Information processing method and device applied to advertising engine

Abstract

The present application discloses an information processing method and device applied to an advertising engine. A specific implementation of the method includes: obtaining the user's advertising demands; determining an executable policy set that meets the advertising demands through a multi-level intelligent agent, wherein each level of the multi-level intelligent agent can determine an executable policy set corresponding to the advertising demands, and the executable policy set includes executable policies corresponding to each of the multiple sub-processes in the entire advertising process; executing the executable policies corresponding to each of the multiple sub-processes in the executable policy set through the business interfaces corresponding to each of the multiple sub-processes. The multi-level intelligent agent in the advertising engine of the present application can automatically process the entire advertising process based on the user's advertising demands, which lowers the participation threshold of the advertising business and improves the user experience and convenience in the advertising processing process.

Description

Technical Field

[0001] The embodiments of the present application relate to the fields of computer technology, specifically to the fields of advertising and artificial intelligence technology, and more particularly to an information processing method, device, computer-readable medium, and electronic device applied to an advertising engine. Background Art

[0002] Advertising, as a traditional business, has long reached a bottleneck in its development. Leveraging information technology, internet advertising has achieved explosive growth in monetization efficiency. However, this exceptionally high monetization efficiency has driven a constant pursuit of excellence in the advertising business. From ad creation to placement, broadcast, and effectiveness verification, the chain of operations is becoming increasingly lengthy, with more and more branches and increasingly sophisticated operations. The immense scale of the advertising industry has spawned a variety of advertising engines, including full-featured, simplified, and automated ones. However, the advertising sector still faces challenges such as a low advertiser experience and high barriers to entry. Summary of the Invention

[0003] The embodiments of the present application provide an information processing method, device, computer-readable medium, and electronic device for an advertising engine.

[0004] In the first aspect, an embodiment of the present application provides an information processing method applied to an advertising engine, including: obtaining the user's advertising demands; determining an executable policy set that meets the advertising demands through a multi-level intelligent agent, wherein the intelligent agent at each level in the multi-level intelligent agent can determine the executable policy set corresponding to the advertising demands, and the executable policy set includes executable policies corresponding to multiple sub-processes in the entire advertising process; executing the executable policies corresponding to multiple sub-processes in the executable policy set through the business interfaces corresponding to the multiple sub-processes.

[0005] In some examples, each level of agents in the multi-level agent includes multiple agents, and the above-mentioned determination of an executable strategy set that meets the advertising demand through the multi-level agent includes: processing multiple sub-demands in the advertising demand according to the execution order of the agents at each level in the multi-level agent, and in response to the presence of a target sub-demand among the multiple sub-demands for which the agent at the current level has not determined an executable strategy based on a preset voting strategy, processing the target sub-demand by the agents at subsequent levels until the executable strategies corresponding to the multiple sub-demands are combined to obtain an executable strategy set.

[0006] In some examples, the first level of agents in the multi-level agent includes an agent group corresponding to each of the multiple sub-processes, and the agent group includes multiple agents, and the above-mentioned execution order of the agents at each level in the multi-level agent is used to process multiple sub-demands in the advertising appeal. In response to the presence of a target sub-demand among the multiple sub-demands for which the agent at the current level has not determined an executable strategy based on the preset voting strategy, the target sub-demand is processed by the agents at subsequent levels until the executable strategies corresponding to the multiple sub-demands are combined to obtain an executable strategy set, including: decomposing the user appeal to determine the sub-demand corresponding to each of the multiple sub-processes; processing the sub-demand corresponding to each of the multiple sub-processes through the agent group corresponding to each of the multiple sub-processes to obtain multiple candidate executable strategies corresponding to each of the multiple sub-processes; for the multiple candidate executable strategies output by each of the multiple agent groups, determining whether an executable strategy can be determined from the multiple candidate executable strategies based on the preset voting strategy; in response to the fact that the executable strategy corresponding to each of the multiple agent groups can be determined, the executable strategy set is obtained by combining the executable strategies corresponding to the multiple agent groups.

[0007] In some examples, the above-mentioned processing of multiple sub-appeals in the advertising appeal according to the execution order of the agents in each layer of the multi-level agents, in response to the presence of a target sub-appeal among the multiple sub-appeals for which the agents in the current layer have not determined an executable strategy based on the preset voting strategy, the target sub-appeal is processed by the agents in the subsequent layers until the executable strategies corresponding to the multiple sub-appeals are combined to obtain an executable strategy set. It also includes: in response to determining that there is a target agent group among the multiple agent groups for which an executable strategy has not been determined, the target sub-appeal is processed by the multiple agents in the subsequent layers until an executable strategy is determined from the candidate executable strategies output by the multiple agents in the current layer based on the preset voting strategy, wherein the multiple agents in the subsequent layers are determined according to the execution order; and the executable strategies determined by the other agent groups except the target agent group in the multiple agent groups and the executable strategies determined by the multiple agents in the current layer are combined to obtain an executable strategy set.

[0008] In some examples, the above-mentioned intelligent agent groups corresponding to each of the multiple sub-processes process the sub-demands corresponding to each of the multiple sub-processes to obtain multiple candidate executable strategies corresponding to each of the multiple sub-processes, including: parsing the sub-demands corresponding to each of the multiple sub-processes through the intelligent agent groups corresponding to each of the multiple sub-processes to obtain advertising terms corresponding to each of the multiple sub-demands; for each of the multiple intelligent agent groups, according to the advertising terms corresponding to the sub-demand to be processed by the intelligent agent group, determining the candidate executable strategy corresponding to the sub-process corresponding to the intelligent agent group from a preset limited result set.

[0009] In some examples, the above-mentioned multi-level intelligent agents are obtained in the following manner: determining the initial models corresponding to each multi-level intelligent agent based on the accuracy and parameter scale of the models in the model set, wherein the level to which the intelligent agent belongs and the accuracy and parameter scale of the initial model corresponding to the level are positively correlated; determining the first knowledge base corresponding to each multi-level intelligent agent, wherein the first knowledge base corresponding to intelligent agents at different levels includes advertising knowledge of different preset dimensions; training the initial models corresponding to each multi-level intelligent agent through the first knowledge base corresponding to each multi-level intelligent agent to obtain the multi-level intelligent agent.

[0010] In some examples, multiple sub-processes include an advertising creative sub-process, an advertising delivery sub-process, an advertising operation sub-process, and an advertising playback sub-process, and the above-mentioned business interfaces corresponding to each of the multiple sub-processes are used to execute the executable policies corresponding to each of the multiple sub-processes in the executable policy set, including: generating advertising creatives through the business interface corresponding to the advertising creative sub-process according to the executable policy corresponding to the advertising creative sub-process in the executable policy set; executing the delivery process of advertising creatives through the business interface corresponding to the advertising delivery sub-process according to the executable policy corresponding to the advertising delivery sub-process in the executable policy set; executing the operation process of advertising creatives through the business interface corresponding to the advertising operation sub-process according to the executable policy corresponding to the advertising operation sub-process in the executable policy set; executing the playback process of advertising creatives through the business interface corresponding to the advertising playback sub-process according to the executable policy corresponding to the advertising playback sub-process in the executable policy set.

[0011] In some examples, the above-mentioned executable strategy corresponding to the advertising delivery sub-process in the executable strategy set executes the advertising creative delivery process, including: determining the target node existing in the advertising delivery graph from the advertising appeal, wherein the advertising delivery graph represents all advertising links in the advertising delivery process; determining the advertising delivery links between the target nodes from the advertising delivery graph; and using the executable strategy corresponding to the advertising delivery sub-process to execute the advertising creative delivery process according to the advertising delivery links.

[0012] In some examples, determining the advertisement delivery links between target nodes from the advertisement delivery graph includes: using a depth-first traversal algorithm to determine the advertisement delivery links between target nodes from the advertisement delivery graph.

[0013] In some examples, the above-mentioned advertising delivery graph is created by extracting the key points related to the delivery business in the advertising delivery sub-process; merging the same key points to obtain an advertising delivery graph, wherein the nodes in the advertising delivery graph include node identifiers and node values ​​representing the advertising revenue of the nodes.

[0014] In some examples, the above-mentioned advertising delivery links include multiple links, and the above-mentioned method further includes: using a preset compression method to compress each advertising delivery link in the multiple advertising delivery links, and displaying the compressed multiple advertising delivery links.

[0015] In some examples, the above method further includes: determining an advertisement delivery link to be executed from multiple advertisement delivery links according to the received link selection operation.

[0016] In some examples, the above method also includes: determining a supplementary appeal based on the received supplementary appeal operation; determining an updated target node from the advertising delivery graph based on the advertising appeal and the supplementary appeal; and determining an advertising delivery link between the updated target nodes from the advertising delivery graph.

[0017] In some examples, the above method also includes: determining creative materials in the advertising creative submitted by the user; determining similar materials of the creative materials from the second knowledge base through the advertising review intelligent agent; generating new materials based on the similar materials through the advertising review intelligent agent; and determining the advertising review results based on the similarity between the new materials and the advertising creatives of the advertising users in the advertising user set.

[0018] In some examples, determining similar materials of the creative material from the second knowledge base includes: determining a first material feature of the creative material; and determining the similar material based on a similarity between the first material feature and a second material feature of a material in the second knowledge base.

[0019] In some examples, the second knowledge base includes an advertising knowledge base and a brand knowledge base, and the determining of similar materials based on the similarity between the first material feature and the second material feature of the material in the second knowledge base includes: determining a plurality of similar materials ranked at the top in similarity based on the similarity between the first material feature and the second material feature of the material in the advertising knowledge base; and determining a plurality of similar materials ranked at the top in similarity and having a similarity exceeding a preset similarity threshold based on the similarity between the first material feature and the second material feature of the material in the brand knowledge base.

[0020] In some examples, the above method also includes: determining the third material feature of the creative material in the historical advertising creative submitted by the user; fusing the first material feature and the third material feature to obtain a fusion feature; and the above-mentioned generating new materials based on similar materials includes: generating new materials based on the second material feature and the fusion feature corresponding to the similar materials.

[0021] In some examples, the above method also includes: adjusting the preset similarity threshold according to the operation status of the advertising engine.

[0022] In some examples, obtaining the user's advertising demands includes obtaining the user's advertising demands represented by natural language.

[0023] In some examples, the above method also includes: determining the agent at each level in the multi-level agent based on the received selection operation.

[0024] In the second aspect, an embodiment of the present application provides an information processing device applied to an advertising engine, including: an acquisition unit, configured to acquire the user's advertising demands; a determination unit, configured to determine a set of executable policies that meet the advertising demands through multi-level intelligent agents, wherein each level of the multi-level intelligent agents can determine the executable policy set corresponding to the advertising demands, and the executable policy set includes executable policies corresponding to each of the multiple sub-processes in the entire advertising process; an execution unit, configured to execute the executable policies corresponding to each of the multiple sub-processes in the executable policy set through the business interfaces corresponding to each of the multiple sub-processes.

[0025] In some examples, the agents at each level in the multi-level agent include multiple agents, and the above-mentioned determination unit is further configured to: process multiple sub-appeals in the advertising appeal in the order of execution of the agents at each level in the multi-level agent, and in response to the presence of a target sub-appeal among the multiple sub-appeals for which the agents at the current level have not determined an executable strategy based on the preset voting strategy, process the target sub-appeal through the agents at subsequent levels until the executable strategies corresponding to the multiple sub-appeals are combined to obtain an executable strategy set.

[0026] In some examples, the first-level agents in the multi-level agents include agent groups corresponding to multiple sub-processes, each agent group includes multiple agents, and the above-mentioned determination unit is further configured to: decompose user demands and determine the sub-demands corresponding to each of the multiple sub-processes; process the sub-demands corresponding to each of the multiple sub-processes through the agent groups corresponding to each of the multiple sub-processes to obtain multiple candidate executable strategies corresponding to each of the multiple sub-processes; for the multiple candidate executable strategies output by each of the multiple agent groups, determine whether an executable strategy can be determined from the multiple candidate executable strategies based on a preset voting strategy; in response to each of the multiple agent groups, the executable strategy corresponding to the agent group can be determined, and the executable strategies corresponding to each of the multiple agent groups are combined to obtain an executable strategy set.

[0027] In some examples, the above-mentioned determination unit is further configured to: in response to determining that there is a target intelligent agent group among multiple intelligent agent groups for which an executable strategy has not been determined, process the target sub-demand through multiple intelligent agents at subsequent levels until an executable strategy is determined from the candidate executable strategies output by each of the multiple intelligent agents at the current level based on a preset voting strategy, wherein the multiple intelligent agents at subsequent levels are determined according to the execution order; and obtain an executable strategy set by combining the executable strategies determined by other intelligent agent groups except the target intelligent agent group among the multiple intelligent agent groups and the executable strategies determined by the multiple intelligent agents at the current level.

[0028] In some examples, the above-mentioned determination unit is further configured to: parse the sub-claims corresponding to the multiple sub-processes through the intelligent agent groups corresponding to the multiple sub-processes, and obtain the advertising terms corresponding to the multiple sub-claims; for each intelligent agent group in the multiple intelligent agent groups, determine the candidate executable strategy corresponding to the sub-process corresponding to the intelligent agent group from the preset limited result set according to the advertising terms corresponding to the sub-claims to be processed by the intelligent agent group.

[0029] In some examples, the above-mentioned multi-level intelligent agents are obtained in the following manner: determining the initial models corresponding to each multi-level intelligent agent based on the accuracy and parameter scale of the models in the model set, wherein the level to which the intelligent agent belongs and the accuracy and parameter scale of the initial model corresponding to the level are positively correlated; determining the first knowledge base corresponding to each multi-level intelligent agent, wherein the first knowledge base corresponding to intelligent agents at different levels includes advertising knowledge of different preset dimensions; training the initial models corresponding to each multi-level intelligent agent through the first knowledge base corresponding to each multi-level intelligent agent to obtain the multi-level intelligent agent.

[0030] In some examples, multiple sub-processes include an advertising creative sub-process, an advertising delivery sub-process, an advertising operation sub-process, and an advertising playback sub-process, and the above-mentioned execution unit is further configured to: generate advertising creatives through the business interface corresponding to the advertising creative sub-process and according to the executable policy corresponding to the advertising creative sub-process in the executable policy set; execute the delivery process of the advertising creatives through the business interface corresponding to the advertising delivery sub-process and according to the executable policy corresponding to the advertising delivery sub-process in the executable policy set; execute the algorithms involved in the advertising process through the business interface corresponding to the advertising operation sub-process and according to the executable policy corresponding to the advertising operation sub-process in the executable policy set; execute the playback process of the advertising creatives through the business interface corresponding to the advertising playback sub-process and according to the executable policy corresponding to the advertising playback sub-process in the executable policy set.

[0031] In some examples, the above-mentioned execution unit is further configured to: determine the target nodes existing in the advertising delivery graph from the advertising appeal, wherein the advertising delivery graph represents all advertising links in the advertising delivery process; determine the advertising delivery links between the target nodes from the advertising delivery graph; and adopt the executable strategy corresponding to the advertising delivery sub-process to execute the advertising creative delivery process according to the advertising delivery link.

[0032] In some examples, the execution unit is further configured to: use a depth-first traversal algorithm to determine the advertising delivery links between target nodes from the advertising delivery graph.

[0033] In some examples, the above-mentioned advertising delivery graph is created by extracting the key points related to the delivery business in the advertising delivery sub-process; merging the same key points to obtain an advertising delivery graph, wherein the nodes in the advertising delivery graph include node identifiers and node values ​​representing the advertising revenue of the nodes.

[0034] In some examples, the above-mentioned advertising delivery links include multiple links, and the above-mentioned device also includes: a compression display unit, configured to use a preset compression method to compress each of the multiple advertising delivery links and display the compressed multiple advertising delivery links.

[0035] In some examples, the apparatus further includes: a selection unit configured to determine an advertisement delivery link to be executed from a plurality of advertisement delivery links according to a received link selection operation.

[0036] In some examples, the above-mentioned device also includes: a supplementing unit, configured to determine a supplementary appeal based on the received supplementary appeal operation; determine an updated target node from the advertising delivery graph based on the advertising appeal and the supplementary appeal; and determine an advertising delivery link between the updated target nodes from the advertising delivery graph.

[0037] In some examples, the above-mentioned device also includes: a review unit, configured to determine the creative materials in the advertising creative submitted by the user; determine similar materials of the creative materials from the second knowledge base through the advertising review intelligent agent; generate new materials based on the similar materials through the advertising review intelligent agent; determine the advertising review result based on the similarity between the new materials and the advertising creative of the advertising users in the advertising user set.

[0038] In some examples, the review unit is further configured to: determine a first material feature of the creative material; and determine similar materials based on a similarity between the first material feature and a second material feature of a material in the second knowledge base.

[0039] In some examples, the second knowledge base includes an advertising knowledge base and a brand knowledge base, and the review unit is further configured to: determine a plurality of similar materials ranked at the top in similarity based on the similarity between the first material feature and the second material feature of the material in the advertising knowledge base; determine a plurality of similar materials ranked at the top in similarity and having a similarity exceeding a preset similarity threshold based on the similarity between the first material feature and the second material feature of the material in the brand knowledge base.

[0040] In some examples, the above-mentioned device also includes: a fusion unit, configured to determine the third material feature of the creative material in the historical advertising creative submitted by the user; fuse the first material feature and the third material feature to obtain a fusion feature; and the above-mentioned review unit, further configured to: generate a new material based on the second material feature and the fusion feature corresponding to similar materials.

[0041] In some examples, the apparatus further includes an adjustment unit configured to adjust a preset similarity threshold according to an operation status of the advertising engine.

[0042] In some examples, the acquisition unit is further configured to: acquire the user's advertising appeal represented by natural language.

[0043] In some examples, the above-mentioned device also includes: an agent determination unit, configured to determine the agent of each level in the multi-level agent according to the received selection operation.

[0044] In a third aspect, an embodiment of the present application provides a computer-readable medium having a computer program stored thereon, wherein when the program is executed by a processor, the method described in any implementation manner of the first aspect is implemented.

[0045] In a fourth aspect, an embodiment of the present application provides an electronic device comprising: one or more processors; a storage device on which one or more programs are stored, and when the one or more programs are executed by one or more processors, the one or more processors implement the method described in any implementation manner of the first aspect.

[0046] The information processing method and device provided in the embodiments of the present application are applied to the advertising engine by obtaining the user's advertising demands; determining an executable policy set that meets the advertising demands through a multi-level intelligent agent, wherein the intelligent agent at each level in the multi-level intelligent agent can determine the executable policy set corresponding to the advertising demands, and the executable policy set includes executable policies corresponding to multiple sub-processes in the entire advertising process; executing the executable policies corresponding to multiple sub-processes in the executable policy set through the business interfaces corresponding to the multiple sub-processes, thereby providing an information processing method applied to the advertising engine, wherein the multi-level intelligent agent in the advertising engine can automatically process the entire advertising process based on the user's advertising demands, lowering the participation threshold of the advertising business and improving the user experience and convenience in the advertising processing process. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Other features, objects and advantages of the present application will become more apparent upon reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0048] Figure 1 is an exemplary system architecture diagram in which an embodiment of the present application may be applied;

[0049] Figure 2 is a flowchart of an embodiment of an information processing method applied to an advertising engine according to the present application;

[0050] Figure 3 is a schematic diagram of the structure of a multi-level intelligent agent according to this embodiment;

[0051] Figure 4 is a schematic diagram of an application scenario of the information processing method applied to an advertising engine according to this embodiment;

[0052] Figure 5 is a flowchart of another embodiment of an information processing method applied to an advertising engine according to the present application;

[0053] Figure 6 is a schematic diagram of an advertisement placement diagram according to this embodiment;

[0054] Figure 7 is a schematic diagram of a user interface of multiple compressed advertisement delivery links according to this embodiment;

[0055] Figure 8 is a flowchart of another embodiment of an information processing method applied to an advertising engine according to the present application;

[0056] Figure 9 is a structural diagram of an embodiment of an information processing device applied to an advertising engine according to the present application;

[0057] Figure 10 It is a structural diagram of a computer system suitable for implementing the embodiments of the present application. DETAILED DESCRIPTION

[0058] The present application will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the relevant invention and are not intended to limit the invention. It should also be noted that, for ease of description, only portions relevant to the relevant invention are shown in the accompanying drawings.

[0059] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0060] It should be noted that the collection, collection, updating, analysis, processing, use, transmission, and storage of user personal information involved in the technical solutions disclosed herein all comply with relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken with respect to user personal information to prevent unauthorized access to user personal information data and to safeguard the security of user personal information, network security, and national security.

[0061] Figure 1 An exemplary architecture 100 is shown to which the information processing method and apparatus for an advertisement engine of the present disclosure may be applied.

[0062] like Figure 1 As shown, system architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. The communication connections between terminal devices 101, 102, and 103 constitute a topological network, and network 104 is used to provide a medium for communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0063] The terminal devices 101, 102, 103 can interact with the server 105 via the network 104 to receive or send data, etc. The terminal devices 101, 102, 103 can be hardware devices or software that support network connection for data interaction and data processing. When the terminal devices 101, 102, 103 are hardware, they can be various electronic devices that support network connection, information acquisition, interaction, display, processing and other functions, including but not limited to smart phones, car computers, tablet computers, e-book readers, laptop computers and desktop computers, etc. When the terminal devices 101, 102, 103 are software, they can be installed in the electronic devices listed above. They can be implemented as multiple software or software modules for providing distributed services, for example, or they can be implemented as a single software or software module. No specific limitation is made here.

[0064] Server 105 can be a server that provides various services, for example, a backend processing server that obtains user advertising demands through terminal devices 101, 102, and 103, and generates and executes an executable policy set that satisfies the advertising demands through a multi-level intelligent agent. As an example, server 105 can be a cloud server.

[0065] It should be noted that the server can be either hardware or software. When the server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the server is software, it can be implemented as multiple software or software modules (e.g., software or software modules for providing distributed services), or as a single software or software module. No specific limitations are given here.

[0066] It should also be noted that the information processing method for an advertising engine provided in the embodiments of the present application can be executed by a server, a terminal device, or a server and a terminal device in cooperation with each other. Accordingly, the various components (e.g., various units) included in the information processing device for an advertising engine can be entirely located in a server, entirely located in a terminal device, or separately located in a server and a terminal device.

[0067] It should be understood that Figure 1 The number of terminal devices, networks, and servers in the system is merely illustrative. Any number of terminal devices, networks, and servers may be provided as needed. When the electronic device on which the information processing method applied to the advertising engine is running does not need to transmit data with other electronic devices, the system architecture may only include the electronic device (e.g., a server or terminal device) on which the information processing method applied to the advertising engine is running.

[0068] Continue to refer Figure 2, shows a process 200 of an embodiment of an information processing method applied to an advertising engine, including the following steps:

[0069] Step 201: Obtain the user's advertising demands.

[0070] In this embodiment, the execution subject of the information processing method applied to the advertisement engine (eg Figure 1 The terminal device or server in the network can obtain the user's advertising demands remotely or locally through a wired network connection or a wireless network connection.

[0071] Users can input advertising appeals into the advertising engine based on advertising terms and / or non-advertising terms. For example, the advertising engine uses a public web service to provide a front-end page and back-end server, and provides login and ad placement functions. Users (e.g., merchants) log in to the advertising engine using their advertising platform account and enter the ad placement function page. The front-end page supports the input function of advertising appeals to support users to enter advertising appeals.

[0072] Advertising appeals can be represented in the form of text, voice, pictures, videos, etc. The advertising engine supports the above multiple input forms for advertising appeals.

[0073] The advertising engine also supports multiple entry of advertising appeals. It provides a separate supplementary entry, allowing users to modify or supplement the initial advertising appeal. The advertising engine modifies the version number and entry time of the initial advertising appeal to determine and process the complete advertising appeal. It should be noted that supplementary appeals obtained through the supplementary entry have a higher priority than the initial advertising appeal, and can be prioritized by sequence number or weight level.

[0074] In some optional implementations of this embodiment, the execution entity may perform step 201 in the following manner: obtaining the user's advertising demands represented by natural language.

[0075] In real life, advertising is extremely complex, and many users lack understanding of the business. Expressing advertising requests in natural language better meets their needs. In practical applications, users can enter advertising requests through a multi-round dialogue. The execution entity can maintain contextual information by introducing a memory network or a long short-term memory (LSTM) network to ensure a comprehensive understanding of the user's question or request.

[0076] For example, the user enters the following advertising appeal: I want to use a budget of 5,000 yuan to promote the product with sku=123, and I hope to have 300 exposures.

[0077] At this time, the record of the above advertising appeal in the database is: id=1, lv=1, v1=Use a budget of 5,000 yuan to promote the product sku=123, and I hope to have 300 exposures.

[0078] The user can make modifications: With a budget of 5,000 yuan, I want to promote product sku=123 and get 600 exposures.

[0079] At this time, the record in the database is: id=1, lv=1, v2=using a budget of 5,000 yuan to promote the product sku=123, and I hope to have 300 exposures.

[0080] The user can add an entry: 20 more conversions are desired.

[0081] The database record at this time is: id = 1, lv = 1, v2 = Use a budget of 5,000 yuan to promote product sku = 123, and I hope to have 300 impressions;

[0082] id=1, lv=2, v1=20 more conversions are expected.

[0083] As in the example above, to ensure the feasibility of user demands, user demands generally include at least two types of demand information: advertising budget and advertising delivery results. The delivery target refers to the revenue that the advertiser hopes to obtain from this advertising.

[0084] A campaign objective can be a performance categorization, such as achieving 1,000 impressions, 500 clicks, or 20 conversions. It can also be a business categorization, such as promoting a store, promoting a product, or promoting a campaign page. It can also be a more vague description, such as achieving good off-site results or focusing on attracting new customers. A campaign objective can be one of these or a combination of multiple. A budget refers to the cost an advertiser can afford, primarily consisting of a dollar amount and a timeframe.

[0085] Step 202: Determine an executable strategy set that meets the advertising demands through multi-level intelligent agents.

[0086] In this embodiment, the execution subject can determine an executable strategy set that meets the advertising requirements through a multi-level intelligent agent. Each level of the multi-level intelligent agent can determine an executable strategy set corresponding to the advertising requirements. The executable strategy set includes executable strategies corresponding to multiple sub-processes in the entire advertising process.

[0087] Each layer of the multi-layered agent includes at least one agent. Examples of these agents include neural network models such as recurrent neural network models and residual neural network models, and large language models such as GPT (Generative Pre-Trained Transformer) and BERT (Bidirectional Encoder Representations from Transformers). Agents at different layers can use the same or different model structures.

[0088] In this embodiment, the process of generating a multi-level agent is as follows:

[0089] First, the specific problems to be solved by each of the multiple sub-processes within the overall advertising pipeline are defined, such as ad text generation, ad creative development, ad effectiveness prediction, and user behavior analysis. Next, advertising data relevant to the defined problems is collected, including ad copy, user comments, and social media content across various ad types and themes. The collected advertising data is then preprocessed and cleaned, including text segmentation, stop word removal, and handling of missing and outliers to ensure data quality. A knowledge graph is then constructed based on the resulting cleansed advertising knowledge, including advertising industry terminology, key concepts, and brand information. This helps the model better understand ad content. Next, large language models suitable for natural language processing tasks are selected, such as GPT-3 and BERT. These models excel in both understanding and generating data. Finally, the selected large language models are fine-tuned using data from the advertising domain to better adapt them to the specific needs of each advertising task within the multiple sub-processes.

[0090] In this embodiment, the execution entity can determine the set of executable policies that meet the advertising demand through any agent in the multi-level agent layer. For example, an agent in one level of the multi-level agent layer can be randomly selected to handle the advertising demand, or an agent in one level of the multi-level agent layer can be selected to handle the advertising demand using a load balancing strategy.

[0091] For example, the multiple sub-processes include an ad creative sub-process, an ad placement sub-process, an ad operations sub-process, and an ad playback sub-process. The ad creative sub-process is used to generate ad creatives that meet advertising needs; the ad placement sub-process guides the ad creative placement process; the ad operations sub-process determines the data processing algorithms involved in the entire ad process; and the ad playback sub-process guides the ad creative playback process.

[0092] The executable strategies of the advertising delivery sub-process mainly include strategies concerned with the advertising delivery business, including advertising delivery, financial settlement, report viewing, etc.

[0093] The executable strategies of the advertising creative sub-process mainly include the strategies concerned with advertising creative production, including advertising plan / unit / creative generation, live broadcast / short video / graphics and other content processing, SKU (Stock Keeping Unit, minimum inventory unit) and other product processing, URL (uniform resource locator) and other landing page processing, etc.

[0094] The executable strategies of the advertising operation sub-process mainly include strategies focused on by advertising data and algorithms, including advertising models, logs, follow-up, anti-fraud, etc.

[0095] The executable strategies of the ad playback sub-process mainly include the strategies concerned with ad playback, including user portrait, adserver (advertising server), userserver (user server), ad recall, gateway, etc.

[0096] As an example, the user's advertising demand is: use a budget of 5,000 yuan to promote the product of sku=123, I hope to have 300 exposures and 20 conversions. Among them, I require the promotion to be targeted at plus members, and do not want it to appear outside the platform. The total run time is 3 days.

[0097] The executable strategies corresponding to the multiple sub-processes are:

[0098] Advertising placement sub-process: merchant pin, budget: 5,000 yuan, promotion plan: 3 days, promotion type: product promotion; automatically complete financial settlement requirements and reporting requirements.

[0099] Ad creative sub-process: SKU: 123, automatically complete the landing page URL, product main image, etc.

[0100] Advertising operation sub-process: Exposure requirement: 300, Conversion requirement: 20

[0101] Advertisement playback sub-process: Targeting requirements: Plus membership, Ad placement requirements: Platform site

[0102] In some optional implementations of this embodiment, each level of the multi-level agent includes multiple agents. The multiple agents may be agents with the same model structure or agents with different model structures.

[0103] The above-mentioned execution entity can execute the above-mentioned step 202 in the following manner: according to the execution order of the intelligent agents at each level in the multi-level intelligent agents, process multiple sub-appeals in the advertising appeal, and in response to the presence of a target sub-appeal among the multiple sub-appeals for which the intelligent agent at the current level has not determined an executable strategy based on the preset voting strategy, process the target sub-appeal through the intelligent agents at subsequent levels until the executable strategies corresponding to the multiple sub-appeals are combined to obtain an executable strategy set.

[0104] As an example, the multi-level agent includes three levels of agents, namely the first level, the second level and the third level, and the execution order is the first level, the second level and the third level.

[0105] Continue to refer Figure 3 , shows a schematic structural diagram of a multi-level intelligent agent, including multiple intelligent agent groups 301 at the first level, multiple intelligent agents 302 at the second level and multiple intelligent agents 303 at the third level.

[0106] For each sub-appeal in the advertising appeal, it is necessary to determine the executable strategy corresponding to the sub-appeal, so as to combine the executable strategies corresponding to multiple sub-appeals to obtain an executable strategy set.

[0107] The preset voting strategy is used to determine the executable strategy in the voting result set for each sub-demand from the candidate executable strategies output by multiple intelligent agents at one level, such as the unanimous recognition strategy, the majority strategy, etc.

[0108] Multiple intelligent agents at the first level generate candidate executable strategies based on each sub-demand in the advertising appeal and vote on multiple candidate executable strategies; for each sub-appeal, an executable strategy is determined from multiple candidate executable strategies corresponding to the sub-appeal and the subsequent executable strategy processing process is terminated.

[0109] In response to the fact that there is a target sub-demand among multiple sub-demands for which multiple intelligent agents at the first level are unable to determine an executable strategy based on a preset voting strategy, multiple intelligent agents at the second level output multiple candidate executable strategies for the target sub-demand, and vote on the multiple candidate executable strategies. In response to determining an executable strategy from them, the subsequent executable strategy processing process is terminated.

[0110] In response to the inability of the multiple agents at the second level to determine an executable strategy for the target sub-demand based on the preset voting strategy, the multiple agents at the third level output multiple candidate executable strategies for the target sub-demand and determine an executable strategy from among them. Because the third level is the bottom layer, an executable strategy for the target sub-demand must be determined.

[0111] In this implementation, multi-level intelligent agents have an execution order. Based on the execution order, the set of executable strategies for the current level is determined through a preset voting strategy. The multi-level intelligent agents cooperate with the voting mechanism to help alleviate the model hallucination problem and improve the response speed of the intelligent agents, as well as the orderliness and efficiency of the intelligent agents in the data processing process.

[0112] In some optional implementations of this embodiment, continue to refer to Figure 3 The first level of the multi-level agent includes multiple agent groups corresponding to each sub-process, and each agent group includes multiple agents. For example, the advertising creative sub-process, the advertising placement sub-process, the advertising operation sub-process, and the advertising playback sub-process each have an agent group corresponding to them.

[0113] In this implementation, the execution subject may perform the process of determining the executable policy set in the following manner:

[0114] First, break down user demands and determine the sub-demands corresponding to multiple sub-processes.

[0115] The user's input demand is broken down to identify the demand segments (sub-demands) belonging to the advertising creative sub-process, advertising delivery sub-process, advertising operation sub-process, and advertising playback sub-process. For each sub-demand corresponding to a sub-process, it is distributed to the agent group under the corresponding sub-process for processing.

[0116] Second, through the agent groups corresponding to the multiple sub-processes, the sub-demands corresponding to the multiple sub-processes are processed to obtain multiple candidate executable strategies corresponding to the multiple sub-processes.

[0117] For each sub-process in the multiple sub-processes, multiple agents in the agent group corresponding to the sub-process respectively process the sub-demands corresponding to the sub-process to obtain multiple candidate executable strategies corresponding to the sub-process.

[0118] Third, for the multiple candidate executable strategies output by the multiple agent groups, determine whether an executable strategy can be determined from the multiple candidate executable strategies based on the preset voting strategy.

[0119] For each of the multiple agent groups, based on a preset voting strategy, vote on the candidate executable strategies output by each of the multiple agents in the current agent group to determine whether an executable strategy can be determined from the multiple candidate executable strategies.

[0120] Fourth, in response to each of the multiple agent groups, an executable strategy corresponding to the agent group can be determined, and the executable strategies corresponding to the multiple agent groups are combined to obtain an executable strategy set.

[0121] When multiple agent groups can determine corresponding executable strategies, the executable strategies corresponding to the multiple agent groups can be combined to obtain an executable strategy set.

[0122] In this implementation, a method for determining an executable strategy set based on a first-level intelligent agent is provided, and multiple sub-processes are assigned to corresponding intelligent agent groups, each performing their respective duties, further improving the efficiency and accuracy of determining the executable strategy set.

[0123] In some optional implementations of this embodiment, the execution subject may perform the process of determining the executable policy set in the following manner:

[0124] First, in response to determining that there is a target agent group among multiple agent groups for which an executable strategy has not been determined, multiple agents at subsequent levels process the target sub-demands until an executable strategy is determined from the candidate executable strategies output by each of the multiple agents at the current level based on a preset voting strategy.

[0125] Among them, multiple agents in subsequent layers are determined according to the execution order.

[0126] In this implementation, agents in subsequent layers can perform the functions of agents in the previous layer. Continuing with the three-layer example, agents in the second and third layers both have the ability to process sub-demands and determine executable strategies for each of the agent groups corresponding to the multiple sub-processes in the first layer.

[0127] For example, the target agent group for which no executable strategy has been determined among the multiple agent groups is the agent group corresponding to the advertising creative sub-process. In this case, the sub-demands corresponding to the advertising creative sub-process are input into the multiple agents at the second level, and a vote is performed on the multiple candidate executable strategies output to determine an executable strategy. Otherwise, the sub-demands corresponding to the advertising creative sub-process are input into the multiple agents at the third level to determine an executable strategy.

[0128] Then, the executable strategies determined by the other agent groups except the target agent group in the multiple agent groups and the executable strategies determined by the multiple agents at the current level are combined to obtain an executable strategy set.

[0129] Continuing with the above example, we combine the executable strategies corresponding to the advertising placement sub-process, advertising operation sub-process, and advertising playback sub-process in the first level, and the executable strategies corresponding to the advertising creative sub-process determined by the intelligent agent in the second level to obtain an executable strategy set.

[0130] In this implementation, a method for determining an executable policy set by combining multiple levels of intelligent agents is provided, which further improves the efficiency and accuracy of determining the executable policy set.

[0131] In some optional implementations of this embodiment, the execution entity may perform the second step in the following manner:

[0132] First, through the intelligent agent groups corresponding to the multiple sub-processes, the sub-demands corresponding to the multiple sub-processes are analyzed to obtain the advertising terms corresponding to the multiple sub-demands.

[0133] In this implementation, the execution entity determines the advertising terms corresponding to each of the multiple sub-appeals based on the following process: (1) Using natural language processing technology, the advertising appeals proposed by the user are understood. Natural language understanding includes steps such as lexical analysis, grammatical analysis, and semantic analysis to extract key information from the advertising appeals. (2) Identify entities related to the advertising field from the advertising appeals, such as brand names, product types, advertising targets, etc. (3) Perform sentiment analysis on the user's language to understand the user's emotional tendencies in order to better understand their needs and expectations. (4) Based on the results of natural language understanding, entity recognition, and sentiment analysis, the advertising appeals are parsed into advertising terms.

[0134] An ad's appeal may contain parts not covered by the aforementioned sub-processes. In these cases, the automated engine can handle these parts to fully and comprehensively analyze the user's ad appeal. The automated engine provides intelligent advertising services. After the user sets the necessary logic for the ad chain, the algorithmic decision-making mechanism helps the user complete the ad action.

[0135] Continuing with the above example, the user's advertising appeal is:

[0136] With a budget of 5,000 yuan, I want to promote product sku=123 and get 300 impressions and 20 conversions. I want to promote it to plus members only and don’t want it to appear outside the platform. The promotion will last for 3 days in total.

[0137] The parsed advertising terms are:

[0138] Merchant pin, budget: 5,000 yuan, sku: 123, promotion type: product promotion, exposure requirement: 300, conversion requirement: 20, targeting requirement: plus member, advertising position requirement: within the platform, promotion plan: 3 days.

[0139] Then, for each of the multiple agent groups, based on the advertising terms corresponding to the sub-appeal to be processed by the agent group, a candidate executable strategy corresponding to the sub-process of the agent group is determined from the preset limited result set.

[0140] The restricted result set includes all executable policies involved in the advertising process. Different sub-processes can have different restricted result sets. The advertising engine has management functions for restricted result sets, allowing relevant technical personnel to add, delete, modify, and query results in the restricted result set.

[0141] Continuing with the above example, based on the advertising terms corresponding to the multiple sub-appeals, the executable strategies corresponding to the multiple sub-processes can be determined as follows:

[0142] Advertising placement sub-process: merchant pin, budget: 5,000 yuan, promotion plan: 3 days, promotion type: product promotion; automatically complete financial settlement requirements and reporting requirements.

[0143] Advertising creative sub-process: sku: 123, automatically complete the landing page URL, product main image, etc.

[0144] Advertising operation sub-process: Exposure requirement: 300, Conversion requirement: 20

[0145] Advertisement playback sub-process: Targeting requirements: Plus membership, Ad placement requirements: Platform site

[0146] In this implementation, the advertising terms of the sub-appeal are first determined and then the executable strategy is determined based on the advertising terms. The output executable strategy is limited by limiting the result set, which further improves the accuracy of the executable strategy and the adaptability of the executable strategy to the sub-process.

[0147] In some optional implementations of this embodiment, the above-mentioned execution subject can obtain a multi-level intelligent agent in the following manner:

[0148] First, the initial models corresponding to each multi-level intelligent agent are determined based on the accuracy and parameter scale of the models in the model set.

[0149] Among them, there is a positive correlation between the level to which the intelligent agent belongs and the accuracy and parameter scale of the initial model corresponding to the level.

[0150] The model collection includes a variety of models suitable for processing advertising tasks. According to the accuracy of the model, the model can be divided into low-precision models, half-precision models and full-precision models; according to the parameter scale of the model, the model can be divided into low-parameter scale models, medium-parameter scale models and high-parameter scale models.

[0151] Continuing with the example of three-level intelligent agents, the first-level intelligent agents tend to choose half-precision, low-parameter-scale models; the second-level intelligent agents tend to choose half / full-precision, medium-parameter-scale models; and the third-level intelligent agents tend to choose full-precision, high-parameter-scale models.

[0152] Then, the first knowledge base corresponding to each of the multi-level intelligent agents is determined.

[0153] Among them, the first knowledge base corresponding to intelligent agents at different levels includes advertising knowledge of different preset dimensions.

[0154] As an example, the preset dimensions include the advertising industry, advertising companies, and advertising business lines within companies. The advertising industry, advertising companies, and advertising business lines correspond to the first-level intelligent agents, the second-level intelligent agents, and the third-level intelligent agents, respectively.

[0155] The advertising knowledge corresponding to the advertising industry is the general knowledge of the advertising industry. For example, billing models include CPM (Cost Per Mille, cost per thousand impressions), CPC (Cost Per Click, cost per click), CPT (Cost Per Time, time cost), CPS (Cost Per Sale, cost per sale), etc.; the advertising knowledge of advertising companies is the general knowledge of the advertising business within the advertising companies. For example, the advertising business of an advertising company includes search advertising express, recommended advertising shopping touchpoints, etc.; the advertising knowledge of advertising business lines is the knowledge related to specific advertising business lines in advertising companies, such as the express advertising plan production method, etc.

[0156] The aforementioned execution entity can provide a hierarchical storage function for the first knowledge base, which can be stored in various public databases (e.g., relational databases, non-relational databases, object-oriented databases). For each piece of knowledge, the database must contain two fields: question and answer, and preferably include the contribution channel and the time of writing of the stored knowledge.

[0157] The first knowledge base has a management backend, where the corresponding knowledge base administrator enters the relevant knowledge and stores it in the first knowledge base's corresponding database. This allows the administrator to manage existing knowledge. For example, a knowledge expiration management function can be added to set an expiration date for each piece of stored knowledge, ensuring that the first knowledge base only provides valid knowledge.

[0158] The first knowledge base has a data synchronization function, which can synchronize data from other database systems to the first knowledge base database to enrich the first knowledge base. In addition, to further enhance the richness of the knowledge in the first knowledge base, advertising-related knowledge data generated by multi-level intelligent agents can be transmitted back to the first knowledge base for storage.

[0159] Finally, the initial models corresponding to the multi-level intelligent agents are trained through the first knowledge bases corresponding to the multi-level intelligent agents to obtain the multi-level intelligent agents.

[0160] In this implementation, a specific method for obtaining a multi-level intelligent agent is provided, which improves the accuracy of the multi-level intelligent agent.

[0161] In some optional implementations of this embodiment, the above-mentioned execution subject may also perform the following operation: determining a multi-level intelligent agent for processing the advertising appeal based on the received selection operation.

[0162] For example, for each of the multiple levels, a user can select at least one agent in the level from the agent set. The agent set includes multiple agents, each of which can independently handle advertising tasks for the entire advertising process, or the agents can cooperate with each other to handle advertising tasks for the entire advertising process.

[0163] In this implementation, users can select intelligent agents at each level based on their own needs, which helps to further improve the user experience and the matching degree between multi-level intelligent agents and users.

[0164] Step 203 : Execute the executable policies corresponding to the multiple sub-processes in the executable policy set through the business interfaces corresponding to the multiple sub-processes.

[0165] In this embodiment, the execution subject may execute the executable policies corresponding to the multiple sub-processes in the executable policy set through the business interfaces corresponding to the multiple sub-processes.

[0166] For example, for each of the multiple sub-processes, the execution entity can transmit the executable policy corresponding to the sub-process to the business interface corresponding to the sub-process. The business interface processes the sub-process according to the execution plan represented by the executable policy. The business interfaces corresponding to the multiple sub-processes cooperate with each other to complete the advertising process.

[0167] In some optional implementations of this embodiment, the execution entity may perform step 203 as follows:

[0168] Generate advertising creatives through the business interface corresponding to the advertising creative sub-process and according to the executable strategy corresponding to the advertising creative sub-process in the executable strategy set.

[0169] Delivery is the most complex part of the advertising process. The business interfaces corresponding to the creative sub-process must include at least three types of interfaces: ad delivery, report viewing, and financials. These interfaces can be either remote procedure call (RPC) or browser call (HTTP). For RPC, a server is required; for browser call, a server is required.

[0170] The advertising delivery interface should provide the following functions: plan making, business selection, and association designation with other businesses; the report viewing interface should provide the following functions: report display; the financial interface should provide the following functions: deductions, account and financial information maintenance.

[0171] Through the business interface corresponding to the advertising delivery sub-process, the advertising creative delivery process is executed according to the executable strategy corresponding to the advertising delivery sub-process in the executable strategy set.

[0172] Advertising creativity mainly adopts a generative approach, that is, providing a creative generation interface, receiving user-specified creative generation demands, such as types (pictures, videos, etc.), and using executable strategies corresponding to the creative generation demands to generate advertising creativity.

[0173] Through the business interface corresponding to the advertising operation sub-process, according to the executable strategy corresponding to the advertising operation sub-process in the executable strategy set, the algorithms involved in the advertising process are executed.

[0174] The business interface corresponding to the advertising operation sub-process mainly provides the specification of models and algorithms. For example, users can choose exposure priority, add-to-cart priority, conversion priority, etc.; they can also choose different types of DMP (Data Management Platform) audience targeting.

[0175] Through the business interface corresponding to the advertisement playing sub-process, the advertisement creative playing process is executed according to the executable strategy corresponding to the advertisement playing sub-process in the executable strategy set.

[0176] The business interface corresponding to the ad playback sub-process mainly provides the specification of the playback link, such as portrait selection, recall sorting logic, etc.

[0177] In this embodiment, each sub-process may correspond to multiple business interfaces, and there is a correspondence between executable policies and business interfaces. The execution entity, or an electronic device communicatively connected to the execution entity, is provided with a policy interface dictionary to determine the business interface that each executable policy matches. Thus, the policy interface dictionary maps executable policies to interface calls. For all executable policies matching business interfaces, an execution order and priority must be established that aligns with the delivery service.

[0178] For policies that do not have a matching interface, they are specially marked so that the business interface that can execute the policy can be completed through the automatic engine.

[0179] In this implementation, a specific execution method of an executable policy set is provided, which improves the orderliness and efficiency of the executable policies during execution.

[0180] Continue to see Figure 4 , Figure 4FIG4 is a schematic diagram 400 of an application scenario of the data processing method according to this embodiment. Figure 4 In the application scenario of FIG, user 401 sends an advertising request to server 403 through terminal device 402. Server 403 first obtains the user's advertising request; then, through a multi-level intelligent agent, it determines an executable policy set that satisfies the advertising request. The intelligent agent at each level of the multi-level intelligent agent can determine an executable policy set corresponding to the advertising request. The executable policy set includes executable policies corresponding to multiple sub-processes in the entire advertising process; finally, through the business interfaces corresponding to the multiple sub-processes, the executable policies corresponding to the multiple sub-processes in the executable policy set are executed.

[0181] The method provided by the above-mentioned embodiments of the present application obtains the user's advertising demands; determines an executable policy set that meets the advertising demands through a multi-level intelligent agent, wherein the intelligent agent at each level in the multi-level intelligent agent can determine the executable policy set corresponding to the advertising demands, and the executable policy set includes executable policies corresponding to multiple sub-processes in the entire advertising process; executes the executable policies corresponding to multiple sub-processes in the executable policy set through the business interfaces corresponding to the multiple sub-processes, thereby providing an information processing method applied to the advertising engine, wherein the multi-level intelligent agent in the advertising engine can automatically process the entire advertising process based on the user's advertising demands, lowering the participation threshold of the advertising business and improving the user experience and convenience in the advertising processing process.

[0182] Continue to refer Figure 5 , shows a schematic process 500 of another embodiment (advertising creative delivery embodiment) of the information processing method applied to the advertising engine according to the present application, including the following steps:

[0183] Step 501: Determine a target node in the advertisement delivery graph from the advertisement appeal.

[0184] In this embodiment, the execution entity can determine the target node in the advertisement delivery graph from the advertisement appeal, wherein the advertisement delivery graph represents all advertisement links in the advertisement delivery process.

[0185] Continue to refer Figure 6 , showing a schematic diagram of an advertisement placement graph 600. The advertisement link includes multiple nodes, and there is a connection relationship between the nodes that have a sequential relationship.

[0186] As an example, the execution entity first determines the nodes included in the advertising appeal; then matches each node included in the advertising appeal with each node in the advertising delivery graph, and determines the matched nodes as target nodes. Generally, the number of target nodes determined is multiple.

[0187] Step 502: Determine the advertisement delivery links between target nodes from the advertisement delivery graph.

[0188] In this embodiment, the execution entity may determine the advertisement delivery link between target nodes from the advertisement delivery graph.

[0189] In this embodiment, the execution entity may determine the feasible connection between target nodes as an advertisement delivery link between the target nodes. A feasible connection between target nodes means: if the target nodes are not directly connected, then a connection that connects all target nodes is a feasible connection; if the target nodes are directly connected, then a connection that extends the connection to form a closed loop is a feasible connection.

[0190] In some optional implementations of this embodiment, the execution entity may perform step 402 in the following manner: using a depth-first traversal algorithm to determine the advertisement delivery links between target nodes from the advertisement delivery graph.

[0191] The Depth-First Search (DFS) algorithm is a graph or tree traversal method that starts from a node and recursively traverses the depth of the graph by exploring its adjacent nodes. Because the advertising delivery links between target nodes are specifically represented by the temporal connections between target nodes, the Depth-First Search algorithm is more suitable for determining the advertising delivery links.

[0192] In this implementation, a depth-first traversal algorithm is used to determine the advertisement delivery link, thereby improving the efficiency and accuracy of determining the advertisement delivery link.

[0193] In some optional implementations of this embodiment, the execution entity may perform the process of determining the advertisement delivery link in the following manner:

[0194] First, starting with the starting point of the advertising service, recursively perform the following search operations on the nodes in the advertising graph:

[0195] Mark the status of the current node as visited and add the current node to the current path; in response to determining that the current node is the target node, add the current path to the advertising delivery link list; in response to determining that there is an unvisited node among the adjacent nodes of the current node, determine the current node corresponding to the next search operation from the unvisited nodes, and perform the next search operation; in response to determining that all adjacent nodes of the current node have been visited, backtrack to the previous node of the current node, mark the access status of the current node as unvisited, and remove the current node from the current path.

[0196] As an example, the determination process includes the following steps:

[0197] 1. Select the starting point of the advertising service as the starting node and mark it as visited.

[0198] 2. At the same time, initialize an empty advertising link list to store all feasible paths found.

[0199] 3. Recursive search: Starting from the starting node, perform the following steps for each adjacent unvisited node:

[0200] 3.1. Mark the current node as visited.

[0201] 3.2. Add the current node to the current path.

[0202] 3.3. Recursively perform the above steps for unvisited adjacent nodes.

[0203] 3.4. If the current node is the target node, add the current path to the feasible path list.

[0204] 4. Backtracking: After all adjacent nodes of the current node have been visited, backtrack to the previous node, unmark the current node, and remove it from the path. Continue recursively searching for other unvisited nodes.

[0205] Repeat steps 3 and 4 until all feasible paths are found. This set is the feasible link set.

[0206] Second, the links in the advertisement delivery link list are determined as advertisement delivery links between target nodes in the advertisement delivery graph.

[0207] For example, if a merchant enters "spend 5,000 yuan to place a certain sku", the calculated advertising placement links include:

[0208] 1. Use 5,000 yuan to place CPC product ads and add X crowd packages to obtain click effects from targeted groups.

[0209] 2. Use 5,000 yuan to place CPS ads, set X commission, and use a certain expert to obtain conversion and investment-output ratio.

[0210] 3. Using 5,000 yuan to place contact ads can get better exposure.

[0211] It should be noted that the above example is only a schematic illustration of the advertising delivery link. In practice, there are advertising delivery businesses involving hundreds or even thousands of business units, and the determined advertising delivery links will be numerous and the structure will be more complex.

[0212] In this implementation, a specific implementation method for determining an advertisement delivery link using a depth-first traversal algorithm is provided, which further improves the efficiency and accuracy of determining the advertisement delivery link.

[0213] Step 503: Use the executable strategy corresponding to the advertisement delivery sub-process and execute the advertisement creative delivery process according to the advertisement delivery link.

[0214] In this embodiment, the execution subject may adopt the executable strategy corresponding to the advertisement delivery sub-process and execute the advertisement creative delivery process according to the advertisement delivery link.

[0215] The advertising delivery link is used to represent the delivery process of advertising creatives. During the advertising delivery process that is executed step by step according to the advertising delivery link, each link in the delivery process is executed according to the executable strategy.

[0216] The method provided by the above-mentioned embodiment of the present application provides a new advertising delivery method that can meet the user's advertising delivery needs by determining the target nodes existing in the advertising delivery graph from the advertising appeal, wherein the advertising delivery graph represents all advertising links in the advertising delivery process; determining the advertising delivery links between the target nodes from the advertising delivery graph; and using the executable strategy corresponding to the advertising delivery sub-process to execute the advertising creative delivery process according to the advertising delivery link.

[0217] In some optional implementations of this embodiment, the advertisement placement graph is created in the following manner:

[0218] First, extract the key points related to the advertising business in the advertising delivery sub-process; then, merge the same key points to obtain the advertising delivery graph. The nodes in the advertising delivery graph include node identifiers and node values ​​representing the advertising revenue of the nodes.

[0219] As an example, we first review various existing advertising delivery processes. Then, we perform the following operations on each of these processes: extract key points related to each business in the delivery process; and merge nodes that share the same key points.

[0220] After breaking down all the advertising delivery processes and performing the above operations, a complete advertising delivery graph knowledge base can be formed.

[0221] Each node in the ad placement graph contains two pieces of information: a node ID and a node value. The node ID represents the ad action for that node, while the node value represents the ad revenue for that node. For example, the ad revenue for a CPS node includes the highest cost-to-performance ratio and the highest conversion rate.

[0222] In this implementation, a method for generating an advertisement placement map is provided, which improves the comprehensiveness and accuracy of the information in the advertisement placement map.

[0223] In some optional implementations of this embodiment, there are multiple advertisement delivery links between target nodes. In this implementation, the execution subject may also perform the following operations:

[0224] A preset compression method is used to compress each of the multiple ad delivery links, and the compressed multiple ad delivery links are displayed. The preset compression method can be set according to actual conditions and is not limited here.

[0225] In this implementation, the execution entity may first construct a data structure of an advertisement delivery link. The data structure mainly includes two parts: feasible delivery behavior and expected revenue.

[0226] Specifically, the execution entity above uses the input information (node ​​identifier and node value) of the nodes in the advertising placement graph as the basis to form a two-dimensional data structure: {feasible placement behavior: expected return}; and then, the feasible links are expressed using the above structure. The following is an example of a set of feasible links:

[0227] {Advertising appeal}-{Feasible delivery behavior 1: Expected return 1}-{Feasible delivery behavior 2: Expected return 2}

[0228] {Advertising appeal}-{Feasible delivery behavior 1: Expected return 1}-{Feasible delivery behavior 3: Expected return 3}

[0229] {Advertising appeal}-{Feasible delivery behavior 2: Expected return 2}-{Feasible delivery behavior 3: Expected return 3}

[0230] The preset compression method is, for example, to select only nodes directly connected to the target node to represent a feasible link, deduplicate repeated expected benefits, and combine the deduplicated results together.

[0231] Continuing with the above advertising appeal of "spending 5000 yuan to place a certain sku" as an example, refer to the attached Figure 7 , shows a user interface 700 of compressed multiple advertisement delivery links. The compressed multiple advertisement delivery links are:

[0232] Feasible link 1: CPC, expected effect XXX

[0233] Feasible link 2: CPS, expected effect XXX

[0234] Viable link 3: Touchpoint advertising, expected effect XXX

[0235] The advertising delivery agent uses the above compression method to output and display the first N feasible links on the front end. Users can view the links and choose to expand them to view the details of each feasible link.

[0236] In this implementation, multiple compressed advertising links are displayed to users, which improves user experience and information acquisition efficiency.

[0237] In some optional implementations of this embodiment, the execution subject may further perform the following operation: determining an advertisement delivery link to be executed from multiple advertisement delivery links according to the received link selection operation.

[0238] Users can use voice commands, click, touch and other action commands to issue link selection operations. Figure 7 The above-mentioned execution subject can directly click the display area corresponding to the advertising delivery link to issue a link selection operation.

[0239] In this implementation, users can flexibly select the advertising delivery link to be executed based on their own demands, which helps to further improve the user experience and meet the user's advertising demands.

[0240] In some optional implementations of this embodiment, the execution entity may further perform the following operations:

[0241] First, the supplementary demand is determined according to the received supplementary demand operation.

[0242] Continue to refer to the above Figure 7 , users can click the "Supplementary Demand" button to enter additional demands.

[0243] Then, based on the advertising appeal and the supplementary appeal, the updated target node is determined from the advertising delivery graph.

[0244] As an example, the execution entity first determines the nodes included in the supplementary appeal; then matches each node included in the advertising appeal and each node included in the supplementary appeal with each node in the advertising placement graph, and determines the matched nodes as the updated target nodes. Generally, there are multiple target nodes.

[0245] Finally, the advertising delivery links between the updated target nodes are determined from the advertising delivery graph.

[0246] In this implementation, the execution entity may refer to the above process of determining the advertisement delivery link based on the target node to determine the advertisement delivery link between the updated target nodes, which will not be described in detail here.

[0247] In this implementation, users can perform multiple appeal supplement operations based on the advertising appeal to express the appeal information as perfectly as possible, so that the final advertising delivery link is more adapted to the user's expectations.

[0248] Continue to refer Figure 8, shows a schematic process 800 of another embodiment (advertising creative review embodiment) of the information processing method applied to the advertising engine according to the present application, including the following steps:

[0249] Step 801: Determine the creative materials in the advertisement creative submitted by the user.

[0250] In this embodiment, the execution entity may determine the creative materials in the advertisement creative submitted by the user.

[0251] Advertising creative refers to user-submitted advertising data, commonly understood as advertising. Creative materials refer to the various materials and resources used in the production, display, and dissemination of advertising campaigns, including copy, images, audio, video, and other content. Creative materials are the concrete manifestation of advertising creativity. Through various forms of advertising materials, users can showcase their products, services, or brand image to their target audiences, conveying the advertiser's desired message and emotions.

[0252] The advertising creativity in this embodiment is provided by the user and is not automatically generated by a multi-level intelligent agent based on the user's advertising demands.

[0253] As an example, the execution entity may analyze and process the advertisement creative submitted by the user to determine the various creative materials included therein.

[0254] Step 802: Using the advertisement review agent, similar materials to the creative material are determined from the second knowledge base.

[0255] In this embodiment, the execution entity may determine similar materials of the creative material from the second knowledge base through the advertisement review agent.

[0256] The advertising review agent can automatically review advertising creatives, using neural network models such as recurrent neural network models and residual network models, as well as large language models such as GPT (Generative Pre-Trained Transformer) and BERT (Bidirectional Encoder Representations from Transformers).

[0257] The second knowledge base includes various types of rich materials. As an example, the execution entity can analyze the advertising creatives of various well-known brands known to date and generate the second knowledge base through statistics.

[0258] In this embodiment, the execution entity can use the advertisement review agent to determine the similarity between each material in the second knowledge base and the creative material; then, based on the similarity, determine similar materials to the creative material from the second knowledge base. The similarity can be determined, for example, by calculating cosine similarity, Euclidean distance, etc.

[0259] In some optional implementations of this embodiment, the execution entity may perform step 802 as follows:

[0260] First, determine the first material characteristics of the creative material.

[0261] As an example, the execution entity may extract features of the creative material through a feature extraction network to generate a first material feature of the creative material. The first material feature may be specifically represented by a feature vector.

[0262] Then, similar materials are determined based on the similarity between the first material feature and the second material feature of the materials in the second knowledge base.

[0263] In this implementation, the execution entity may use the same feature extraction network to extract the second material feature of the materials in the second knowledge base, further determine the similarity between the first material feature and the second material feature of the materials in the second knowledge base, and determine the materials in the second knowledge base whose similarity exceeds a preset threshold as similar materials. The preset threshold can be set according to actual circumstances and is not limited here.

[0264] In this implementation, a specific implementation method for determining similar materials based on material characteristics is provided, which improves the accuracy of determining similar materials.

[0265] In some optional implementations of this embodiment, the second knowledge base includes an advertising knowledge base and a brand knowledge base. The advertising knowledge base refers to advertising knowledge in the advertising field, such as advertising knowledge related to the advertising industry and advertising companies. The brand knowledge base includes various types of knowledge derived from various types of advertising users (e.g., various advertising brands), such as brand trademarks, names, product names, creative images, creative videos, etc. For both the advertising knowledge base and the brand knowledge base, the advertising engine supports manual entry or import to write data to the two types of knowledge bases.

[0266] In this implementation, the execution entity may perform the similar material determination process as follows:

[0267] First, based on the similarity between the first material feature and the second material feature of the material in the advertising knowledge base, a plurality of similar materials ranked top in similarity are determined.

[0268] In this implementation, the execution entity may sort the materials in the advertising knowledge base in descending order of similarity, and then determine the multiple materials ranked highest in similarity as similar materials.

[0269] Then, according to the similarity between the first material feature and the second material feature of the material in the brand knowledge base, a plurality of similar materials with a top similarity ranking and a similarity exceeding a preset similarity threshold are determined.

[0270] In this implementation, the execution entity may sort the materials in the brand knowledge base in descending order of similarity, and then determine multiple materials with higher similarity rankings and a similarity exceeding a preset similarity threshold as similar materials.

[0271] In this implementation, the second knowledge base includes an advertising knowledge base and a brand knowledge base, and similar materials are recalled separately for the two types of knowledge bases, thereby improving the richness and accuracy of similar materials.

[0272] In some optional implementations of this embodiment, the execution entity may further perform the following operation: adjusting a preset similarity threshold according to the operation status of the advertisement engine.

[0273] The operation status is, for example, whether the user has subscribed to a service for adjusting a preset similarity threshold, wherein the degree of adjustment of the preset similarity threshold corresponds to different fees.

[0274] It can be understood that after an advertising user pays, the preset similarity threshold corresponding to the advertising user can be increased, so that the similarity between multiple similar materials returned based on the brand knowledge base and the creative materials is greater, which is more helpful to make the generated new materials similar to the advertising creativity of the advertising user, and improve the advertising review of the advertising creativity submitted by the user for the advertising user.

[0275] In this implementation, the preset similarity threshold can be flexibly adjusted based on the operating conditions to adjust the advertising creatives submitted by users to improve the advertising review of the advertising user, which helps to improve the flexibility of the review process and user experience.

[0276] Step 803: Generate new materials based on similar materials through the advertisement review agent.

[0277] In this embodiment, the above-mentioned execution entity can generate new materials based on similar materials through the advertisement review intelligent agent.

[0278] For example, Ad Review Intelligence has AIGC (Artificial Intelligence Generated Content), which can generate new materials based on similar materials. The Ad Review Agent extracts features from similar materials, obtains similar material characteristics, and then generates new materials based on these similar material characteristics.

[0279] In some optional implementations of this embodiment, the above-mentioned execution entity may further perform the following operations: determining the third material feature of the creative material in the historical advertising creative submitted by the user; and fusing the first material feature and the third material feature.

[0280] As an example, the above-mentioned execution entity can use the same feature extraction network to extract features of creative materials in historical advertising creatives submitted by users and up to the current preset time period to obtain third material features of the creative materials in the historical advertising creatives.

[0281] In this implementation, the execution entity may perform step 803 in the following manner: generating a new material according to the second material feature and the fusion feature corresponding to the similar material.

[0282] As an example, through the advertising review agent, features of similar materials are extracted to obtain similar material features, and then new materials are generated based on similar material features and fusion features.

[0283] In this implementation, historical advertising ideas submitted by users are taken into consideration during the production of new materials, and new materials are generated for advertising review by combining historical advertising ideas submitted by users with currently submitted advertising ideas, which helps to further improve the accuracy of the review results.

[0284] Step 804: Determine the advertisement review result based on the similarity between the new material and the advertisement creative ideas of the advertisement users in the advertisement user set.

[0285] In this embodiment, the execution entity may determine the advertisement review result based on the similarity between the new material and the advertisement creative ideas of the advertisement users in the advertisement user set.

[0286] The advertising user set includes all advertising users up to now, and records the advertising creative data of each advertising user.

[0287] In this embodiment, the execution entity can determine the similarity between the new material and the advertising creative of each advertiser in the set of advertisers; and then, based on the similarity, determine the ad review result. For example, if the similarity between the new material and a particular advertiser's advertising creative exceeds a preset threshold, it is determined that the user-submitted advertising creative is "piggybacking" on the advertiser. The similarity can be determined, for example, by calculating cosine similarity, Euclidean distance, or the like.

[0288] In this embodiment, creative materials in the advertising creative submitted by the user are determined; similar materials of the creative materials are determined from the second knowledge base through the advertising review intelligent body; new materials are generated based on the similar materials through the advertising review intelligent body; and the advertising review result is determined based on the similarity between the new materials and the advertising creative of the advertising users in the advertising user set, thereby providing a new advertising review method and improving the accuracy of advertising review.

[0289] Continue to refer Figure 9 As an implementation of the methods shown in the above figures, the present application provides an embodiment of an information processing device applied to an advertising engine. The device embodiment corresponds to the method embodiments shown in Figures 200, 500, and 800, and the device can be specifically applied to various electronic devices.

[0290] like Figure 9 As shown, the information processing device 900 applied to the advertising engine includes: an acquisition unit 901, configured to acquire the user's advertising demands; a determination unit 902, configured to determine an executable policy set that meets the advertising demands through multi-level intelligent agents, wherein the intelligent agents at each level in the multi-level intelligent agents can determine the executable policy set corresponding to the advertising demands, and the executable policy set includes executable policies corresponding to multiple sub-processes in the entire advertising process; an execution unit 903, configured to execute the executable policies corresponding to multiple sub-processes in the executable policy set through the business interfaces corresponding to the multiple sub-processes.

[0291] In some optional implementations of this embodiment, the agents at each level in the multi-level agent include multiple agents, and the above-mentioned determination unit 902 is further configured to: process multiple sub-appeals in the advertising appeal in the order of execution of the agents at each level in the multi-level agent, and in response to the presence of a target sub-appeal among the multiple sub-appeals for which the agent at the current level has not determined an executable strategy based on the preset voting strategy, process the target sub-appeal through the agents at subsequent levels until the executable strategies corresponding to the multiple sub-appeals are combined to obtain an executable strategy set.

[0292] In some optional implementations of this embodiment, the first-level agents in the multi-level agent include agent groups corresponding to multiple sub-processes, each agent group includes multiple agents, and the above-mentioned determination unit 902 is further configured to: decompose user demands and determine the sub-demands corresponding to the multiple sub-processes; process the sub-demands corresponding to the multiple sub-processes through the agent groups corresponding to the multiple sub-processes to obtain multiple candidate executable strategies corresponding to the multiple sub-processes; for the multiple candidate executable strategies output by each of the multiple agent groups, determine whether an executable strategy can be determined from the multiple candidate executable strategies based on a preset voting strategy; in response to each of the multiple agent groups, the executable strategy corresponding to the agent group can be determined, and the executable strategy set is obtained by combining the executable strategies corresponding to the multiple agent groups.

[0293] In some optional implementations of this embodiment, the above-mentioned determination unit 902 is further configured to: in response to determining that there is a target intelligent agent group among multiple intelligent agent groups for which an executable strategy has not been determined, process the target sub-demand through multiple intelligent agents at subsequent levels until an executable strategy is determined from the candidate executable strategies output by each of the multiple intelligent agents at the current level based on a preset voting strategy, wherein the multiple intelligent agents at subsequent levels are determined according to the execution order; and combine the executable strategies determined by other intelligent agent groups except the target intelligent agent group among the multiple intelligent agent groups and the executable strategies determined by the multiple intelligent agents at the current level to obtain an executable strategy set.

[0294] In some optional implementations of this embodiment, the above-mentioned determination unit 902 is further configured to: parse the sub-claims corresponding to the multiple sub-processes through the intelligent agent groups corresponding to the multiple sub-processes, and obtain the advertising terms corresponding to the multiple sub-claims; for each intelligent agent group in the multiple intelligent agent groups, determine the candidate executable strategy corresponding to the sub-process corresponding to the intelligent agent group from the preset limited result set according to the advertising terms corresponding to the sub-claim to be processed by the intelligent agent group.

[0295] In some optional implementations of this embodiment, the above-mentioned multi-level intelligent agents are obtained in the following manner: determining the initial models corresponding to each multi-level intelligent agent based on the accuracy and parameter scale of the models in the model set, wherein the level to which the intelligent agent belongs and the accuracy and parameter scale of the initial model corresponding to the level are positively correlated; determining the first knowledge base corresponding to each multi-level intelligent agent, wherein the first knowledge base corresponding to intelligent agents at different levels includes advertising knowledge of different preset dimensions; training the initial models corresponding to each multi-level intelligent agent through the first knowledge base corresponding to each multi-level intelligent agent to obtain the multi-level intelligent agent.

[0296] In some optional implementations of this embodiment, multiple sub-processes include an advertising creative sub-process, an advertising delivery sub-process, an advertising operation sub-process and an advertising playback sub-process, and the above-mentioned execution unit 903, which is further configured to: generate advertising creativity through the business interface corresponding to the advertising creative sub-process and according to the executable policy corresponding to the advertising creative sub-process in the executable policy set; execute the delivery process of the advertising creativity through the business interface corresponding to the advertising delivery sub-process and according to the executable policy corresponding to the advertising delivery sub-process in the executable policy set; execute the algorithms involved in the advertising process through the business interface corresponding to the advertising operation sub-process and according to the executable policy corresponding to the advertising operation sub-process in the executable policy set; execute the playback process of the advertising creativity through the business interface corresponding to the advertising playback sub-process and according to the executable policy corresponding to the advertising playback sub-process in the executable policy set.

[0297] In some optional implementations of this embodiment, the above-mentioned execution unit 903 is further configured to: determine the target nodes existing in the advertising delivery graph from the advertising appeal, wherein the advertising delivery graph represents all advertising links in the advertising delivery process; determine the advertising delivery links between the target nodes from the advertising delivery graph; adopt the executable strategy corresponding to the advertising delivery sub-process, and execute the advertising creative delivery process according to the advertising delivery link.

[0298] In some optional implementations of this embodiment, the execution unit 903 is further configured to: use a depth-first traversal algorithm to determine the advertisement delivery links between target nodes from the advertisement delivery graph.

[0299] In some optional implementations of this embodiment, the above-mentioned advertising delivery graph is created in the following manner: extracting key points related to the delivery business in the advertising delivery sub-process; merging the same key points to obtain an advertising delivery graph, wherein the nodes in the advertising delivery graph include node identifiers and node values ​​representing the advertising revenue of the nodes.

[0300] In some optional implementations of this embodiment, the above-mentioned advertising delivery links include multiple links, and the above-mentioned device also includes: a compression display unit (not shown in the figure), which is configured to use a preset compression method to compress each of the multiple advertising delivery links and display the compressed multiple advertising delivery links.

[0301] In some optional implementations of this embodiment, the apparatus further includes: a selection unit (not shown in the figure), configured to determine an advertisement delivery link to be executed from a plurality of advertisement delivery links according to a received link selection operation.

[0302] In some optional implementations of this embodiment, the above-mentioned device also includes: a supplementary unit (not shown in the figure), which is configured to determine the supplementary appeal based on the received supplementary appeal operation; determine the updated target node from the advertising delivery graph based on the advertising appeal and the supplementary appeal; and determine the advertising delivery link between the updated target nodes from the advertising delivery graph.

[0303] In some optional implementations of this embodiment, the above-mentioned device also includes: an audit unit (not shown in the figure), which is configured to determine the creative materials in the advertising creativity submitted by the user; determine similar materials of the creative materials from the second knowledge base through the advertising audit intelligent body; generate new materials based on the similar materials through the advertising review intelligent body; determine the advertising audit results based on the similarity between the new materials and the advertising creativity of the advertising users in the advertising user set.

[0304] In some optional implementations of this embodiment, the above-mentioned review unit (not shown in the figure) is further configured to: determine the first material feature of the creative material; and determine similar materials based on the similarity between the first material feature and the second material feature of the material in the second knowledge base.

[0305] In some optional implementations of this embodiment, the above-mentioned second knowledge base includes an advertising knowledge base and a brand knowledge base, and the above-mentioned review unit (not shown in the figure) is further configured to: determine a plurality of similar materials with a high similarity ranking based on the similarity between the first material feature and the second material feature of the material in the advertising knowledge base; determine a plurality of similar materials with a high similarity ranking and a similarity exceeding a preset similarity threshold based on the similarity between the first material feature and the second material feature of the material in the brand knowledge base.

[0306] In some optional implementations of this embodiment, the above-mentioned device also includes: a fusion unit (not shown in the figure), configured to determine the third material feature of the creative material in the historical advertising creativity submitted by the user; fuse the first material feature and the third material feature to obtain a fusion feature; and the above-mentioned review unit is further configured to: generate a new material based on the second material feature and the fusion feature corresponding to similar materials.

[0307] In some optional implementations of this embodiment, the apparatus further includes: an adjustment unit (not shown in the figure), configured to adjust the preset similarity threshold according to the operation status of the advertising engine.

[0308] In some optional implementations of this embodiment, the acquisition unit 901 is further configured to: acquire the user's advertising demands represented by natural language.

[0309] In some examples, the apparatus further includes an agent determination unit (not shown in the figure), configured to determine an agent at each level of the multi-level agent based on the received selection operation.

[0310] In this embodiment, an acquisition unit in an information processing device applied to an advertising engine acquires the user's advertising demands; a determination unit determines an executable policy set that satisfies the advertising demands through a multi-level intelligent agent, wherein the intelligent agent at each level in the multi-level intelligent agent can determine the executable policy set corresponding to the advertising demands, and the executable policy set includes executable policies corresponding to each of the multiple sub-processes in the entire advertising process; the execution unit executes the executable policies corresponding to each of the multiple sub-processes in the executable policy set through the business interfaces corresponding to each of the multiple sub-processes, thereby providing an information processing device applied to an advertising engine, wherein the multi-level intelligent agent in the advertising engine can automatically process the entire advertising process based on the user's advertising demands, thereby lowering the participation threshold of the advertising business and improving the user experience and convenience in the advertising processing process.

[0311] Reference below Figure 10 , which shows a device suitable for implementing the embodiments of the present application (eg Figure 1 Schematic diagram of the structure of the computer system 1000 of the devices 101, 102, 103, 105 shown. Figure 10 The device shown is only an example and should not limit the functions and scope of use of the embodiments of the present application.

[0312] like Figure 10 As shown, the computer system 1000 includes a processor (e.g., CPU, central processing unit) 1001, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage unit 1008 into a random access memory (RAM) 1003. Various programs and data required for the operation of the system 1000 are also stored in the RAM 1003. The processor 1001, ROM 1002, and RAM 1003 are connected to each other via a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0313] The following components are connected to the I / O interface 1005: an input section 1006 including a keyboard, a mouse, and the like; an output section 1007 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and a speaker; a storage section 1008 including a hard disk; and a communication section 1009 including a network interface card such as a LAN card or a modem. The communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to the I / O interface 1005 as needed. A removable medium 1011, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 1010 as needed, so that computer programs read therefrom can be installed into the storage section 1008 as needed.

[0314] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart 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 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 via the communication section 1009 and / or installed from the removable medium 1011. When the computer program is executed by the processor 1001, the above-mentioned functions defined in the method of the present application are performed.

[0315] It should be noted that the computer-readable medium of the present application may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, device, or device. In the present application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. Program code embodied on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wireline, optical fiber cable, RF, or any suitable combination thereof.

[0316] Computer program code for performing the operations of the present application may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the client computer, partially on the client computer, as a stand-alone software package, partially on the client computer and partially on a remote computer, or entirely on the remote computer or server. In cases involving a remote computer, the remote computer may be connected to the client computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0317] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the devices, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0318] The units involved in the embodiments described in the present application can be implemented by software or by hardware. The described units can also be set in a processor. For example, they can be described as: a processor comprising an acquisition unit, a determination unit and an execution unit. Among them, the names of these units do not constitute a limitation on the unit itself under certain circumstances. For example, the determination unit can also be described as "determining an executable strategy set that meets the advertising demands through a multi-level intelligent agent, wherein each level of the intelligent agent in the multi-level intelligent agent can determine the executable strategy set corresponding to the advertising demands, and the executable strategy set includes units of executable strategies corresponding to multiple sub-processes in the entire advertising process."

[0319] As another aspect, the present application also provides a computer-readable medium, which may be included in the device described in the above embodiment; or it may exist independently and not be assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by the device, the computer device: obtains the user's advertising demands; determines an executable policy set that meets the advertising demands through a multi-level intelligent agent, wherein each level of the multi-level intelligent agent can determine the executable policy set corresponding to the advertising demands, and the executable policy set includes executable policies corresponding to each of the multiple sub-processes in the entire advertising process; and executes the executable policies corresponding to each of the multiple sub-processes in the executable policy set through the business interfaces corresponding to each of the multiple sub-processes.

[0320] The above description is merely a preferred embodiment of the present application and an illustration of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but also encompasses other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the above-mentioned inventive concept. For example, a technical solution formed by replacing the above-mentioned features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. An information processing method applied to an advertising engine, comprising: Obtain users’ advertising demands; Determining an executable strategy set that satisfies the advertising appeal through multi-level intelligent agents, including: processing multiple sub-appeals in the advertising appeal in the order of execution of intelligent agents at each level in the multi-level intelligent agents, and in response to the presence of a target sub-appeal among the multiple sub-appeals for which the intelligent agents at the current level have not determined an executable strategy based on a preset voting strategy, processing the target sub-appeal through intelligent agents at subsequent levels until the executable strategies corresponding to the multiple sub-appeals are combined to obtain the executable strategy set, wherein each level includes multiple intelligent agents, and the intelligent agents at each level in the multi-level intelligent agents can determine the executable strategy set corresponding to the advertising appeal, and the executable strategy set includes the executable strategies corresponding to the multiple sub-processes in the entire advertising process; The executable policies corresponding to the multiple sub-processes in the executable policy set are executed through the business interfaces corresponding to the multiple sub-processes.

2. The method according to claim 1, wherein The first level of agents in the multi-level agent layer includes an agent group corresponding to each of the plurality of sub-processes, wherein the agent group includes a plurality of agents, and The method further comprises processing the plurality of sub-appeals in the advertisement appeal in the order of execution of the agents at each level of the multi-level agents, and in response to a target sub-appeal among the plurality of sub-appeals for which the agents at the current level have not determined an executable strategy based on a preset voting strategy, processing the target sub-appeal by the agents at subsequent levels until the executable strategies corresponding to the plurality of sub-appeals are combined to obtain the executable strategy set, including: Decomposing the advertising appeal and determining the sub-appeals corresponding to the multiple sub-processes; Processing the sub-demands corresponding to the multiple sub-processes by the agent groups corresponding to the multiple sub-processes to obtain multiple candidate executable strategies corresponding to the multiple sub-processes; For multiple candidate executable strategies output by each of the multiple agent groups, determining whether an executable strategy can be determined from the multiple candidate executable strategies based on the preset voting strategy; In response to the fact that for each of the multiple agent groups, an executable strategy corresponding to the agent group can be determined, the executable strategy set is obtained by combining the executable strategies corresponding to each of the multiple agent groups.

3. The method according to claim 1, wherein The method further includes processing the plurality of sub-appeals in the advertisement appeal in the order of execution of the agents at each level of the multi-level agents, and in response to a target sub-appeal among the plurality of sub-appeals for which the agents at the current level have not determined an executable strategy based on a preset voting strategy, processing the target sub-appeal by agents at subsequent levels until the executable strategies corresponding to the plurality of sub-appeals are combined to obtain the executable strategy set. In response to determining that there is a target agent group among the multiple agent groups for which no executable strategy has been determined, processing the target sub-demand by multiple agents at subsequent levels until an executable strategy is determined from candidate executable strategies output by each of the multiple agents at the current level based on the preset voting strategy; The executable strategy set is obtained by combining the executable strategies determined by the other agent groups except the target agent group in the multiple agent groups and the executable strategies determined by the multiple agents at the current level.

4. The method according to claim 2, wherein: The method further includes processing the plurality of sub-appeals in the advertisement appeal in the order of execution of the agents at each level of the multi-level agents, and in response to a target sub-appeal among the plurality of sub-appeals for which the agents at the current level have not determined an executable strategy based on a preset voting strategy, processing the target sub-appeal by agents at subsequent levels until the executable strategies corresponding to the plurality of sub-appeals are combined to obtain the executable strategy set. In response to determining that there is a target agent group among the multiple agent groups for which no executable strategy has been determined, processing the target sub-demand by multiple agents at subsequent levels until an executable strategy is determined from candidate executable strategies output by each of the multiple agents at the current level based on the preset voting strategy; The executable strategy set is obtained by combining the executable strategies determined by the other agent groups except the target agent group in the multiple agent groups and the executable strategies determined by the multiple agents at the current level.

5. The method according to claim 2, wherein: The step of processing the sub-demands corresponding to the plurality of sub-processes by the agent groups corresponding to the plurality of sub-processes to obtain the plurality of candidate executable strategies corresponding to the plurality of sub-processes includes: parsing the sub-demands corresponding to the multiple sub-processes through the agent groups corresponding to the multiple sub-processes to obtain advertising terms corresponding to the multiple sub-demands; For each of the multiple agent groups, based on the advertising terms corresponding to the sub-appeal to be processed by the agent group, a candidate executable strategy corresponding to the sub-process of the agent group is determined from a preset limited result set.

6. The method according to any one of claims 1 to 5, wherein The multi-level agent is obtained in the following way: Determine the initial model corresponding to each of the multiple levels of agents based on the accuracy and parameter scale of the models in the model set, wherein the accuracy and parameter scale of the initial model corresponding to the level to which the agent belongs are positively correlated; Determining first knowledge bases corresponding to the multi-level intelligent agents, wherein the first knowledge bases corresponding to intelligent agents at different levels include advertising knowledge of different preset dimensions; The initial models corresponding to the multi-level intelligent agents are trained by using the first knowledge bases corresponding to the multi-level intelligent agents to obtain the multi-level intelligent agents.

7. The method according to claim 1, wherein The multiple sub-processes include an advertising creative sub-process, an advertising delivery sub-process, an advertising operation sub-process and an advertising playback sub-process, and The executing, through the business interfaces corresponding to the multiple sub-processes, the executable policies corresponding to the multiple sub-processes in the executable policy set, includes: Generate an advertising creative through the business interface corresponding to the advertising creative sub-process and according to the executable strategy corresponding to the advertising creative sub-process in the executable strategy set; Executing the advertising creative delivery process according to the executable strategy corresponding to the advertising delivery sub-process in the executable strategy set through the business interface corresponding to the advertising delivery sub-process; Executing the operation process of the advertising creative according to the executable strategy corresponding to the advertising operation sub-process in the executable strategy set through the business interface corresponding to the advertising operation sub-process; The advertisement playing process is executed through the service interface corresponding to the advertisement playing sub-process and according to the executable policy corresponding to the advertisement playing sub-process in the executable policy set.

8. The method according to claim 7, wherein: The process of executing the advertisement delivery process according to the executable policy corresponding to the advertisement delivery sub-process in the executable policy set includes: Determining a target node in an advertisement delivery graph from the advertisement appeal, wherein the advertisement delivery graph represents all advertisement links in an advertisement delivery process; Determining the advertisement delivery links between the target nodes from the advertisement delivery graph; The advertisement delivery process is executed by adopting the executable strategy corresponding to the advertisement delivery sub-process and following the advertisement delivery link.

9. The method according to claim 8, wherein Determining the advertisement delivery link between the target nodes from the advertisement delivery graph includes: Advertisement delivery links between the target nodes are determined from the advertisement delivery graph using a depth-first traversal algorithm.

10. The method according to claim 8, wherein The advertising graph is created in the following way: Extracting key points related to the advertising delivery business in the advertising delivery sub-process; The same key points are merged to obtain the advertisement placement graph, wherein the nodes in the advertisement placement graph include node identifiers and node values ​​representing the advertisement revenue of the nodes.

11. The method according to claim 8, wherein The advertising delivery link includes multiple links, and The method further comprises: A preset compression method is used to compress each of the multiple advertising delivery links, and the compressed multiple advertising delivery links are displayed.

12. The method according to claim 11, wherein Also includes: According to the received link selection operation, an advertisement delivery link to be executed is determined from the multiple advertisement delivery links.

13. The method according to any one of claims 8 to 12, wherein: Also includes: Determine the supplementary demands based on the received supplementary demands operation; Determining an updated target node from the advertising placement graph according to the advertising appeal and the supplementary appeal; The advertisement delivery links between the updated target nodes are determined from the advertisement delivery graph.

14. The method according to claim 1, wherein Also includes: Determining the creative materials in the advertising creative submitted by the user; Determining, by an advertisement review agent, similar materials to the creative material from a second knowledge base; Generate new materials based on the similar materials through the advertisement review agent; An advertisement review result is determined based on similarities between the new material and advertisement creative ideas of advertisement users in the advertisement user set.

15. The method according to claim 14, wherein Determining similar materials to the creative material from the second knowledge base includes: determining a first material characteristic of the creative material; The similar material is determined based on the similarity between the first material feature and the second material feature of the material in the second knowledge base.

16. The method according to claim 15, wherein The second knowledge base includes an advertising knowledge base and a brand knowledge base, and The determining the similar material according to the similarity between the first material feature and the second material feature of the material in the second knowledge base includes: Determining a plurality of similar materials ranked top in similarity based on similarity between the first material feature and a second material feature of a material in the advertising knowledge base; According to the similarity between the first material feature and the second material feature of the material in the brand knowledge base, a plurality of similar materials with a top similarity ranking and a similarity exceeding a preset similarity threshold are determined.

17. The method according to claim 15 or 16, wherein Also includes: Determining a third material feature of the creative material in the historical advertising creative submitted by the user; fusing the first material feature and the third material feature to obtain a fused feature; as well as The generating of new materials based on the similar materials includes: A new material is generated according to the second material feature corresponding to the similar material and the fusion feature.

18. The method according to claim 16, wherein Also includes: The preset similarity threshold is adjusted according to the operation status of the advertising engine.

19. The method according to claim 1, wherein The obtaining of the user's advertising demands includes: Acquire the user's advertising demands represented by natural language.

20. The method according to claim 1, wherein Also includes: According to the received selection operation, the agent of each level in the multi-level agent is determined.

21. An information processing device for an advertising engine, comprising: an acquisition unit, configured to acquire a user's advertising demands; A determination unit is configured to determine an executable strategy set that satisfies the advertising appeal through a multi-level intelligent agent, including: processing a plurality of sub-appeals in the advertising appeal in the order of execution of the intelligent agents at each level in the multi-level intelligent agent, and in response to the presence of a target sub-appeal among the multiple sub-appeals for which the intelligent agent at the current level has not determined an executable strategy based on a preset voting strategy, processing the target sub-appeal through the intelligent agents at subsequent levels until the executable strategies corresponding to the plurality of sub-appeals are combined to obtain the executable strategy set, wherein each level includes a plurality of intelligent agents, and the intelligent agents at each level in the multi-level intelligent agent can determine the executable strategy set corresponding to the advertising appeal, and the executable strategy set includes the executable strategies corresponding to the plurality of sub-processes in the entire advertising process; The execution unit is configured to execute the executable policies corresponding to the multiple sub-processes in the executable policy set through the business interfaces corresponding to the multiple sub-processes.

22. A computer-readable medium having a computer program stored thereon, wherein: When the program is executed by a processor, the method according to any one of claims 1 to 20 is implemented.

23. An electronic device comprising: one or more processors; a storage device having one or more programs stored thereon, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 20.

24. A computer program product comprising: A computer program which, when executed by a processor, implements the method according to any one of claims 1 to 20.

Images