Advertisement data processing method and device, electronic equipment and storage medium

Abstract

The invention provides an advertisement data processing method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining a newly-created target advertisement and target advertisement data of the target advertisement; based on the target advertisement data, determining a module gain between the target advertisement and each community node in a target advertisement map, each historical advertisement in each community node in the target advertisement map satisfying a preset correlation; determining the maximum module gain in all the module gains; determining the community node corresponding to the maximum module gain as a target community node to which the target advertisement belongs; and taking the target identifier corresponding to the target community node as the feature of the target advertisement. Through the method and the device, the problem that the new advertisement lacks historical interaction data in related technologies is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of advertisement pushing, and in particular to an advertisement data processing method and device, an electronic device, and a storage medium. BACKGROUND

[0002] Performance advertising is an advertising mode with measurable user behaviors (such as clicks, conversions, downloads, purchases, etc.) as the core target. Advertisers only pay for actual performance (such as pay-per-click CPC, pay-per-conversion CPA, etc.), rather than simply paying for exposure (CPM). Its core features are data-driven, accurate delivery, and performance attribution. In performance advertising, new ads lack historical interaction data (such as clicks, conversions, etc.), which prevents algorithms from accurately estimating their performance (such as click-through rate CTR, conversion rate CVR, ROI, etc.), thereby affecting the accuracy of delivery strategies, which is the cold start problem.

[0003] Therefore, the related art has the technical problem that new ads lack historical interaction data. SUMMARY

[0004] The present application provides an advertisement data processing method and device, an electronic device, and a storage medium to at least solve the problem that new ads lack historical interaction data in the related art.

[0005] According to an aspect of an embodiment of the present application, an advertisement data processing method is provided, comprising: obtaining a newly created target advertisement and target advertisement data of the target advertisement; determining a module gain between the target advertisement and each community node in a target advertisement graph based on the target advertisement data, wherein each historical advertisement in each community node in the target advertisement graph satisfies a preset correlation; determining a maximum module gain from all the module gains; determining a target community node to which the target advertisement belongs as a community node corresponding to the maximum module gain; using a target identifier corresponding to the target community as a feature of the target advertisement.

[0006] Optionally, as described above, the method further comprises: constructing an original advertisement graph by taking each existing advertisement in a plurality of existing advertisements as an advertisement node, wherein an edge weight between any two advertisement nodes is used to indicate a similarity between two existing advertisements corresponding to the two advertisement nodes; The community division is performed on the advertisement nodes in the original advertisement graph by a community discovery algorithm to obtain the target advertisement graph including at least one community node, and a community identifier of a community node in which each advertisement node is located is allocated to each advertisement node.

[0007] Optionally, in the method described above, the method for determining the edge weight between any two advertisement nodes comprises: determining a first multi-modal embedding vector of a first advertisement corresponding to a first advertisement node of the two advertisement nodes; and determining a second multi-modal embedding vector of a second advertisement corresponding to a second advertisement node of the two advertisement nodes; determining a cosine similarity between the first multi-modal embedding vector and the second multi-modal embedding vector; based on a preset weight distribution ratio, performing weighted summation on the cosine similarity, a first number of users who click on both the first advertisement and the second advertisement, a second number of users who are converted on both the first advertisement and the second advertisement, and a third number of users who are converted through the first advertisement and the second advertisement, to determine the edge weight between the first advertisement node and the second advertisement node.

[0008] Optionally, in the method described above, the method for taking the target identifier corresponding to the target community as a feature of the target advertisement comprises: assigning the target identifier of the target community node to the target advertisement, and taking the target identifier as an advertisement feature of the target advertisement; The method further comprises: taking the target identifier as an advertisement feature of the target advertisement and inputting the target identifier into an advertisement recommendation system; in a case where an advertisement acquisition request is acquired, and a predicted click rate and / or a predicted conversion rate of the target advertisement pushed to an advertisement position corresponding to the advertisement acquisition request based on the target identifier corresponding to the target advertisement meet a preset requirement, pushing the target advertisement to the advertisement position.

[0009] According to another aspect of the embodiments of the present application, an advertisement data processing apparatus is further provided, comprising: an acquisition module configured to acquire a newly created target advertisement and target advertisement data of the target advertisement; a first determination module configured to determine a module gain between the target advertisement and each community node in a target advertisement graph based on the target advertisement data, wherein each historical advertisement in each community node in the target advertisement graph meets a preset correlation; a second determination module configured to determine a maximum module gain from all the module gains; a target community node determination module configured to determine a community node corresponding to the maximum module gain as a target community node to which the target advertisement belongs. a feature module, configured to take the target identifier corresponding to the target community as a feature of the target advertisement.

[0010] Optionally, the apparatus as described in the preceding, further comprises: an original advertisement graph construction module, configured to construct an original advertisement graph by taking each of a plurality of existing advertisements as an advertisement node, wherein an edge weight between any two advertisement nodes is used to indicate a similarity between two existing advertisements corresponding to the two advertisement nodes respectively; a community division module, configured to divide the advertisement nodes in the original advertisement graph through a community discovery algorithm to obtain the target advertisement graph comprising at least one community node, and assign each advertisement node a community identifier of the community node in which the advertisement node is located.

[0011] Optionally, the apparatus as described in the preceding, further comprises an edge weight determination module, configured to: determine a first multi-modal embedding vector of a first advertisement corresponding to a first advertisement node of the two advertisement nodes; and determine a second multi-modal embedding vector of a second advertisement corresponding to a second advertisement node of the two advertisement nodes; determine a cosine similarity between the first multi-modal embedding vector and the second multi-modal embedding vector; based on a preset weight distribution ratio, perform a weighted summation on the cosine similarity, a first number of users who have clicked on both the first advertisement and the second advertisement, a second number of users who have converted on both the first advertisement and the second advertisement, and a third number of users who have converted through the first advertisement and the second advertisement, to determine the edge weight between the first advertisement node and the second advertisement node.

[0012] Optionally, the apparatus as described in the preceding: the feature module is configured to take the target identifier as an advertisement feature of the target advertisement by assigning the target identifier of the target community node to the target advertisement; the apparatus further comprises a recommendation module, configured to: take the target identifier as an advertisement feature of the target advertisement into an advertisement recommendation system; and in a case where an advertisement acquisition request is acquired and a prediction click rate and / or a prediction conversion rate of the target advertisement pushed to an advertisement position corresponding to the advertisement acquisition request based on the target identifier corresponding to the target advertisement satisfy a preset requirement, push the target advertisement to the advertisement position.

[0013] According to a further aspect of the embodiments of the present application, an electronic device is also provided, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete the communication among each other through the communication bus; wherein the memory is configured to store a computer program; and the processor is configured to execute the method steps in any of the above embodiments by running the computer program stored in the memory.

[0014] According to a further aspect of the embodiments of the present application, a computer readable storage medium is also provided, which stores a computer program, wherein the computer program is configured to execute the method steps in any of the above embodiments when running.

[0015] In the embodiments of the present application, the newly created target advertisement and the target advertisement data of the target advertisement are obtained; based on the target advertisement data, the module gain between the target advertisement and each community node in the target advertisement graph is determined, wherein each historical advertisement in each community node in the target advertisement graph satisfies a preset correlation; the maximum module gain is determined among all the module gains; the community node corresponding to the maximum module gain is determined as the target community node to which the target advertisement belongs; and the target identifier corresponding to the target community is taken as the feature of the target advertisement. Since the historical advertisements in each community node in the target advertisement graph satisfy the preset correlation, that is, the same community node is aggregated by each historical advertisement, and the target identifier of the same target community represents that all the historical interaction features and multi-modal features of the advertisements under the target community satisfy the similarity, therefore, in the subsequent recommendation system, the embedding vector corresponding to the feature of the target identifier can simultaneously learn the commonality of the historical interaction features and multi-modal features of all the advertisements in the community; so that even if the target advertisement is newly generated, the purpose of the historical interaction features of other similar advertisements can be utilized, and thus the problem of lack of historical interaction data of new advertisements in the related art is solved. BRIEF DESCRIPTION OF DRAWINGS

[0016] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and serve to explain the principles of the present application together with the specification.

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows, and obviously, other drawings can also be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0018] Figure 1 is a schematic diagram of a hardware environment of an optional advertisement data processing method according to the embodiments of the present application; Figure 2 is a flow diagram of an optional advertisement data processing method according to an embodiment of the application; Figure 3 is a schematic diagram of an optional advertisement according to an embodiment of the application; Figure 4 is a structural block diagram of an optional advertisement data processing apparatus according to an embodiment of the application; Figure 5 is a structural block diagram of an optional electronic device according to an embodiment of the application. DETAILED DESCRIPTION

[0019] In order to enable persons skilled in the art to better understand the scheme of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by persons skilled in the art without creative labor should fall within the scope of protection of the present application.

[0020] It should be noted that the terms “first”, “second”, and the like in the specification and claims of the present application and in the above-described drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms “include” and “have” and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to only those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to the process, method, product, or device.

[0021] According to an aspect of an embodiment of the present application, an advertisement data processing method is provided. Optionally, in the present embodiment, the above-mentioned advertisement data processing method can be applied in a hardware environment composed of a terminal 1402 and a server 1404 as shown in Figure 1 As shown in Figure 1 , the server 1404 is connected with the terminal 1402 through a network, and can be used to provide services (such as game services, application services, etc.) for the terminal or a client installed on the terminal, and a database can be set on the server or independently of the server, to provide data storage services for the server 1404.

[0022] The aforementioned networks may include, but are not limited to, at least one of the following: wired network, wireless network. The aforementioned wired network may include, but is not limited to, at least one of the following: wide area network (WAN), metropolitan area network (MAN), local area network (LAN). The aforementioned wireless network may include, but is not limited to, at least one of the following: Wi-Fi (Wireless Fidelity), Bluetooth. The terminal is not limited to PCs, mobile phones, tablets, etc.

[0023] The advertising data processing method of this application embodiment can be executed by a server, a terminal, or both. Alternatively, the terminal can execute the advertising data processing method of this application embodiment via a client installed on it.

[0024] Taking the advertising data processing method in this embodiment as an example, which is executed by a server, Figure 2 An advertising data processing method provided in this application includes the following steps: Step S202: Obtain the newly created target ad and the target ad data of the target ad.

[0025] The advertising data processing method in this embodiment can be applied to scenarios where interactive features corresponding to newly created advertisements need to be generated.

[0026] Targeted advertising data can include, but is not limited to: the target audience, target group, copy content, and advertising platform—data that can be obtained without actually running the targeted advertising campaign.

[0027] Step S204: Based on the target ad data, determine the module gain between the target ad and each community node in the target ad graph, wherein the historical ads in each community node of the target ad graph satisfy a preset correlation.

[0028] Specifically, a target advertising map can be obtained in advance, and each community node in the target advertising map can be obtained by aggregating various historical advertisements.

[0029] Furthermore, after determining the target ad data, module gain can be calculated between the target ad data and each community node to determine the module gain between each community node in the target ad graph. Module gain is a key local optimization metric that indicates how much the modularity of the entire network increases after moving a node to another community. In community detection algorithms (e.g., the Louvain community detection algorithm), modularity is a core evaluation metric used to quantify the quality of the network's community structure. It measures how significantly dense the internal connections of the current community partitioning are compared to a completely random network.

[0030] Specifically, the module gain between the target advertisement and each community node in the target advertisement graph can be determined using the following method: For a target advertisement and any community node, merge the target advertisement with each of its neighboring nodes j to obtain multiple node combinations. Calculate the module gain for each node combination and determine the largest module gain. ,like Then the target advertisement will be merged into the community c where node j is located; here It is the sum of the weights of all edges within community c. It is the sum of the weights of all edges connected to nodes in community c (i.e., the sum of the weights of all edges connected to each node in community c). It is the sum of the links connecting the target ad to the nodes in community c (i.e., the sum of the edge weights connecting the target ad to each node in community c). It is the sum of edge weights associated with the target ad, and n is the sum of all edge weights.

[0031] As an alternative implementation, the target advertisement image can be constructed through the following steps, as described above: The original ad graph is constructed by treating each existing ad as an ad node. The edge weight between any two ad nodes is used to indicate the similarity between the two existing ads corresponding to the two ad nodes.

[0032] As an optional implementation, as described above, the method for determining the edge weight between any two ad nodes includes: determining the first multimodal embedding vector of the first ad corresponding to the first ad node; determining the second multimodal embedding vector of the second ad corresponding to the second ad node; determining the cosine similarity between the first and second multimodal embedding vectors; and, based on a preset weight allocation ratio, performing a weighted summation of the cosine similarity, the number of first users who clicked on both the first and second ads, the number of second users who converted together, and the number of third users who converted through deep conversion, to determine the edge weight between the first and second ad nodes.

[0033] In other words, the rich information of the first advertisement is compressed into a fixed-length numerical vector to obtain the first multimodal embedding vector corresponding to the first advertisement node, making it computationally and comparable by the machine. Similarly, the second multimodal embedding vector corresponding to the second advertisement node is obtained. Then, the cosine similarity between the first and second multimodal embedding vectors is calculated to quantify the semantic similarity between the two advertisements (i.e., the first and second advertisements). Since semantic similarity alone is insufficient—two advertisements may have similar content, but user responses may be completely different—user behavior data (i.e., the number of first users who clicked on both the first and second advertisements, the number of second users who converted, and the number of third users who made deep conversions) must be introduced as "post-hoc" verification. The number of users who clicked on both advertisements refers to the number of independent users who clicked on both the first and second advertisements. The number of users who made deep conversions refers to the number of independent users who clicked on both advertisements and made deep conversions. "Deep conversion" can be defined as: high-priced purchases, repeat purchases, membership conversions, etc. For example, each advertisement can be considered as a node. Here, n is the total number of ads, and P is the set of all ads. and edge weight . These are the multimodal embedding vectors obtained from these two advertisements using CLIP. and cosine similarity . , , These are the collaborative information of the two ads: the number of first users who clicked on both ads, the number of second users who converted on both ads, and the number of third users who made a deep conversion on both ads. This refers to the corresponding weights assigned manually based on specific circumstances. The final advertisement image is as follows: Figure 3 As shown.

[0034] The advertising nodes in the original advertising graph are divided into communities using a community detection algorithm to obtain a target advertising graph that includes at least one community node. Each advertising node is assigned a community identifier of the community node it belongs to.

[0035] Specifically, the Louvain community detection algorithm can be used to divide the advertising nodes in the original advertising graph into communities to obtain a target advertising graph that includes at least one community node.

[0036] (1-1) Initialization: Treat each point in the original ad graph as a community; (1-2) Merge each ad node i with other adjacent ad nodes j in turn, and calculate the maximum module gain among them. ,like Then, merge advertising node i into the community c where advertising node j is located; here It is the sum of the weights of all edges within community c. It is the sum of the weights of all edges connected to nodes in community c (i.e., the sum of the weights of all edges connected to each node in community c). It is the sum of the links connecting node i to nodes in community c (i.e., the sum of the edge weights connecting node i to each node in community c). is the sum of edge weights associated with node i, and n is the sum of all edge weights.

[0037] (1-3) Repeat (1-2) until the communities to which all advertising nodes belong no longer change.

[0038] (1-4) Compress all nodes in each community into one node, and combine all edge weights between nodes in two communities as the latest inter-community weight.

[0039] (1-5) Repeat the above steps until all communities no longer change.

[0040] Furthermore, the algorithm obtains the community label that maximizes the module degree increment. Here, we supplement the definition of module degree. .

[0041] Step S206: Determine the largest module gain among all module gains.

[0042] After calculating all module gains, the largest module gain can be determined by sorting all module gains by size.

[0043] Step S208: The community node corresponding to the largest module gain is determined as the target community node to which the target advertisement belongs.

[0044] In other words, the community node with the greatest module gain is directly identified as the target community node to which the target advertisement belongs.

[0045] Optionally, the module gain between the target ad and each community node in the target ad graph can be determined through the following steps: (2-1) For newly created advertisements All calculable ones should be advertisements Edge weight with all other ads And each new ad can be seen as a new community.

[0046] (2-2) Sequentially add each new ad node i (i.e., ad ... The corresponding node is merged with other adjacent nodes j to obtain multiple node combinations. The module gain is calculated for each node combination, and the largest module gain is determined. ,like Then the new advertising node i will be merged into the community c where node j is located; here It is the sum of the weights of all edges within community c. It is the sum of the weights of all edges connected to nodes in community c (i.e., the sum of the weights of all edges connected to each node in community c). It is the sum of the links connecting node i to nodes in community c (i.e., the sum of the edge weights connecting node i to each node in community c). is the sum of edge weights associated with node i, and n is the sum of all edge weights.

[0047] (2-3) Repeat step (2-2) until the communities to which all new nodes belong no longer change.

[0048] (2-4) Compress all nodes in each community into one node, and combine all edge weights between nodes in two communities as the latest inter-community weight.

[0049] (2-5) Repeat the above steps until all communities no longer change.

[0050] This allows us to determine the target community node to which the target advertisement belongs.

[0051] Step S210: Use the target identifier corresponding to the target community as the feature of the target advertisement.

[0052] Once the target community node is identified, the target identifier corresponding to the target community can be used as the feature of the target advertisement. This allows for learning the click-through rate / conversion rate of the target advertisement in the later stage, making full use of its inherent multimodal attributes and the historical interaction characteristics of other historical advertisements to improve the efficiency of recommending new target advertisements.

[0053] In this embodiment, the following steps are taken: acquiring newly created target ads and their target ad data; determining the module gain between the target ad and each community node in the target ad graph, wherein historical ads in each community node of the target ad graph satisfy a preset correlation; determining the largest module gain among all module gains; identifying the community node corresponding to the largest module gain as the target community node to which the target ad belongs; and using the historical interaction features of historical ads in the target community node as the interaction features of the target ad. Since the historical ads in each community node of the target ad graph satisfy a preset correlation, meaning that the same community node is obtained by aggregating various historical ads, and the target identifier of the same target community represents that the historical interaction features and multimodal features of all ads under that target community satisfy similarity, the embedding vector corresponding to the feature of this target identifier in the subsequent recommendation system can simultaneously learn the commonalities of the historical interaction features and multimodal features of all ads in that community. This allows the use of the historical interaction features of other similar ads even if the target ad is newly generated, thus solving the problem of lack of historical interaction data for new ads in related technologies.

[0054] As an optional implementation, the method described above can be implemented by using the target identifier corresponding to the target community as a feature of the target advertisement through the following steps: By assigning the target identifier of a target community node to a target advertisement, the historical interaction features of historical advertisements within the target community node are used as the interaction features of the target advertisement. In other words, each community node has its own identifier (ID). Therefore, the target identifier of a target community node can be assigned to a target advertisement to indicate that the target advertisement is within the target community node. Based on the target identifier assigned to the target advertisement, the embedding vector corresponding to the features of that target identifier in the subsequent recommendation system can simultaneously learn the commonalities of the historical interaction features and multimodal features of all advertisements in that community.

[0055] The method also includes: The target identifier is used as an advertising feature of the target advertisement and is then fed into the advertising recommendation system.

[0056] By using the methods described in the aforementioned embodiments, historical advertisements with similar interaction characteristics are clustered into community nodes, so that each community node represents a set of advertisements with similar user groups or content characteristics. For example, when recommending a target advertisement belonging to a target community, in addition to the target advertisement's own advertising characteristics, the aforementioned target identifier is also used as an input feature of the target advertisement. In feature engineering, this is equivalent to adding a set of "group context features" or "community prior features" to the target advertisement.

[0057] These features (i.e., ad features and historical interaction features) are fed into the ranking model of the ad recommendation system (such as DeepFM, DIN, Graph Neural Network). The model learns that "ads belonging to a high CTR community may have potential, even if they have little data."

[0058] If an ad request is received, and the predicted click-through rate and / or predicted conversion rate of the target ad to the ad slot corresponding to the ad request are determined based on the target identifier of the target ad, the target ad will be pushed to the ad slot.

[0059] When a user triggers an ad display request in a certain scenario (such as watching short videos or browsing shopping websites), the ad retrieval request contains rich contextual information: user characteristics: ID, profile, historical behavior; scenario characteristics: time, location, device, page type; ad placement characteristics: location, size, exposure environment, etc.

[0060] Since the core task of an advertising recommendation system is to predict the performance of each candidate ad under the current request, this can include: predicted click-through rate (pCTR): the probability that the user will click on the ad; and predicted conversion rate (pCVR): the probability that the user will complete a purchase or other target behavior after the ad is clicked. When the predicted click-through rate and / or predicted conversion rate of the target ad in the ad slot corresponding to the ad acquisition request is determined based on the interaction characteristics and ad characteristics of the target ad, and it is determined that the target ad ranks first among all candidate ads (i.e., it meets the preset requirements), the target ad is pushed to the ad slot.

[0061] According to another aspect of the embodiments of this application, an electronic device is also provided, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; wherein the memory is used to store a computer program; and the processor is used to execute the method steps of any of the above embodiments by running the computer program stored in the memory.

[0062] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, wherein a computer program is stored therein, wherein the computer program is configured to execute the method steps of any of the above embodiments when running.

[0063] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0064] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM (Read-Only Memory) / RAM (Random Access Memory), magnetic disk, optical disk), and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0065] According to another aspect of the embodiments of this application, an advertising data processing apparatus for implementing the above-described advertising data processing method is also provided. Figure 4 This is a structural block diagram of an optional advertising data processing apparatus according to an embodiment of this application, such as... Figure 4 As shown, the device may include: Module 41 is used to obtain newly created target ads and target ad data of target ads; The first determining module 42 is used to determine the module gain between the target advertisement and each community node in the target advertisement graph, wherein each historical advertisement in each community node in the target advertisement graph satisfies a preset correlation. The second determining module 43 is used to determine the largest module gain among all module gains; The target community node determination module 44 is used to determine the community node corresponding to the largest module gain as the target community node to which the target advertisement belongs; Feature module 45 is used to use the target identifier corresponding to the target community as the feature of the target advertisement.

[0066] It should be noted that the acquisition module 41 in this embodiment can be used to perform the above step S202, the first determination module 42 in this embodiment can be used to perform the above step S204, the second determination module 43 in this embodiment can be used to perform the above step S206, the target community node determination module 44 in this embodiment can be used to perform the above step S208, and the feature module 45 in this embodiment can be used to perform the above step S210.

[0067] As an optional implementation, the apparatus as described above further includes: The original ad graph construction module is used to construct the original ad graph by treating each existing ad in a plurality of existing ads as an ad node. The edge weight between any two ad nodes is used to indicate the similarity between the two existing ads corresponding to the two ad nodes. The community segmentation module is used to segment the advertising nodes in the original advertising graph into communities using a community discovery algorithm, resulting in a target advertising graph that includes at least one community node, and assigning a community identifier to each advertising node.

[0068] As an optional implementation, the apparatus as described above further includes an edge weight determination module, used for: Determine the first multimodal embedding vector of the first advertisement corresponding to the first advertisement node in the two advertisement nodes; determine the second multimodal embedding vector of the second advertisement corresponding to the second advertisement node in the two advertisement nodes; Determine the cosine similarity between the first multimodal embedding vector and the second multimodal embedding vector; Based on a preset weight allocation ratio, the weights of the first and second ad nodes are determined by weighted summation of cosine similarity, the number of first users who clicked on both the first and second ads, the number of second users who converted together, and the number of third users who converted through deep conversion.

[0069] As an alternative implementation, the apparatus as described above: The feature module is used to assign the target identifier of the target community node to the target advertisement, and use the target identifier as the advertisement feature of the target advertisement. It also includes a recommendation module, which is used to: input the target identifier as an advertising feature of the target advertisement into the advertising recommendation system; and push the target advertisement to the advertising position when an advertisement acquisition request is obtained and the predicted click-through rate and / or predicted conversion rate of the target advertisement to be pushed to the advertising position corresponding to the advertisement acquisition request is determined based on the target identifier corresponding to the target advertisement.

[0070] In addition to the modules described above, the apparatus in this embodiment may also include modules that perform any method in any of the aforementioned advertising data processing methods.

[0071] It should be noted that the examples and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the content disclosed in the above embodiments. It should also be noted that the above modules, as part of a device, can operate in environments such as... Figure 1 The hardware environment shown can be implemented through software or hardware, and the hardware environment includes the network environment.

[0072] According to another aspect of the embodiments of this application, an electronic device for implementing the above-described advertising data processing method is also provided. The electronic device may be a server, a terminal, or a combination thereof.

[0073] According to another embodiment of this application, an electronic device is also provided, comprising: Figure 5 As shown, the electronic device may include: a processor 1501, a communication interface 1502, a memory 1503, and a communication bus 1504, wherein the processor 1501, the communication interface 1502, and the memory 1503 communicate with each other through the communication bus 1504.

[0074] Memory 1503 is used to store computer programs; When processor 1501 executes the program stored in memory 1503, it performs the following steps: Step S202: Obtain the newly created target ad and the target ad data of the target ad.

[0075] Step S204: Based on the target ad data, determine the module gain between the target ad and each community node in the target ad graph, wherein the historical ads in each community node of the target ad graph satisfy a preset correlation.

[0076] Step S206: Determine the largest module gain among all module gains.

[0077] Step S208: The community node corresponding to the largest module gain is determined as the target community node to which the target advertisement belongs.

[0078] Step S210: Use the target identifier corresponding to the target community as the feature of the target advertisement.

[0079] Optionally, in this embodiment, the communication bus can be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. This communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is used to represent it in the figure, but this does not mean that there is only one bus or one type of bus. The communication interface is used for communication between the aforementioned electronic device and other devices.

[0080] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0081] As an example, the memory 1503 described above may include, but is not limited to, the acquisition module 41, the first determination module 42, the second determination module 43, the target community node determination module 44, and the feature module 45 from the aforementioned advertising data processing device. Furthermore, it may include, but is not limited to, other module units from the aforementioned advertising data processing device, which will not be elaborated upon in this example.

[0082] The processor mentioned above can be a general-purpose processor, including but not limited to: CPU (Central Processing Unit), NP (Network Processor), etc.; it can also be DSP (Digital Signal Processor), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0083] This application also provides a computer-readable storage medium, which includes a stored program, wherein the program executes the method steps of the above method embodiments when it runs.

[0084] Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as USB flash drives, ROMs, RAMs, portable hard drives, magnetic disks, or optical disks.

[0085] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0086] If the integrated units in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in the aforementioned computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause one or more computer devices (which may be personal computers, servers, or network devices, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0087] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0088] In the several embodiments provided in this application, it should be understood that the disclosed client can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection between units or modules, and may be electrical or other forms.

[0089] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the solution provided in this embodiment, depending on actual needs.

[0090] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0091] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for processing advertising data, characterized in that, include: Retrieve the newly created target ad and the target ad data of the target ad; Based on the target ad data, determine the module gain between the target ad and each community node in the target ad graph, wherein each historical ad in each community node of the target ad graph satisfies a preset correlation. The largest module gain is determined from all the module gains described. The community node corresponding to the largest module gain is determined as the target community node to which the target advertisement belongs; The target identifier corresponding to the target community is used as a feature of the target advertisement.

2. The method according to claim 1, characterized in that, The method further includes: The original ad graph is constructed by treating each existing ad as an ad node. The edge weight between any two ad nodes is used to indicate the similarity between the two existing ads corresponding to the two ad nodes. The advertising nodes in the original advertising graph are divided into communities using a community discovery algorithm to obtain the target advertising graph that includes at least one community node, and each advertising node is assigned a community identifier of the community node to which it belongs.

3. The method according to claim 2, characterized in that, The method for determining the edge weight between any two advertising nodes includes: Determine the first multimodal embedding vector of the first advertisement corresponding to the first advertisement node among the two advertisement nodes; determine the second multimodal embedding vector of the second advertisement corresponding to the second advertisement node among the two advertisement nodes; Determine the cosine similarity between the first multimodal embedding vector and the second multimodal embedding vector; Based on a preset weight allocation ratio, the cosine similarity, the number of first users who clicked on both the first and second ads, the number of second users who converted together, and the number of third users who converted through deep conversion are weighted and summed to determine the edge weight between the first ad node and the second ad node.

4. The method according to claim 1, characterized in that, The step of using the target identifier corresponding to the target community as a feature of the target advertisement includes: By assigning the target identifier of the target community node to the target advertisement, the target identifier is used as the advertisement feature of the target advertisement; The method further includes: The target identifier is transmitted to the advertising recommendation system as an advertising feature of the target advertisement; If an ad acquisition request is received, and the predicted click-through rate and / or predicted conversion rate of the target ad to the ad slot corresponding to the ad acquisition request is determined based on the target identifier corresponding to the target ad, the target ad will be pushed to the ad slot.

5. An advertising data processing device, characterized in that, include: The acquisition module is used to acquire the newly created target advertisement and the target advertisement data of the target advertisement; The first determining module is used to determine the module gain between the target advertisement and each community node in the target advertisement graph based on the target advertisement data, wherein each historical advertisement in each community node in the target advertisement graph satisfies a preset correlation. The second determining module is used to determine the largest module gain among all the module gains; The target community node determination module is used to determine the community node corresponding to the largest module gain as the target community node to which the target advertisement belongs; The feature module is used to use the target identifier corresponding to the target community as a feature of the target advertisement.

6. The apparatus according to claim 5, characterized in that, Also includes: The original ad graph construction module is used to construct the original ad graph by treating each existing ad in a plurality of existing ads as an ad node, wherein the edge weight between any two ad nodes is used to indicate the similarity between the two existing ads corresponding to the two ad nodes respectively; The community segmentation module is used to segment the advertising nodes in the original advertising graph into communities using a community discovery algorithm, thereby obtaining the target advertising graph that includes at least one community node, and assigning a community identifier to each advertising node.

7. The apparatus according to claim 6, characterized in that, It also includes an edge weight determination module, used for: Determine the first multimodal embedding vector of the first advertisement corresponding to the first advertisement node among the two advertisement nodes; determine the second multimodal embedding vector of the second advertisement corresponding to the second advertisement node among the two advertisement nodes; Determine the cosine similarity between the first multimodal embedding vector and the second multimodal embedding vector; Based on a preset weight allocation ratio, the cosine similarity, the number of first users who clicked on both the first and second ads, the number of second users who converted together, and the number of third users who converted through deep conversion are weighted and summed to determine the edge weight between the first ad node and the second ad node.

8. The apparatus according to claim 5, characterized in that: The feature module is used to assign the target identifier of the target community node to the target advertisement, and use the target identifier as the advertisement feature of the target advertisement. It also includes a recommendation module, used to: input the target identifier as an advertising feature of the target advertisement into the advertising recommendation system; and when an advertisement acquisition request is obtained, and the predicted click-through rate and / or predicted conversion rate of the target advertisement to the advertisement position corresponding to the advertisement acquisition request is determined based on the target identifier corresponding to the target advertisement and meets the preset requirements, push the target advertisement to the advertisement position.

9. An electronic device comprising a processor, a communication interface, a memory, and a communication bus, wherein, The processor, the communication interface, and the memory communicate with each other via the communication bus, characterized in that... The memory is used to store computer programs; The processor is configured to perform the method of any one of claims 1 to 7 by running the computer program stored in the memory.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, wherein the computer program is configured to execute the method described in any one of claims 1 to 7 when run on a processor.

Images