Information processing method, device and computer-readable storage medium

Abstract

The embodiments of the present application disclose an information processing method, apparatus, and computer-readable storage medium. The embodiments of the present application collect operational information of each dynamic push message; generate a first target beta distribution corresponding to the click-through rate of each dynamic push message based on exposure data and click data; select a preset number of dynamic push messages based on the first target beta distribution; obtain target predicted click-through rates of the preset number of dynamic push messages, and select target dynamic push messages for push based on the target predicted click-through rates. In this way, the operational information of each dynamic push message is collected in real time, a first target beta distribution corresponding to the click-through rate of each dynamic push message is generated based on the Thompson sampling concept, a preset number of dynamic push messages are selected based on the first target beta distribution and the corresponding target click-through rates are obtained, and accurate target dynamic push messages are selected for push based on the target predicted click-through rates, thereby greatly improving the accuracy of information processing.

Description

Technical Field

[0001] The present application relates to the field of information processing technology, and in particular to an information processing method, device, and computer-readable storage medium. Background Art

[0002] With the development of the Internet and the widespread use of computers, the online information push market has expanded rapidly. Push publishers can pay publishers (push platforms) to distribute their push information through web pages, search engines, browsers, or other online media, effectively promoting their products.

[0003] In the existing technology, in order to better achieve better creative selection, the push host can create multiple dynamic push information under one push information to form different creative ideas, and randomly push multiple dynamic push information online at the same time. According to the performance of each dynamic push information on the client, the best performing dynamic push information is selected to achieve creative optimization.

[0004] During the research and practice of the prior art, the inventors of this application discovered that in the prior art, due to the random push method, exposure opportunities would be wasted on low-quality dynamic push information, affecting the results of creative selection and resulting in low accuracy of information processing. Summary of the Invention

[0005] The embodiments of the present application provide an information processing method, device, and computer-readable storage medium, which can improve the accuracy of information processing.

[0006] To solve the above technical problems, the embodiments of the present application provide the following technical solutions:

[0007] An information processing method, comprising:

[0008] Collecting statistics on the operation information of each dynamic push message, wherein the operation information includes at least exposure data and click data;

[0009] generating a first target beta distribution corresponding to a click rate of each dynamic push information based on the exposure data and the click data;

[0010] Selecting a preset number of dynamic push information according to the first target beta distribution;

[0011] The target predicted click rates of the preset number of dynamic push information are obtained, and target dynamic push information is selected for push according to the target predicted click rates.

[0012] An information processing device, comprising:

[0013] A statistics unit, configured to collect statistics on operation information of each dynamic push message, wherein the operation information includes at least exposure data and click data;

[0014] A generating unit, configured to generate a first target beta distribution corresponding to a click rate of each dynamic push information based on the exposure data and the click data;

[0015] a first coarse sorting unit, configured to select a preset number of dynamic push information according to the first target beta distribution;

[0016] The fine sorting unit is used to obtain the target predicted click-through rates of the preset number of dynamic push information, and select the target dynamic push information for push according to the target predicted click-through rates.

[0017] In some embodiments, the operation information further includes virtual expense data, and the apparatus further includes:

[0018] a cost control unit, configured to calculate virtual cost data for each dynamically pushed information based on the virtual expense data and the conversion data;

[0019] The dynamic push information corresponding to the virtual cost data being greater than the preset virtual data is frozen.

[0020] In some embodiments, the fine sorting unit comprises:

[0021] A prediction subunit, configured to predict the preset number of dynamic push information using a target click rate prediction model to obtain target predicted click rates for the preset number of dynamic push information;

[0022] Exposure subunit, used to obtain target exposure data for each dynamic push information;

[0023] The combining subunit is used to select target dynamic push information for push based on the target predicted click rate and target exposure data.

[0024] In some embodiments, the binding subunit is used to:

[0025] When it is detected that the target exposure data is less than a second preset threshold, normalizing the target predicted click rate to obtain target prediction vector information of a preset number of dimensions;

[0026] Divide the probability intervals based on the target prediction vector information, randomly access the probability intervals, and determine the dynamic push information corresponding to the accessed probability intervals as target dynamic push information for push;

[0027] When it is detected that the target exposure data is not less than a second preset threshold, the dynamic push information with the highest target predicted click rate is determined as the target dynamic push information for push.

[0028] In some embodiments, the apparatus further comprises:

[0029] An updating unit, configured to collect update information of dynamic push information according to a preset period;

[0030] An update operation is performed on the dynamic push information according to the update information.

[0031] A computer-readable storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor to execute the steps in the above-mentioned information processing method.

[0032] The embodiment of the present application counts the operational information of each dynamic push message; generates a first target beta distribution corresponding to the click-through rate of each dynamic push message based on exposure data and click data; selects a preset number of dynamic push messages based on the first target beta distribution; obtains the target predicted click-through rates of the preset number of dynamic push messages, and selects target dynamic push messages for push based on the target predicted click-through rates. In this way, the operational information of each dynamic push message is counted in real time, and a first target beta distribution corresponding to the click-through rate of each dynamic push message is generated based on the Thompson sampling concept. A preset number of dynamic push messages are selected based on the first target beta distribution and the corresponding target click-through rates are obtained. Accurate target dynamic push messages are selected for push based on the target predicted click-through rates, greatly improving the accuracy of information processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0034] Figure 1 This is a schematic diagram of a scenario of an information processing system provided by an embodiment of the present application;

[0035] Figure 2a This is a flowchart of the information processing method provided by the embodiment of the present application;

[0036] Figure 2b A product diagram of the information processing method provided in an embodiment of the present application;

[0037] Figure 2c Another product schematic diagram of the information processing method provided in an embodiment of the present application;

[0038] Figure 2d Another product schematic diagram of the information processing method provided in an embodiment of the present application;

[0039] Figure 2e Another product schematic diagram of the information processing method provided in an embodiment of the present application;

[0040] Figure 3 is another flowchart of the information processing method provided in an embodiment of the present application;

[0041] Figure 4a This is a schematic diagram of the framework of the information processing method provided in the embodiment of the present application;

[0042] Figure 4b This is another schematic diagram of the information processing method provided by the embodiment of the present application;

[0043] Figure 5 is a schematic structural diagram of an information processing device provided in an embodiment of the present application;

[0044] Figure 6 It is a structural diagram of the server provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0046] Embodiments of the present application provide an information processing method, apparatus, and computer-readable storage medium.

[0047] See also Figure 1 , Figure 1 The scenario diagram of the information processing system provided in the embodiment of the present application includes: terminal A, and a server (the information processing system may also include other terminals in addition to terminal A, and the specific number of terminals is not limited here). Terminal A and the server can be connected through a communication network. The communication network may include a wireless network and a wired network, wherein the wireless network includes a combination of one or more of a wireless wide area network, a wireless local area network, a wireless metropolitan area network, and a wireless personal network. The network includes network entities such as routers and gateways, which are not shown in the figure. Terminal A can exchange information with the server through the communication network. For example, when terminal A is running an application containing various types of push information, such as video, short video, microblog and shopping applications, terminal A can detect the user's operation information on the push information, which includes at least exposure data and click data, and send the operation information to the server.

[0048] The information processing system may include an information processing device, which may be integrated into a server. Figure 1 In the embodiment, the server is mainly used to receive operation information uploaded by terminal A, count the operation information of each dynamic push information, the operation information includes at least exposure data and click data, generate a first target beta distribution corresponding to the click rate of each dynamic push information based on the exposure data and click data, select a preset number of dynamic push information according to the first target beta distribution, obtain the target predicted click rate of the preset number of dynamic push information, select the target dynamic push information for push according to the target predicted click rate, the target dynamic push information is more in line with the user's preferences, and can achieve a better creative optimization effect.

[0049] The information processing system may also include terminal A, which can install various applications required by users, such as video, short video, Weibo, shopping and other applications. For example, when terminal A runs a video application, terminal A can display push information, detect user operation information on the push information, which operation information includes at least exposure data and click data, and send the operation information to the server.

[0050] It should be noted that Figure 1 The scenario diagram of the information processing system shown is only an example. The information processing system and scenario described in the embodiment of the present application are intended to more clearly illustrate the technical solution of the embodiment of the present application, and do not constitute a limitation on the technical solution provided by the embodiment of the present application. Ordinary technicians in this field can know that with the evolution of information processing systems and the emergence of new business scenarios, the technical solution provided by the embodiment of the present application is also applicable to similar technical problems.

[0051] It should be noted that the serial numbers of the following embodiments are not intended to limit the preferred order of the embodiments.

[0052] Example 1

[0053] In this embodiment, the description will be made from the perspective of an information processing device. The information processing device can be specifically integrated into a server with a storage unit and a microprocessor installed and has computing capabilities. The server can be an independent physical server or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, as well as big data and artificial intelligence platforms.

[0054] See also Figure 2a , Figure 2a: is a flow chart of the information processing method provided in the embodiment of the present application. The information processing method includes:

[0055] In step 101, operation information of each dynamic push information is counted.

[0056] It should be noted that the push information is the introductory information about the product that the push host pays to the push platform through web pages, search engines, browsers or online media. The push information can be an advertisement, etc. The dynamic creative push information can be a dynamic creative advertisement (DC). The dynamic push information is a creative display form of push information that can display different advertising ideas to users. Users can create different dynamic push information for a push information. In order to better illustrate the dynamic push information of the embodiment of the present application, please refer to the following description:

[0057] Please also refer to Figure 2b As shown, Figure 2b This is a product diagram of the information processing method provided in an embodiment of the present application. The user can click the dynamic creative advertising control 11 on the client to create dynamic push information for pushing information. The settings of the dynamic push information, such as the location, scheduling, and bidding, are no different from those of ordinary advertisements.

[0058] Please continue reading Figure 2c As shown, Figure 2c Another product diagram of the information processing method provided in an embodiment of the present application, the user can add multiple pictures through the picture adding control 13 in the creative picture area 12, and add multiple creative copy through the copy adding control 15 in the creative copy area 14. The user can also add multiple titles.

[0059] Please continue reading Figure 2d As shown, Figure 2d Another product diagram of the information processing method provided in the embodiment of the present application, the client can combine the multiple pictures, multiple texts and multiple titles to form multiple dynamic push information ideas, such as Figure 2d As shown in the figure, 4 images, 3 titles and 4 creative copy (i.e. description) can form 4*3*4=48 dynamic push information. In the subsequent push information, one dynamic push information (i.e. creative) will be selected and sent to the user. In order to maximize the use of advertisers' resources, it is necessary to select the dynamic push information with the best creative among multiple dynamic push information for display. For example, please continue to refer to Figure 2e As shown, Figure 2e Another product diagram of the information processing method provided in the embodiment of the present application, after the dynamic push information is delivered, Figure 2eThe client page shown detects the feedback data of each creative idea. The dimensions may include various aspects that the push host wants to know about the dynamic push information, such as exposure data, click data, conversion data, click-through rate, virtual cost data (i.e., expenditure) of each creative idea, etc., to achieve an intuitive understanding of the pros and cons of the overall creative idea and help the push host select excellent dynamic push information.

[0060] Cloud technology is a general term for network technologies, information technologies, integration technologies, management platform technologies, and application technologies based on the cloud computing business model. It can form a resource pool that can be used flexibly and conveniently on demand. Cloud computing technology will become a crucial support. Backend services for technical network systems, such as video websites, image websites, and more portals, require extensive computing and storage resources. With the rapid development and application of the internet industry, every item will likely have its own unique identifier, requiring transmission to backend systems for logical processing. Different levels of data will be processed separately, and data from various industries will require a strong system backend, which can only be achieved through cloud computing.

[0061] Among them, the present application can use cloud technology to count the operation information of each dynamic push information in the same push information in real time. The operation information may include at least exposure data and click data. The exposure data is the number generated by each dynamic push information when it is exposed (i.e., displayed). The larger the exposure data, the more times the dynamic push information is displayed. The smaller the exposure data, the fewer times the dynamic push information is displayed. The click data is the number of clicks generated by the dynamic push information when it is exposed. When the client is exposed to the dynamic push information, the user can click on the dynamic push information according to his own interests and jump to the corresponding push page, such as the application download page, the recharge page, etc. If the user is not interested, he can also close the dynamic push information and not generate click data. The click data can reflect the user's interest in the dynamic push information.

[0062] In step 102 , a first target beta distribution corresponding to the click rate of each dynamic push information is generated based on the exposure data and the click data.

[0063] Artificial Intelligence (AI) refers to the theories, methods, techniques, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, to perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that seeks to understand the essence of intelligence and produce new intelligent machines that can respond in a manner similar to human intelligence. AI also studies the design principles and implementation methods of various intelligent machines, enabling them to possess the capabilities of perception, reasoning, and decision-making.

[0064] Artificial intelligence (AI) technology is a comprehensive discipline encompassing a wide range of fields, encompassing both hardware and software technologies. Foundational AI technologies generally include sensors, specialized AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0065] Machine learning (ML) is a multidisciplinary field that encompasses probability theory, statistics, approximation theory, convex analysis, and algorithmic complexity theory. It specifically studies how computers can simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is at the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications span all areas of AI. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and self-learning.

[0066] The solutions provided in the embodiments of this application involve technologies such as deep learning of artificial intelligence, and are specifically described through the following embodiments:

[0067] In related technologies, when a server pushes dynamic push information, it can adopt a random optimization strategy and a optimization strategy based on the previous click-through rate of the creative.

[0068] (a) Random selection: In each selection process, any creative is randomly selected from all dynamic push information with equal probability for exposure. An obvious disadvantage of this method is that it wastes exposure opportunities on low-quality creatives and does not have the ability to iteratively evolve and optimize based on creative feedback data.

[0069] (b) The optimization strategy based on the previous click-through rate of creatives is to score and sort each dynamic push information according to its historical click-through rate, and select any creative with the highest historical click-through rate for recommendation. However, the disadvantage of this strategy is that when the initial data of dynamic push information is low, the bad dynamic push information will be played in a concentrated manner due to accidental factors in the playback, resulting in a poor optimization result.

[0070] Among them, Thompson Sampling: Each dynamic push information does not have a clear push recommendation strength. The Thompson sampling can solve the problem of how to obtain the recommendation strength of each dynamic push information. The core of the Thompson sampling is the Beta distribution, which is a continuous probability distribution defined on the interval [0, 1]. The prior distribution information and posterior distribution information of the Beta distribution have a unified form. Assuming that the prior distribution is Beta (α, β), after s successes and f failures, its posterior distribution information can be Beta (s+α, f+β), realizing the addition of the posterior distribution information to the prior distribution information, so that the entire data is updated with real changes.

[0071] Furthermore, through the posterior distribution information Prior distribution information CTR~beta(α, β) is added to generate a first target beta distribution corresponding to the click rate of each dynamic push information. The first target beta distribution represents the actual click rate of each dynamic push information in practice.

[0072] In some embodiments, the step of generating a first target beta distribution corresponding to the click-through rate of each dynamic push information based on the exposure data and the click data includes:

[0073] (1) obtaining prior distribution information corresponding to click rates of historical dynamic push information, and generating a first beta distribution corresponding to the click rate of each dynamic push information based on the prior distribution information;

[0074] (2) Count the click data and non-click data generated by each dynamic push information when it is exposed, and generate the corresponding posterior distribution information of the click rate;

[0075] (3) The first beta distribution is adjusted according to the posterior distribution information corresponding to the click rate to obtain a first target beta distribution corresponding to the click rate of each dynamic push information.

[0076] Among them, since the click-through rate of all historical dynamic push information has an approximate range prediction, that is, it conforms to the prediction of an approximate range, the click-through rate of historical dynamic push information can be obtained to generate the corresponding prior distribution information CTR~beta(α, β), and the first beta distribution corresponding to the click-through rate of each dynamic push information is generated based on the prior distribution information.

[0077] Furthermore, statistics are collected on the click data generated when each dynamic push information is exposed and the corresponding non-click data when the user did not click on it. The non-click data can be generated by subtracting the click data from the exposure data. In this way, the posterior distribution information corresponding to the click rate is generated based on the click data and non-click data. The first beta distribution is adjusted according to the posterior distribution information corresponding to the click rate to obtain a first target beta distribution corresponding to the click rate of each dynamic push information.

[0078] In step 103, a preset number of dynamic push information is selected according to the first target beta distribution.

[0079] Among them, because the first target beta distribution is constantly combined with the click habits of users during actual use, it will become more and more in line with actual click usage as it continues to learn. The peak value of the first target beta distribution for dynamic push information with excellent click-through rate performance will become higher and higher, while the peak value of the first target beta distribution for dynamic push information with poor click-through rate performance will become lower and lower. In turn, different dynamic push information can be continuously differentiated according to actual use, which can better accurately express the click-through rate of dynamic push information. Based on this, a preset number of dynamic push information with the best click-through rate performance can be selected according to the first target beta distribution corresponding to each dynamic push information. This preset number can be 5, 10, etc.

[0080] In one embodiment, the step of selecting a preset number of dynamic push information according to the first target beta distribution may include:

[0081] (1) obtaining a predicted click rate for each dynamic push message according to the first target beta distribution;

[0082] (2) Selecting a preset number of target dynamic push information in descending order of the predicted click-through rates.

[0083] Among them, the sampling value of the first target beta distribution is the predicted click-through rate of the dynamic push information. Therefore, the sampling value of the first target beta distribution corresponding to each dynamic push information can be obtained respectively to obtain the predicted click-through rate of each dynamic push information. The higher the predicted click-through rate, the more popular the corresponding dynamic recommendation information is among users. The lower the predicted click-through rate, the less popular the corresponding dynamic recommendation information is among users. The best preset number of target dynamic push information is selected in order of the predicted click-through rate from high to low.

[0084] In step 104, target predicted click-through rates of a preset number of dynamic push information are obtained, and target dynamic push information is selected for push based on the target predicted click-through rates.

[0085] Among them, when obtaining a preset number of dynamic push information, the embodiment of the present application can also accurately sort the preset number of dynamic push information through an advertising click-through rate prediction (Predict Click-Through Rate, Pctr) model. The Pctr model is a mature advertising prediction model that can predict the precise target predicted click-through rate of dynamic push information. In this way, the target predicted click-through rate of a preset number of dynamic push information can be obtained through the Pctr model. The higher the target predicted click-through rate, the greater the click probability of the dynamic push information. The lower the target predicted click-through rate, the lower the click probability of the dynamic push information, thereby achieving accurate secondary prediction.

[0086] Furthermore, the preset number of dynamic push information can be sorted according to the target predicted click rate, and the target dynamic push information ranked first or second can be selected for push to achieve secondary screening. Therefore, the target dynamic push information is the best creative and can be pushed to the user first.

[0087] In one embodiment, the step of obtaining target predicted click-through rates of a preset number of dynamic push information and selecting target dynamic push information for push based on the target predicted click-through rates may include:

[0088] (1) predicting the preset number of dynamic push messages using a target click rate prediction model to obtain a target predicted click rate for the preset number of dynamic push messages;

[0089] (2) Sum the exposure data of each dynamic push information to obtain the target exposure data;

[0090] (3) Select the target dynamic push information and push it based on the target predicted click rate and target exposure data.

[0091] The target click-through rate prediction model (Pctr) is used to perform a precise ranking prediction on a preset number of dynamic push messages to obtain the target predicted click-through rates for the preset number of dynamic push messages. The exposure data of each dynamic push message is summed to obtain the total target exposure data, which reflects the exposure of the dynamic push message.

[0092] Furthermore, if in the early stages of exposure, that is, the target exposure data is less than a certain number, the exposure of each dynamic push information's creative is limited. Therefore, based on the idea of ​​random sampling, the corresponding target dynamic push information can be selected from a preset number of dynamic push information. The higher the target predicted click-through rate, the higher the probability of being selected, and the lower the target predicted click-through rate, the lower the probability of being selected. This ensures that in the early stages of exposure, all preset dynamic push information have the opportunity to be pushed. If in the later stages of exposure, that is, the target exposure data exceeds a certain number, and the accuracy of the Pctr model estimation reaches a certain level, the target dynamic push information with the highest target predicted click-through rate can be directly selected for push.

[0093] As can be seen from the above, the embodiment of the present application counts the operational information of each dynamic push message; generates a first target beta distribution corresponding to the click-through rate of each dynamic push message based on exposure data and click data; selects a preset number of dynamic push messages based on the first target beta distribution; obtains the target predicted click-through rate of the preset number of dynamic push messages, and selects the target dynamic push message for push based on the target predicted click-through rate. In this way, the operational information of each dynamic push message is counted in real time, the first target beta distribution corresponding to the click-through rate of each dynamic push message is generated based on the Thompson sampling concept, the preset number of dynamic push messages are selected based on the first target beta distribution and the corresponding target click-through rate is obtained, and the precise target dynamic push message is selected for push based on the target predicted click-through rate, which greatly improves the accuracy of information processing.

[0094] Example 2

[0095] The method described in Example 1 is further described in detail below with examples.

[0096] In this embodiment, the information processing device is specifically integrated into a server as an example for description.

[0097] See also Figure 3 , Figure 3 This is another flow chart of the information processing method provided in an embodiment of the present application. The method flow may include:

[0098] In step 201, the server collects statistics on the operation information of each dynamic push message.

[0099] Please also refer to Figure 4a , Figure 4aThe framework diagram of the information processing method provided in the embodiment of the present application is as follows. It should be noted that each dynamic creative advertisement includes multiple exclusive dynamic push information, which can be understood as advertising creativity. The data between different dynamic creative advertisements are isolated. The server obtains the operation information of each dynamic push information in each dynamic creative advertisement in real time through the real-time data stream module 21. The operation information includes at least exposure data, click data, conversion data, virtual spending data, etc.

[0100] Furthermore, the operation information of each dynamic push information is counted by summarizing the data stream, and the statistical time period can be the current day, the last three days, or all the summary data, etc.

[0101] In step 202, the server collects update information of the dynamic push information according to a preset period, and updates the dynamic push information according to the update information.

[0102] Please continue to see Figure 4a The server collects the update information of the push host on the dynamic push information through the real-time subscription module 23 according to a preset period, such as every 1 minute. The update information includes addition, deletion, query, and modification information. According to the update information, the push information stored in the server is replaced in real time to ensure that the push host's operations on the dynamic push information can be fed back to the background in a timely manner.

[0103] In step 203 , the server calculates virtual cost data for each dynamic push message based on the virtual expense data and the conversion data, and freezes the dynamic push messages corresponding to the virtual cost data that is greater than the preset virtual cost data.

[0104] The pusher sets a target CPA for each dynamic push message, and the server controls the bid price during the actual exposure, ensuring that the cost of each conversion behavior is within 1.2 times the pusher's expected cost. Figure 4a The cost control module 24 is responsible for suppressing the dynamic push information with poor cost performance in the virtual cost data of the dynamic push information, so as to achieve the goal of controlling the expected cost price of the push master.

[0105] Furthermore, when the conversion volume of the dynamic push information exceeds a certain amount, the cost control module 24 calculates the virtual cost data of each dynamic push information based on the ratio of the virtual cost data (i.e., cost) to the conversion data. For example, the calculation formula is as follows:

[0106]

[0107] The cpa is the virtual cost data, the sun (cost) is the virtual expenditure data, and the sum (conversion) is the conversion data. The virtual cost data cpa of each dynamic push information can be calculated through the ratio of the virtual expenditure data and the conversion data. The preset virtual data can be 1.2 times the expected cost price. When the virtual cost data is greater than the expected cost price, it means that the cost is running high, and the exposure opportunity of the dynamic push information with the cost running high should be reduced. That is, the dynamic push information corresponding to the virtual cost data greater than the preset virtual data can be frozen and will not participate in subsequent exposure, thereby protecting the interests of the push owner. The information filtered by the cost control module 24 can be quickly transmitted to the rough sorting module 26 through the data stream 25.

[0108] In step 204 , the server obtains prior distribution information corresponding to the click-through rates of historical dynamic push information, and generates a first beta distribution corresponding to the click-through rate of each dynamic push information based on the prior distribution information.

[0109] Among them, since the click-through rate of historical dynamic push information has an approximate range, before pushing the dynamic push information, an average probability prediction can be made for the dynamic push information. In this way, the prior distribution information CTR~beta(α, β) corresponding to the click-through rate of historical dynamic push information can be obtained, and the first beta distribution corresponding to the click-through rate of each dynamic push information is generated according to the prior distribution information. The first beta distribution is the average click-through rate distribution.

[0110] In step 205 , the server counts click data and non-click data generated when each dynamic push information is exposed, and generates posterior distribution information corresponding to the click rate.

[0111] The server counts the click data generated by each dynamic push information when it is exposed, and obtains the non-click data by subtracting the click data from the exposure data. Based on the click data and non-click data, the posterior distribution information corresponding to the click rate is generated.

[0112] In step 206 , the server adjusts the first beta distribution according to the posterior distribution information corresponding to the click rate, and obtains a first target beta distribution corresponding to the click rate of each dynamic push information.

[0113] Among them, the posterior distribution information of the server click rate The curve of the first beta distribution is adjusted to obtain the first target beta distribution corresponding to the click rate of each dynamic push. The first target beta distribution is continuously combined with the clicking habits of users during actual use, so the first target beta distribution will become more and more in line with the actual usage.

[0114] In step 207 , the server obtains target conversion data of each dynamic push information.

[0115] Among them, in the relevant technologies for pushing dynamic push information, the effects at the conversion data level are not taken into consideration. Conversion data is the behavioral data of the push owner's expected events such as application downloads or virtual recharges after the user clicks on the dynamic push information. This conversion effect is the core interest of the push owner, so corresponding improvements are needed.

[0116] In one embodiment, the step of obtaining target conversion data for each dynamic push information may include summing the conversion data for each dynamic push information to obtain the target conversion data.

[0117] In the embodiment of the present application, since the conversion data accumulates at a relatively slow speed, there may be no conversion data or little conversion data in the early stage for a long time. Therefore, the premise for considering the conversion data at the push level is that the target conversion data reaches a certain number. In this way, the server can sum the conversion data of each dynamic push information to obtain the total target conversion data.

[0118] In another embodiment, the step of obtaining target conversion data of each dynamically pushed information may further include counting average conversion data of the dynamically pushed information and determining the average conversion data as the target conversion data.

[0119] In the embodiment of the present application, since conversion data accumulates relatively slowly, there may be no conversion data or very little conversion data in the early stages for a long time. Therefore, the premise for considering conversion data at the push level is that the average conversion number of each dynamic push information reaches a certain number. For example, when the average conversion number of each dynamic push information in the same dynamic advertisement is greater than 1, the condition is met. In this way, the server can count the total conversion data of all dynamic push information, calculate the ratio of this total to the number of dynamic push information (i.e., the number of creatives), obtain the average conversion data, and determine this average conversion data as the target conversion data.

[0120] In step 208 , the server detects whether the target conversion data is less than a first preset threshold.

[0121] The first preset threshold is a critical value for determining whether the target conversion data has reached a certain number. The number of the first preset threshold can be the number of dynamic push information. When the server detects that the target conversion data is less than the first preset threshold, step 209 is executed. When the server detects that the target conversion data is not less than the first preset threshold, step 210 is executed.

[0122] In step 209 , the server obtains the predicted click rate of each dynamic push information according to the first target beta distribution, and selects a preset number of target dynamic push information in descending order of predicted click rate.

[0123] Among them, when the server detects that the target conversion data is less than the first preset threshold, it means that the conversion data does not meet the requirements. In this way, the server can obtain the sampling value of the first target beta distribution corresponding to each dynamic push information respectively to obtain the predicted click-through rate of each dynamic push information, and select a preset number of target dynamic push information in order from high to low according to the predicted click-through rate, thereby realizing a rough selection method of the rough sorting module 26.

[0124] In step 210 , the server collects statistics on conversion data and non-conversion data generated when each dynamic push information is clicked, and generates posterior distribution information corresponding to the conversion rate.

[0125] Among them, when the server detects that the target conversion data is not less than the first preset threshold, it means that the conversion data meets the requirements, and the conversion rate of each dynamic push information can be estimated. Based on the Thompson sampling principle, the conversion data and non-conversion data generated by each dynamic push information when it is clicked are counted. The conversion data is the number of times the user completes the expected operation after clicking the dynamic push information, and the non-conversion data is the number of times the user fails to complete the expected operation after clicking the dynamic push information. Since the data has a certain scale, the corresponding posterior distribution information CVR_post~beta (conversion, Click-conversion) of the conversion rate can be directly generated based on the conversion data and non-conversion data, where the conversion is the conversion data and the click is the click data.

[0126] In step 211 , the server generates a second target beta distribution corresponding to the conversion rate of each dynamic push information based on the posterior distribution information corresponding to the conversion rate.

[0127] Among them, the second target beta distribution corresponding to the conversion rate of each dynamic push information is generated based on the posterior distribution information CVR_post~beta (conversion, Click-conversion) of the conversion rate. The second target beta distribution can be continuously combined with the conversion habits of users during actual use. Therefore, the second target beta distribution will become more and more in line with the actual conversion usage with continuous use and learning.

[0128] In step 212, the server obtains the predicted click rate of each dynamic push message based on the first target beta distribution, obtains the predicted conversion rate of each dynamic push message based on the second target beta distribution, combines the predicted click rate and the predicted conversion rate of each dynamic push message to obtain a combination rate, and selects a preset number of dynamic push messages in descending order of the combination rates.

[0129] Among them, the server obtains the sampling value of the first target beta distribution corresponding to each dynamic push information respectively, obtains the predicted click-through rate of each dynamic push information, obtains the sampling value of the second target beta distribution corresponding to each dynamic push information respectively, obtains the predicted conversion rate of each dynamic push information, combines the predicted click-through rate and the predicted conversion rate of each dynamic push information, such as multiplying the predicted click-through rate and the predicted conversion rate to obtain the combination rate, selects a preset number of dynamic push information in order of the combination rate from high to low, and realizes the effect of combining the preset number of dynamic push information with the conversion data level, and selects the dynamic push information based on the click-through rate dimension and the conversion rate dimension to realize another rough selection method of the rough sorting module 26.

[0130] In step 213 , the server predicts a preset number of dynamic push messages using a target click rate prediction model to obtain target predicted click rates for the preset number of dynamic push messages and target exposure data for each dynamic push message.

[0131] Please also refer to Figure 4a and Figure 4b The coarse ranking module 26 includes a lightweight click-through rate prediction model (Litectr) and a lightweight conversion prediction model (Litecvr). The lightweight click-through rate prediction model can obtain the predicted click-through rate of each dynamic push information, and the lightweight conversion prediction model can obtain the predicted conversion rate of each dynamic push information. The effective cost per thousand impressions (ECPM) indicator of the advertisement is then obtained. All advertisements are sorted from high to low according to the ECPM, and the top N advertisements are selected. For example, please refer to the following formula:

[0132] ecpm=ocpa_bid*Litectr*Litecvr

[0133] The ocpa_bid is the ad bid, the Litectr is the predicted click-through rate, and the Litecvr is the predicted conversion rate. The product of the ad bid, the predicted click-through rate, and the predicted conversion rate is calculated using the above formula to calculate the rough ecpm indicator of each ad.

[0134] Furthermore, since the first N advertisements include dynamic creative advertisements, and the dynamic creative advertisements contain multiple creative ideas (i.e., dynamic push information), it is necessary to optimize the advertising creative ideas for each dynamic creative advertisement in the first N advertisements. The specific method of optimizing advertising creative ideas refers to the above-mentioned step of selecting a preset number of target dynamic push information. After selecting the preset number of target dynamic push information, it is sent to the precision ranking module 27. The precision ranking module 27 includes a target click-through rate prediction model (Pctr) and a target conversion rate prediction model (Pcvr). The preset number of dynamic push information is predicted by the target click-through rate prediction model to obtain the target predicted click-through rate of the preset number of dynamic push information. The preset number of dynamic push information is predicted by the target conversion rate prediction model to obtain the target predicted conversion rate of the preset number of dynamic push information. Since the precision ranking module 27 inputs more features of Pctr and Pcvr, the prediction accuracy of Pctr and Pcvr will be higher.

[0135] In one embodiment, the step of obtaining target exposure data for each dynamic push information may include summing the exposure data for each dynamic push information to obtain the target exposure data.

[0136] In an embodiment of the present application, the server may sum the exposure data of each dynamic push information to obtain the total target exposure data.

[0137] In another embodiment, the step of obtaining target conversion data of each dynamic push information may further include counting average exposure data of the dynamic push information and determining the average exposure data as the target exposure data.

[0138] In an embodiment of the present application, the server can count the total exposure data of all dynamic push information, calculate the ratio of the total to the number of dynamic push information (i.e., the number of creatives), obtain the average exposure data, and determine the average exposure data as the target exposure data.

[0139] In step 214 , the server detects whether the target exposure data is less than a second preset threshold.

[0140] Among them, since the learning of the target click-through rate prediction model requires certain data support, in the early stage of exposure, that is, when the target exposure data is small, the target click-through rate prediction model may not be accurate, but in the late stage of exposure, that is, when the target exposure data is large, the target click-through rate prediction model is accurate, so it is necessary to set a second preset threshold to distinguish whether the current stage is in the early or late stage of exposure. The second preset threshold can be 5000 or 300, etc. When the server detects that the target exposure data is less than the second preset threshold, step 215 is executed. When the server detects that the target exposure data is not less than the second preset threshold, step 216 is executed.

[0141] In step 215, the server normalizes the target predicted click rate to obtain target prediction vector information of a preset number of dimensions, divides the probability interval based on the target prediction vector information, randomly accesses the probability interval, and determines the dynamic push information corresponding to the accessed probability interval as the target dynamic push information for push.

[0142] Among them, when the server detects that the target exposure data is less than the second preset threshold, it means that the fine ranking module 27 is in the early stage of exposure, the exposure of the preset number of dynamic push information is limited, and the Pctr model has inaccuracies. At this time, it is not possible to fully optimize according to the Pctr model. It is necessary to normalize the target predicted click rate through the softmax function to obtain the target prediction vector information of the preset number of dimensions. The probability range of the vector elements of each dimension is between (0, 1), and the sum of the vector elements of all dimensions is 1. The larger the target predicted click rate, the larger the probability range after conversion. The softmax function is as follows:

[0143]

[0144] The Win_rate i is the target prediction vector information, the e is a constant, the pctr i For the target predicted click-through rate, numof creatives is the preset number minus 1.

[0145] Furthermore, the probability interval of (0, 1) is divided according to the probability of each element in the target prediction vector information. The higher the probability, the larger the assigned probability interval, and the lower the probability, the smaller the assigned probability. The probability interval after random access is accessed. It should be noted that the larger the probability interval, the greater the probability of being accessed, and the smaller the probability interval, the smaller the probability of being accessed. The dynamic push information corresponding to the accessed probability interval is determined as the target dynamic push information for push, that is, the target dynamic push information represents the dynamic creative advertisement in subsequent bidding. In this way, the dynamic push information with a low target predicted click-through rate but a high conversion rate can also be creatively recommended, which fully takes into account the interests of the pusher and diversifies the push of the target dynamic push information.

[0146] In step 216 , the server determines the dynamic push information with the highest target predicted click rate as the target dynamic push information for push.

[0147] Among them, when the server detects that the target exposure data is not less than the second preset threshold, it means that the precise ranking module 27 is in the late exposure stage, the exposure of the preset number of dynamic push information meets the requirements, and the Pctr model prediction is accurate. The dynamic push information with the highest target predicted click-through rate can be directly determined as the target push information for push, that is, the target dynamic push information will represent the dynamic creative advertisement in the subsequent bidding.

[0148] Please continue reading Figure 4b After the fine ranking module 27 determines the optimal target dynamic push information of each dynamic creative advertisement as a representative, it further calculates a more accurate target ecpm index of the top N advertisements, and selects the best 1 to 2 advertisements to push to the user based on the target ecpm index. For example, please refer to the following formula:

[0149] ecpm=ocpa_bid*Pctr*Pcvr

[0150] The ocpa_bid is the ad bid, the Pctr is the target predicted click-through rate, and the Litecvr is the target predicted conversion rate. The ad bid, target predicted click-through rate, and target predicted conversion rate are calculated using the above formula to calculate the precise ranking ecpm index for each ad. Based on the precise ranking ecpm index, the best 1 to 2 ads are selected and pushed to users.

[0151] In some embodiments, the refined ranking module 27 predicts a preset number of dynamic push information using a target click rate prediction model. After obtaining the target predicted click rates of the preset number of dynamic push information, the refined ranking module 27 may further include:

[0152] (1) A preset number of dynamic push messages are predicted using a target conversion rate (PCVR) model to obtain a target predicted conversion rate for the preset number of dynamic push messages. The target predicted click rate and the target predicted conversion rate are multiplied to obtain a target binding rate. The exposure data of each dynamic push message is summed to obtain target exposure data.

[0153] (2) The server detects whether the target exposure data is less than a second preset threshold.

[0154] When the server detects that the target exposure data is less than the second preset threshold, step (3) is executed; when the server detects that the target exposure data is not less than the second preset threshold, step (4) is executed.

[0155] (3) The server normalizes the target binding rate to obtain the binding prediction vector information of a preset number of dimensions, divides the probability interval based on the binding prediction vector information, randomly accesses the probability interval, and determines the dynamic push information corresponding to the accessed probability interval as the target dynamic push information for push.

[0156] (4) The server determines the dynamic push information with the highest target binding rate as the target dynamic push information and pushes it.

[0157] Due to the introduction of the combination of target predicted conversion rate and target predicted click-through rate, the refined ranking module 27 further introduces preferential recommendation of target dynamic push information at the conversion level, so that the target dynamic push information is more accurate. The same description part refers to the above and will not be elaborated here.

[0158] As can be seen from the above, the embodiment of the present application counts the operational information of each dynamic push message; generates a first target beta distribution corresponding to the click-through rate of each dynamic push message based on exposure data and click data; selects a preset number of dynamic push messages based on the first target beta distribution; obtains the target predicted click-through rate of the preset number of dynamic push messages, and selects the target dynamic push message for push based on the target predicted click-through rate. In this way, the operational information of each dynamic push message is counted in real time, the first target beta distribution corresponding to the click-through rate of each dynamic push message is generated based on the Thompson sampling concept, the preset number of dynamic push messages are selected based on the first target beta distribution and the corresponding target click-through rate is obtained, and the precise target dynamic push message is selected for push based on the target predicted click-through rate, which greatly improves the accuracy of information processing.

[0159] Furthermore, since the conversion data level is introduced in the selection process to screen the dynamic push information, the push of the target dynamic push information is more in line with the requirements of the pusher, the effect of creative selection is improved, and the accuracy of information processing is further improved.

[0160] Example 3:

[0161] To facilitate better implementation of the information processing method provided in the embodiments of the present application, the embodiments of the present application also provide a device based on the above information processing method. The meanings of the terms are the same as those in the above information processing method, and the specific implementation details can be referred to the description in the method embodiment.

[0162] See also Figure 5 , Figure 5 This is a structural diagram of an information processing device provided in an embodiment of the present application, wherein the information processing device may include a statistical unit 301, a generation unit 302, a first rough sorting unit 303, and a fine sorting unit 304, etc.

[0163] The statistics unit 301 is used to count the operation information of each dynamic push information, and the operation information at least includes exposure data and click data.

[0164] The generating unit 302 is configured to generate a first target beta distribution corresponding to a click rate of each dynamic push information based on the exposure data and the click data.

[0165] In some embodiments, the generation unit 302 is used to: obtain prior distribution information corresponding to the click-through rate of historical dynamic push information, and generate a first beta distribution corresponding to the click-through rate of each dynamic push information based on the prior distribution information; count the click data and non-click data generated by each dynamic push information when it is exposed, and generate posterior distribution information corresponding to the click-through rate; adjust the first beta distribution based on the posterior distribution information corresponding to the click-through rate, and obtain a first target beta distribution corresponding to the click-through rate of each dynamic push information.

[0166] The first coarse sorting unit 303 is configured to select a preset number of dynamic push information according to the first target beta distribution.

[0167] In some embodiments, the first coarse sorting unit 303 is used to: obtain a predicted click rate of each dynamic push information according to the first target beta distribution; and select a preset number of target dynamic push information in descending order of the predicted click rates.

[0168] The fine sorting unit 304 is configured to obtain target predicted click-through rates of the preset number of dynamic push information, and select target dynamic push information for push based on the target predicted click-through rates.

[0169] In some embodiments, the fine sorting unit 304 includes:

[0170] A prediction subunit, configured to predict the preset number of dynamic push information using a target click rate prediction model to obtain target predicted click rates for the preset number of dynamic push information;

[0171] Exposure subunit, used to obtain target exposure data for each dynamic push information;

[0172] The combining subunit is used to select target dynamic push information for push based on the target predicted click rate and target exposure data.

[0173] In some embodiments, the combining sub-unit is used to: when it is detected that the target exposure data is less than a second preset threshold, normalize the target predicted click-through rate to obtain target prediction vector information of a preset number of dimensions; divide the probability interval based on the target prediction vector information, randomly access the probability interval, and determine the dynamic push information corresponding to the accessed probability interval as the target dynamic push information for push; when it is detected that the target exposure data is not less than the second preset threshold, determine the dynamic push information with the highest target predicted click-through rate as the target dynamic push information for push.

[0174] In some embodiments, the operation information further includes conversion data, and the apparatus further includes:

[0175] A conversion unit, used to obtain target conversion data for each dynamic push information;

[0176] The first coarse sorting unit is configured to, when detecting that the target conversion data is less than a first preset threshold, execute the step of selecting a preset number of dynamic push information according to the first target beta distribution;

[0177] The second coarse sorting unit is used to count the conversion data and non-conversion data generated when each dynamic push information is clicked, and generate posterior distribution information corresponding to the conversion rate when it is detected that the target conversion data is not less than a first preset threshold value; generate a second target beta distribution corresponding to the conversion rate of each dynamic push information based on the posterior distribution information corresponding to the conversion rate; and select a preset number of dynamic push information in combination with the first target beta distribution and the second target beta distribution.

[0178] In some embodiments, the second coarse sorting unit is further used to: when it is detected that the target conversion data is not less than a first preset threshold, count the conversion data and non-conversion data generated by each dynamic push information when it is clicked, and generate posterior distribution information corresponding to the conversion rate; generate a second target beta distribution corresponding to the conversion rate of each dynamic push information based on the posterior distribution information corresponding to the conversion rate; obtain the predicted click-through rate of each dynamic push information based on the first target beta distribution; obtain the predicted conversion rate of each dynamic push information based on the second target beta distribution; combine the predicted click-through rate and the predicted conversion rate of each dynamic push information to obtain a combination rate; and select a preset number of dynamic push information in descending order according to the combination rate.

[0179] In some embodiments, the operation information further includes virtual expense data, and the apparatus further includes:

[0180] The cost control unit is used to calculate the virtual cost data of each dynamic push information according to the virtual expense data and the conversion data; and freeze the dynamic push information corresponding to the virtual cost data being greater than the preset virtual data.

[0181] In some embodiments, the apparatus further comprises:

[0182] The updating unit is used to collect update information of the dynamic push information according to a preset period; and update the dynamic push information according to the update information.

[0183] The specific implementation of each of the above units can be found in the previous embodiments and will not be described again here.

[0184] As can be seen from the above, the embodiment of the present application uses the statistical unit 301 to count the operation information of each dynamic push information; the generation unit 302 generates the first target beta distribution corresponding to the click rate of each dynamic push information based on the exposure data and the click data; the first coarse sorting unit 303 selects a preset number of dynamic push information according to the first target beta distribution; the fine sorting unit 304 obtains the target predicted click rate of the preset number of dynamic push information, and selects the target dynamic push information for push based on the target predicted click rate. In this way, the operation information of each dynamic push information is counted in real time, the first target beta distribution corresponding to the click rate of each dynamic push information is generated based on the Thompson sampling idea, the preset number of dynamic push information is selected according to the first target beta distribution and the corresponding target click rate is obtained, and the accurate target dynamic push information is selected for push based on the target predicted click rate, which greatly improves the accuracy of information processing.

[0185] Example 4:

[0186] The embodiment of the present application also provides a server, such as Figure 6 As shown, it shows a schematic diagram of the structure of the server involved in the embodiment of the present application, specifically:

[0187] The server may include one or more processing core processors 401, one or more computer-readable storage media memories 402, a power supply 403, an input unit 404 and other components. Those skilled in the art will appreciate that Figure 6 The server structure shown in the figure does not constitute a limitation to the server, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0188] in:

[0189] Processor 401 is the server's control center, connecting various components of the server using various interfaces and circuits. By running or executing software programs and / or modules stored in memory 402 and accessing data stored in memory 402, it performs various server functions and processes data, thereby performing overall server testing. Optionally, processor 401 may include one or more processing cores; preferably, processor 401 may integrate an application processor and a modem processor, with the application processor primarily processing the operating system, user interface, and application programs, while the modem processor primarily handles wireless communications. It is understood that the modem processor may not be integrated into processor 401.

[0190] The memory 402 can be used to store software programs and modules. The processor 401 executes various functional applications and data processing by running the software programs and modules stored in the memory 402. The memory 402 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the server, etc. In addition, the memory 402 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage device. Accordingly, the memory 402 may also include a memory controller to provide the processor 401 with access to the memory 402.

[0191] The server also includes a power supply 403 for supplying power to various components. Preferably, the power supply 403 can be logically connected to the processor 401 via a power management system, thereby enabling the power management system to manage charging, discharging, and power consumption. The power supply 403 can also include one or more DC or AC power supplies, a recharging system, a power failure detection circuit, a power converter or inverter, a power status indicator, and other arbitrary components.

[0192] The server may further include an input unit 404, which may be configured to receive input digital or character information and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

[0193] Although not shown, the server may further include a display unit, etc., which will not be described in detail here. Specifically, in this embodiment, the processor 401 in the server will load the executable files corresponding to one or more application processes into the memory 402 according to the following instructions, and the processor 401 will run the application stored in the memory 402 to implement various functions as follows:

[0194] Statistics are provided for the operation information of each dynamic push message, where the operation information includes at least exposure data and click data; a first target beta distribution corresponding to the click rate of each dynamic push message is generated based on the exposure data and click data; a preset number of dynamic push messages are selected according to the first target beta distribution; target predicted click rates of the preset number of dynamic push messages are obtained, and target dynamic push messages are selected for push according to the target predicted click rates.

[0195] In the above embodiments, the description of each embodiment has its own focus. For the part that is not described in detail in a certain embodiment, please refer to the detailed description of the information processing method above and will not be repeated here.

[0196] As can be seen from the above, the server of the embodiment of the present application can count the operational information of each dynamic push message; generate a first target beta distribution corresponding to the click-through rate of each dynamic push message based on exposure data and click data; select a preset number of dynamic push messages based on the first target beta distribution; obtain the target predicted click-through rate of the preset number of dynamic push messages, and select the target dynamic push message for push based on the target predicted click-through rate. In this way, the operational information of each dynamic push message is counted in real time, the first target beta distribution corresponding to the click-through rate of each dynamic push message is generated based on the Thompson sampling concept, the preset number of dynamic push messages are selected based on the first target beta distribution and the corresponding target click-through rate is obtained, and the precise target dynamic push message is selected for push based on the target predicted click-through rate, which greatly improves the accuracy of information processing.

[0197] Example 5

[0198] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be accomplished by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0199] To this end, an embodiment of the present application provides a computer-readable storage medium storing a plurality of instructions that can be loaded by a processor to execute the steps of any of the information processing methods provided in the embodiments of the present application. For example, the instructions can execute the following steps:

[0200] Statistics are provided for the operation information of each dynamic push message, where the operation information includes at least exposure data and click data; a first target beta distribution corresponding to the click rate of each dynamic push message is generated based on the exposure data and click data; a preset number of dynamic push messages are selected according to the first target beta distribution; target predicted click rates of the preset number of dynamic push messages are obtained, and target dynamic push messages are selected for push according to the target predicted click rates.

[0201] The computer-readable storage medium may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0202] Since the instructions stored in the computer-readable storage medium can execute the steps in any information processing method provided in the embodiments of the present application, the beneficial effects that can be achieved by any information processing method provided in the embodiments of the present application can be achieved. Please refer to the previous embodiments for details and will not be repeated here.

[0203] The above is a detailed introduction to an information processing method, device and computer-readable storage medium provided in the embodiments of the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for technical personnel in this field, based on the ideas of the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. An information processing method, characterized in that: include: Collecting statistics on the operation information of each dynamic push message, wherein the operation information includes at least exposure data, conversion data, and click data; generating a first target beta distribution corresponding to the click rate of each dynamic push information based on the exposure data and the click data, wherein the first target beta distribution represents the actual click rate of each dynamic push information in practice; Obtain target conversion data for each dynamic push information, where the target conversion data is obtained by summing the conversion data of each dynamic push information, or is an average conversion data obtained by counting the dynamic push information; When it is detected that the target conversion data is less than a first preset threshold, executing the step of selecting a preset number of dynamic push information according to the first target beta distribution; When it is detected that the target conversion data is not less than a first preset threshold, the conversion data and non-conversion data generated when each dynamic push information is clicked are counted to generate posterior distribution information corresponding to the conversion rate; generating a second target beta distribution corresponding to the conversion rate of each dynamic push information based on the posterior distribution information corresponding to the conversion rate, wherein the second target beta distribution fits the actual conversion usage of each dynamic push information during actual use; Selecting a preset number of dynamic push information in combination with the first target beta distribution and the second target beta distribution; The target predicted click rates of the preset number of dynamic push information are obtained, and target dynamic push information is selected for push according to the target predicted click rates.

2. The information processing method according to claim 1, wherein: The step of generating a first target beta distribution corresponding to the click rate of each dynamic push information based on the exposure data and the click data includes: Obtaining prior distribution information corresponding to click rates of historical dynamic push information, and generating a first beta distribution corresponding to the click rate of each dynamic push information based on the prior distribution information; Count the click data and non-click data generated by each dynamic push information when it is exposed, and generate the corresponding posterior distribution information of the click rate; The first beta distribution is adjusted according to the posterior distribution information corresponding to the click-through rate to obtain a first target beta distribution corresponding to the click-through rate of each dynamically pushed information.

3. The information processing method according to claim 2, wherein: The step of selecting a preset number of dynamic push information according to the first target beta distribution includes: Obtaining a predicted click rate for each dynamic push message according to the first target beta distribution; A preset number of target dynamic push information is selected in descending order of the predicted click rates.

4. The information processing method according to claim 3, wherein: The step of selecting a preset number of dynamic push information in combination with the first target beta distribution and the second target beta distribution includes: Obtaining a predicted click rate for each dynamic push message according to the first target beta distribution; Obtaining a predicted conversion rate for each dynamic push message according to the second target beta distribution; Combine the predicted click rate and predicted conversion rate of each dynamic push information to obtain the combined rate; A preset number of dynamic push information is selected in descending order of the combination rates.

5. The information processing method according to claim 3, wherein: The operation information also includes virtual expense data, and the method further includes: Calculating virtual cost data for each dynamic push information based on the virtual spending data and the conversion data; The dynamic push information corresponding to the virtual cost data being greater than the preset virtual data is frozen.

6. The information processing method according to any one of claims 1 to 5, characterized in that: The step of obtaining target predicted click-through rates of the preset number of dynamic push information, and selecting target dynamic push information to push according to the target predicted click-through rates, includes: Predicting the preset number of dynamic push messages using a target click rate prediction model to obtain target predicted click rates for the preset number of dynamic push messages; Obtain target exposure data for each dynamic push information; Target dynamic push information is selected and pushed in combination with the target predicted click rate and target exposure data.

7. The information processing method according to claim 6, characterized in that: The step of selecting target dynamic push information for push based on the target predicted click-through rate and target exposure data includes: When it is detected that the target exposure data is less than a second preset threshold, normalizing the target predicted click rate to obtain target prediction vector information of a preset number of dimensions; Divide the probability intervals based on the target prediction vector information, randomly access the probability intervals, and determine the dynamic push information corresponding to the accessed probability intervals as target dynamic push information for push; When it is detected that the target exposure data is not less than a second preset threshold, the dynamic push information with the highest target predicted click rate is determined as the target dynamic push information for push.

8. The information processing method according to any one of claims 1 to 5, characterized in that: The method further comprises: Collect updated information of dynamic push information according to a preset period; An update operation is performed on the dynamic push information according to the update information.

9. An information processing device, characterized in that include: A statistics unit, configured to collect statistics on operation information of each dynamic push message, wherein the operation information includes at least exposure data, conversion data, and click data; a generating unit, configured to generate a first target beta distribution corresponding to a click-through rate of each dynamic push information based on the exposure data and the click data, wherein the first target beta distribution represents an actual click-through rate of each dynamic push information in practice; a first coarse sorting unit, configured to select a preset number of dynamic push information according to the first target beta distribution; A sorting unit, configured to obtain target predicted click-through rates of the preset number of dynamic push information, and select target dynamic push information for push based on the target predicted click-through rates; The device further comprises: a conversion unit, configured to obtain target conversion data for each dynamically pushed information, wherein the target conversion data is obtained by summing the conversion data of each dynamically pushed information, or is an average conversion data obtained by statistically analyzing the dynamic pushed information; The first coarse sorting unit is configured to, when detecting that the target conversion data is less than a first preset threshold, execute the step of selecting a preset number of dynamic push information according to the first target beta distribution; The second coarse sorting unit is configured to, when detecting that the target conversion data is not less than a first preset threshold, count the conversion data and non-conversion data generated when each dynamic push information is clicked, and generate posterior distribution information corresponding to the conversion rate; generating a second target beta distribution corresponding to the conversion rate of each dynamic push information based on the posterior distribution information corresponding to the conversion rate, wherein the second target beta distribution fits the actual conversion usage of each dynamic push information during actual use; A preset number of dynamic push information is selected in combination with the first target beta distribution and the second target beta distribution.

10. The information processing device according to claim 9, wherein The generating unit is configured to: Obtaining prior distribution information corresponding to click rates of historical dynamic push information, and generating a first beta distribution corresponding to the click rate of each dynamic push information based on the prior distribution information; Count the click data and non-click data generated by each dynamic push information when it is exposed, and generate the corresponding posterior distribution information of the click rate; The first beta distribution is adjusted according to the posterior distribution information corresponding to the click-through rate to obtain a first target beta distribution corresponding to the click-through rate of each dynamically pushed information.

11. The information processing device according to claim 10, wherein: The first coarse row unit is used to: Obtaining a predicted click rate for each dynamic push message according to the first target beta distribution; A preset number of target dynamic push information is selected in descending order of the predicted click rates.

12. The information processing device according to claim 9, wherein The second coarse row unit is further used for: When it is detected that the target conversion data is not less than a first preset threshold, the conversion data and non-conversion data generated when each dynamic push information is clicked are counted to generate posterior distribution information corresponding to the conversion rate; generating a second target beta distribution corresponding to the conversion rate of each dynamic push information according to the posterior distribution information corresponding to the conversion rate; Obtaining a predicted click rate for each dynamic push message according to the first target beta distribution; Obtaining a predicted conversion rate for each dynamic push message according to the second target beta distribution; Combine the predicted click rate and predicted conversion rate of each dynamic push information to obtain the combined rate; A preset number of dynamic push information is selected in descending order of the combination rates.

13. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor to execute the steps of the information processing method according to any one of claims 1 to 8.

Images