is a model obtained by training an initial prediction model using a sample account, the sample account generates first sample object data within the first historical duration, and generates second sample object data within a second historical duration, the time length of the second historical duration is the same as the time length of the target future duration, and the first sample object data and the second sample object data are used together to train the initial prediction model;

[0219] The determination module 1006 is used to determine that the first account has triggered a target conversion event when the target account tag indicates that the account type of the first account is a preset account type, wherein the triggering of the target conversion event indicates that the first account is a converted account of the target application.

[0220] As an optional scheme, the device is also used to: input the target object data into a pre-trained target prediction model, and before determining the target account label corresponding to the first account, obtain the first sample object data, wherein the first sample object data is used to represent the object data generated by the sample account in the first historical period in the target application; input the first sample object data into the initial prediction model to obtain a prediction label, wherein the prediction label is used to represent the account type predicted for the sample account after the second historical period; obtain the second sample object data generated by the sample account within the second historical period, wherein the second sample object data is used to represent the object data generated by the sample account in the second historical period in the target application; determine a sample label based on the second sample object data, wherein the sample label is used to represent the actual account type of the sample account after the second historical period; and train the initial prediction model based on the sample label and the prediction label.

[0221] As an optional scheme, the device is used to determine the sample label according to the second sample object data in the following manner, including at least one of the following: when the value of the first parameter in the second sample object data satisfies the first preset value condition, labeling the sample account with a first sample label, wherein the first parameter represents the social ability of the sample object, and the sample label includes the first sample label; when the value of the second parameter in the second sample object data satisfies the second preset value condition, labeling the sample account with a second sample label, wherein the second parameter represents the business ability of the sample object, and the sample label includes the second sample label; when the value of the third parameter in the second sample object data satisfies the third preset value condition, labeling the sample account with a third sample label, wherein the third parameter represents the consumption ability of the sample object, and the sample label includes the third sample label.

[0222] As an optional scheme, the device is used to label the sample account with a first sample label when the value of the first parameter in the second sample object data satisfies a first preset value condition in the following manner: obtaining a first sub-parameter set from the second sample object data, wherein the first sub-parameter set includes at least one of the following: the number of accounts invited by the sample account, the length of time that the sample account and the friend account use the target application together, and the number of friend accounts of the sample account; performing a weighted operation on at least two sub-parameters in the first sub-parameter set to determine a first weighted result, determining the first weighted result as the first parameter, and labeling the sample account with a first sample label when the value of the first weighted result satisfies the first preset value condition.

[0223] As an optional scheme, the device is used to perform a weighted operation on at least two sub-parameters in the first sub-parameter set to determine a first weighted result in the following manner: performing regression modeling based on the second sample object data to determine the weight coefficient of each sub-parameter in the first sub-parameter set, wherein the target of the regression modeling is the sum of the consumption capabilities or the sum of the activity capabilities of the sample account and the friend account; and using the weight coefficient to perform a weighted operation on at least two sub-parameters in the first sub-parameter set to determine the first weighted result.

[0224] As an optional scheme, the device is used to label the sample account with a second sample label when the value of the second parameter in the second sample object data satisfies the second preset value condition in the following manner: obtaining a second sub-parameter set from the second sample object data, wherein the second sub-parameter set includes at least one of the following: the number of times the sample account is the account with the highest business score in a business of the target application, the number of accounts defeated by the sample account in the target application, and the number of tasks completed by the sample account in the target application; performing a weighted operation on at least two sub-parameters in the second sub-parameter set to determine a second weighted result, determining the second weighted result as the second parameter, and labeling the sample account with a second sample label when the value of the second weighted result satisfies the second preset value condition.

[0225] As an optional scheme, the device is also used to: when the sample label corresponding to the first sample account is determined, determine the second sample account according to the second sample object data corresponding to the first sample account, wherein the similarity between the second sample object data corresponding to the first sample account and the second sample object data corresponding to the second sample account satisfies a preset similarity condition; and mark the second sample account with the same sample label as the first sample account.

[0226] As an optional scheme, the device is used to input the target object data into a pre-trained target prediction model to determine the target account label corresponding to the first account in the following manner: perform a feature extraction operation on the target object data to obtain initial feature data; select target feature data from the initial feature data based on the first historical duration and the target future duration; input the target feature data into the target prediction model to obtain a target classification label or a target evaluation label, wherein the target classification label indicates whether the first account is the preset account type, and the target evaluation label is used to evaluate the probability of whether the first account will perform a preset behavior after the target future duration. When the target evaluation label meets the preset evaluation parameter conditions, the account type of the first account is an account type that will perform the preset behavior.

[0227] As an optional solution, the device is also used for: when the target account tag indicates that the account type of the first account is a preset account type, after determining that the first account has triggered a target conversion event, sending the account data of the first account to a media information publishing platform to determine a second account, wherein the second account is a similar account found by the media information publishing platform based on the account data of the first account; determining the target media information based on the preset account type; and publishing the target media information to the first account and the second account through the media information publishing platform.

[0228] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0229] According to one aspect of the present application, a computer program product is provided, the computer program product comprising a computer program / instruction, the computer program / instruction comprising a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication part 1109, and / or installed from a removable medium 1111. When the computer program is executed by the central processor 1101, various functions provided in the embodiments of the present application are executed.

[0230] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0231] Fig.11 The structure block diagram of a computer system for implementing an electronic device according to an embodiment of the present application is schematically shown.

[0232] It should be noted that Fig.11The computer system 1100 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0233] like Fig.11 As shown, the computer system 1100 includes a central processing unit 1101 (CPU), which can perform various appropriate actions and processes according to the program stored in the read-only memory 1102 (ROM) or the program loaded from the storage part 1108 to the random access memory 1103 (RAM). Various programs and data required for system operation are also stored in the random access memory 1103. The central processing unit 1101, the read-only memory 1102 and the random access memory 1103 are connected to each other through a bus 1104. An input / output interface 1105 (Input / Output interface, i.e., I / O interface) is also connected to the bus 1104.

[0234] The following components are connected to the input / output interface 1105: an input section 1106 including a keyboard, a mouse, etc.; an output section 1107 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1108 including a hard disk, etc.; and a communication section 1109 including a network interface card such as a LAN card, a modem, etc. The communication section 1109 performs communication processing via a network such as the Internet. A drive 1110 is also connected to the input / output interface 1105 as needed. A removable medium 1111, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1110 as needed so that a computer program read therefrom is installed into the storage section 1108 as needed.

[0235] In particular, according to an embodiment of the present application, the process described in each method flow chart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer readable medium, and the computer program contains a program code for executing the method shown in the flow chart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 1109, and / or installed from the removable medium 1111. When the computer program is executed by the central processor 1101, various functions defined in the system of the present application are executed.

[0236] According to another aspect of the embodiment of the present application, an electronic device for implementing the above-mentioned method for processing a conversion event is also provided. The electronic device may be Figure 1 The terminal device or server shown in the figure. This embodiment is described by taking the electronic device as a terminal device as an example. Fig.12 As shown, the electronic device includes a memory 1202 and a processor 1204. The memory 1202 stores a computer program, and the processor 1204 is configured to execute the steps in any of the above method embodiments through the computer program.

[0237] Optionally, in this embodiment, the electronic device may be located in at least one network device among a plurality of network devices of a computer network.

[0238] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0239] S1, obtaining target object data generated by a first account in a target application, wherein the target object data is used to represent object data generated by the first account within a first historical period;

[0240] S2, input the target object data into the pre-trained target prediction model, and determine the target account label corresponding to the first account, wherein the target account label is used to indicate the account type predicted for the first account after the target future duration, the time length of the first historical duration is less than the time length of the target future duration, the target prediction model is a model obtained by training the initial prediction model using a sample account, the sample account generates first sample object data within the first historical duration, and generates second sample object data within the second historical duration, the time length of the second historical duration is the same as the time length of the target future duration, and the first sample object data and the second sample object data are used together to train the initial prediction model;

[0241] S3, when the target account tag indicates that the account type of the first account is a preset account type, determining that the first account has triggered a target conversion event, wherein the triggering of the target conversion event indicates that the first account is a converted account of the target application.

[0242] Alternatively, a person skilled in the art may understand that: Fig.12 The structure shown is for illustration only, and the electronic device may also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a PDA, a mobile Internet device (MID), a PAD, and other terminal devices. Fig.12 The electronic device and the electronic equipment described above are not limited in structure. Fig.12More or fewer components (such as network interfaces, etc.) as shown in, or with Fig.12 Different configurations are shown.

[0243] Among them, the memory 1202 can be used to store software programs and modules, such as the program instructions / modules corresponding to the method and device for processing conversion events in the embodiments of the present application. The processor 1204 executes various functional applications and data processing by running the software programs and modules stored in the memory 1202, that is, to implement the above-mentioned method for processing conversion events. The memory 1202 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 1202 may further include a memory remotely located relative to the processor 1204, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. Among them, the memory 1202 may specifically be, but is not limited to, used to store information such as object data generated by the first account within the first historical period. As an example, such as Fig.12 As shown, the memory 1202 may include, but is not limited to, the acquisition module 1002, prediction module 1004, and determination module 1006 in the conversion event processing device. In addition, it may also include, but is not limited to, other module units in the conversion event processing device, which will not be repeated in this example.

[0244] Optionally, the transmission device 1206 is used to receive or send data via a network. Specific examples of the network may include a wired network and a wireless network. In one example, the transmission device 1206 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices and routers via a network cable so as to communicate with the Internet or a local area network. In one example, the transmission device 1206 is a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0245] In addition, the electronic device further includes: a display 1208 for displaying the target object data; and a connection bus 1210 for connecting various module components in the electronic device.

[0246] In other embodiments, the terminal device or server may be a node in a distributed system, wherein the distributed system may be a blockchain system, and the blockchain system may be a distributed system formed by connecting the multiple nodes through network communication. The nodes may form a peer-to-peer network, and any form of computing device, such as a server, terminal or other electronic device, may become a node in the blockchain system by joining the peer-to-peer network.

[0247] According to one aspect of the present application, a computer-readable storage medium is provided, and a processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the conversion event processing method provided in various optional implementations of the above-mentioned conversion event aspects.

[0248] Optionally, in this embodiment, the computer-readable storage medium may be configured to store a computer program for performing the following steps:

[0249] S1, obtaining target object data generated by a first account in a target application, wherein the target object data is used to represent object data generated by the first account within a first historical period;

[0250] S2, input the target object data into the pre-trained target prediction model, and determine the target account label corresponding to the first account, wherein the target account label is used to indicate the account type predicted for the first account after the target future duration, the time length of the first historical duration is less than the time length of the target future duration, the target prediction model is a model obtained by training the initial prediction model using a sample account, the sample account generates first sample object data within the first historical duration, and generates second sample object data within the second historical duration, the time length of the second historical duration is the same as the time length of the target future duration, and the first sample object data and the second sample object data are used together to train the initial prediction model;

[0251] S3, when the target account tag indicates that the account type of the first account is a preset account type, determining that the first account has triggered a target conversion event, wherein the triggering of the target conversion event indicates that the first account is a converted account of the target application.

[0252] Optionally, in this embodiment, a person of ordinary skill in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by instructing hardware related to the terminal device through a program, and the program may be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc.

[0253] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0254] If the integrated units in the above embodiments are implemented in the form of software functional units and sold or used as independent products, they can be stored in the above computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling one or more computer devices (which may be personal computers, servers, or network devices, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application.

[0255] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0256] In the several embodiments provided in the present application, it should be understood that the disclosed client can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0257] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0258] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0259] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for processing a conversion event, characterized in that: include: Acquire target object data generated by the first account in the target application, wherein the target object data is used to represent object data generated by the first account within a first historical period; Input the target object data into a pre-trained target prediction model to determine a target account label corresponding to the first account, wherein the target account label is used to indicate the account type predicted for the first account after a target future duration, the time length of the first historical duration is less than the time length of the target future duration, the target prediction model is a model obtained by training an initial prediction model using a sample account, the sample account generates first sample object data within the first historical duration, and generates second sample object data within the second historical duration, the time length of the second historical duration is the same as the time length of the target future duration, and the first sample object data and the second sample object data are used together to train the initial prediction model; When the target account tag indicates that the account type of the first account is a preset account type, it is determined that the first account has triggered a target conversion event, wherein the triggering of the target conversion event indicates that the first account is a converted account of the target application.

2. The method according to claim 1, characterized in that Before inputting the target object data into a pre-trained target prediction model to determine a target account label corresponding to the first account, the method further includes: Acquire the first sample object data, wherein the first sample object data is used to represent object data generated by the sample account within the first historical period in the target application; Inputting the first sample object data into the initial prediction model to obtain a prediction label, wherein the prediction label is used to indicate the account type predicted for the sample account after the second historical period; Acquire the second sample object data generated by the sample account within the second historical period, wherein the second sample object data is used to represent the object data generated by the sample account within the second historical period in the target application; Determine a sample label according to the second sample object data, wherein the sample label is used to represent the real account type of the sample account after the second historical period; The initial prediction model is trained according to the sample labels and the prediction labels.

3. The method according to claim 2, characterized in that The determining of the sample label according to the second sample object data includes at least one of the following: When the value of the first parameter in the second sample object data satisfies the first preset value condition, labeling the sample account with a first sample label, wherein the first parameter represents the social ability of the sample object, and the sample label includes the first sample label; When the value of the second parameter in the second sample object data satisfies a second preset value condition, marking the sample account with a second sample label, wherein the second parameter represents the business capability of the sample object, and the sample label includes the second sample label; When the value of the third parameter in the second sample object data satisfies the third preset value condition, a third sample label is marked for the sample account, wherein the third parameter represents the consumption capacity of the sample object, and the sample label includes the third sample label.

4. The method according to claim 3, characterized in that When the value of the first parameter in the second sample object data satisfies a first preset value condition, marking the sample account with a first sample label includes: Acquire a first sub-parameter set from the second sample object data, wherein the first sub-parameter set includes at least one of the following: the number of accounts invited by the sample account, the duration of the sample account and the friend account jointly using the target application, and the number of friend accounts of the sample account; Perform a weighted operation on at least two sub-parameters in the first sub-parameter set to determine a first weighted result, determine the first weighted result as the first parameter, and mark the sample account with a first sample label when the value of the first weighted result satisfies the first preset value condition.

5. The method according to claim 4, characterized in that The performing a weighted operation on at least two sub-parameters in the first sub-parameter set to determine a first weighted result includes: Performing regression modeling based on the second sample object data to determine the weight coefficient of each sub-parameter in the first sub-parameter set, wherein the target of the regression modeling is the sum of the consumption capacity or the sum of the activity capacity of the sample account and the friend account; A weighting operation is performed on at least two sub-parameters in the first sub-parameter set using the weight coefficient to determine the first weighted result.

6. The method according to claim 3, characterized in that When the value of the second parameter in the second sample object data satisfies a second preset value condition, marking the sample account with a second sample label includes: Acquire a second sub-parameter set from the second sample object data, wherein the second sub-parameter set includes at least one of the following: the number of times the sample account is the account with the highest business score in a business of the target application, the number of accounts defeated by the sample account in the target application, and the number of tasks completed by the sample account in the target application; Perform a weighted operation on at least two sub-parameters in the second sub-parameter set to determine a second weighted result, determine the second weighted result as the second parameter, and if the value of the second weighted result satisfies the second preset value condition, mark the sample account with a second sample label.

7. The method according to claim 2, characterized in that The method further comprises: When the sample label corresponding to the first sample account is determined, determining the second sample account according to the second sample object data corresponding to the first sample account, wherein the similarity between the second sample object data corresponding to the first sample account and the second sample object data corresponding to the second sample account satisfies a preset similarity condition; The second sample account is labeled with the same sample label as the first sample account.

8. The method according to claim 1, characterized in that The step of inputting the target object data into a pre-trained target prediction model to determine a target account label corresponding to the first account includes: Performing a feature extraction operation on the target object data to obtain initial feature data; Selecting target feature data from the initial feature data based on the first historical duration and the target future duration; The target feature data is input into the target prediction model to obtain a target classification label or a target evaluation label, wherein the target classification label indicates whether the first account is the preset account type, and the target evaluation label is used to evaluate the probability of whether the first account will perform a preset behavior after the target future duration. When the target evaluation label meets the preset evaluation parameter conditions, the account type of the first account is the account type that will perform the preset behavior.

9. The method according to claim 1, characterized in that: In the case where the target account tag indicates that the account type of the first account is a preset account type, after determining that the first account has triggered a target conversion event, the method further includes: Sending the account data of the first account to the media information publishing platform to determine a second account, wherein the second account is a similar account found by the media information publishing platform based on the account data of the first account; Determining target media information according to the preset account type; The target media information is published to the first account and the second account through the media information publishing platform.

10. A device for processing a conversion event, characterized in that: include: An acquisition module, configured to acquire target object data generated by a first account in a target application, wherein the target object data is used to represent object data generated by the first account within a first historical period; a prediction module, used for inputting the target object data into a pre-trained target prediction model, and determining a target account label corresponding to the first account, wherein the target account label is used to indicate the account type predicted for the first account after a target future duration, the time length of the first historical duration is less than the time length of the target future duration, the target prediction model is a model obtained by training an initial prediction model using a sample account, the sample account generates first sample object data within the first historical duration, and generates second sample object data within a second historical duration, the time length of the second historical duration is the same as the time length of the target future duration, and the first sample object data and the second sample object data are used together to train the initial prediction model; A determination module is used to determine that the first account has triggered a target conversion event when the target account tag indicates that the account type of the first account is a preset account type, wherein the triggering of the target conversion event indicates that the first account is a converted account of the target application.

11. The device according to claim 10, characterized in that Before the device inputs the target object data into a pre-trained target prediction model to determine the target account label corresponding to the first account, the device is further used to: Acquire the first sample object data, wherein the first sample object data is used to represent object data generated by the sample account within the first historical period in the target application; Inputting the first sample object data into the initial prediction model to obtain a prediction label, wherein the prediction label is used to indicate the account type predicted for the sample account after the second historical period; Acquire the second sample object data generated by the sample account within the second historical period, wherein the second sample object data is used to represent the object data generated by the sample account within the second historical period in the target application; Determine a sample label according to the second sample object data, wherein the sample label is used to represent the real account type of the sample account after the second historical period; The initial prediction model is trained according to the sample labels and the prediction labels.

12. The device according to claim 11, characterized in that The apparatus is configured to determine a sample label according to the second sample object data in the following manner, including at least one of the following: When the value of the first parameter in the second sample object data satisfies the first preset value condition, labeling the sample account with a first sample label, wherein the first parameter represents the social ability of the sample object, and the sample label includes the first sample label; When the value of the second parameter in the second sample object data satisfies a second preset value condition, marking the sample account with a second sample label, wherein the second parameter represents the business capability of the sample object, and the sample label includes the second sample label; When the value of the third parameter in the second sample object data satisfies the third preset value condition, a third sample label is marked for the sample account, wherein the third parameter represents the consumption capacity of the sample object, and the sample label includes the third sample label.

13. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program can be executed by a terminal device or a computer to execute the method described in any one of claims 1 to 9.

14. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method described in any one of claims 1 to 9 are implemented.

15. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 9 through the computer program.

Images