Landing page generation method and device, electronic equipment, medium and program product

Abstract

The present disclosure provides a landing page generation method and device, electronic equipment, medium and program product, relates to the technical field of computers, and particularly relates to the field of information pushing. The specific implementation scheme is as follows: a landing page generation method, comprising: obtaining at least two groups of candidate materials, wherein each group of candidate materials comprises a creative title and at least one multimedia data; generating a first data set based on the at least two groups of candidate materials, wherein the first data set comprises first landing pages formed by each creative title and corresponding target multimedia data in the at least two groups of candidate materials; the target multimedia data corresponding to a first creative title comprises at least one multimedia data corresponding to the first creative title and at least one multimedia data corresponding to a second creative title; and generating a target landing page based on the first data set. The present disclosure can improve the quality of the generated landing page.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and more particularly to the field of information push. Specifically, it relates to a landing page generation method, apparatus, electronic device, medium, and program product. Background Technology

[0002] A landing page is a detail page in the advertising process that carries the advertisement and conveys the advertiser's key information. Among landing pages, the app download landing page is an important conversion form, playing a crucial role in recommending apps that match the user's current interests and preferences, attracting the user to download the app, and completing the user conversion. In related technologies, a default landing page is typically generated uniformly by the push platform and then pushed to users. Summary of the Invention

[0003] This disclosure provides a landing page generation method, apparatus, electronic device, medium, and program product.

[0004] According to a first aspect of this disclosure, a landing page generation method is provided, comprising:

[0005] Obtain at least two sets of candidate materials, wherein each set of candidate materials includes a creative title and at least one multimedia data, and the creative title and the at least one multimedia data correspond to each other;

[0006] A first dataset is generated based on the at least two sets of candidate materials, wherein the first dataset includes a first landing page formed by each creative title and its corresponding target multimedia data in the at least two sets of candidate materials; the target multimedia data corresponding to the first creative title includes at least one multimedia data corresponding to the first creative title, and at least one multimedia data corresponding to the second creative title; the first creative title is a creative title in any one set of candidate materials in the at least two sets of candidate materials, and the second creative title is a creative title in the at least two sets of candidate materials whose semantic relevance to the first creative title is greater than a preset value;

[0007] Generate the target landing page based on the first dataset.

[0008] According to a second aspect of this disclosure, a landing page generation apparatus is provided, comprising:

[0009] The acquisition module is used to acquire at least two sets of candidate materials, wherein each set of candidate materials includes a creative title and at least one multimedia data, and the creative title and the at least one multimedia data correspond to each other;

[0010] A first generation module is configured to generate a first dataset based on the at least two sets of candidate materials, wherein the first dataset includes a first landing page formed by each creative title and its corresponding target multimedia data in the at least two sets of candidate materials; the target multimedia data corresponding to the first creative title includes at least one multimedia data corresponding to the first creative title, and at least one multimedia data corresponding to the second creative title; the first creative title is a creative title in any one set of candidate materials in the at least two sets of candidate materials, and the second creative title is a creative title in the at least two sets of candidate materials whose semantic relevance to the first creative title is greater than a preset value;

[0011] The second generation module is used to generate the target landing page based on the first dataset.

[0012] According to a fifth aspect of this disclosure, an electronic device is provided, comprising:

[0013] At least one processor; and

[0014] A memory communicatively connected to the at least one processor; wherein,

[0015] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method described in the first aspect above.

[0016] According to a sixth aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are configured to cause the computer to perform the method described in the first aspect above.

[0017] According to a seventh aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the method described in the first aspect.

[0018] In this embodiment of the disclosure, when generating the first dataset based on the at least two sets of candidate materials, since the creative title can not only be combined with its corresponding multimedia data to form a landing page, but also combined with the multimedia data of other creative titles with high semantic relevance to form a landing page, this is beneficial to enriching the diversity of the generated landing pages and thus improving the quality of the generated landing pages. Attached Figure Description

[0019] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0020] Figure 1 This is one of the flowcharts of a landing page generation method provided in this embodiment of the disclosure;

[0021] Figure 2 This is a second flowchart of a landing page generation method provided in this embodiment of the disclosure;

[0022] Figure 3 This is a schematic diagram of a landing page generation system provided in an embodiment of this disclosure;

[0023] Figure 4 This is a flowchart illustrating the training process of the semantic conversion model in this embodiment of the disclosure;

[0024] Figure 5 This is a schematic diagram of the structure of a landing page generation device provided in an embodiment of this disclosure;

[0025] Figure 6 This is a schematic diagram of the structure of the second generation module in an embodiment of this disclosure;

[0026] Figure 7 This is a schematic diagram of the structure of the first generation module in an embodiment of this disclosure;

[0027] Figure 8 This is a schematic diagram of the structure of the combined submodule in an embodiment of this disclosure;

[0028] Figure 9 This disclosure provides a block diagram of an electronic device for implementing a landing page generation method. Detailed Implementation

[0029] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0030] Please see Figure 1 , Figure 1 This is a flowchart illustrating a landing page generation method provided in an embodiment of the present disclosure. The landing page generation method includes the following steps:

[0031] Step S101: Obtain at least two sets of candidate materials, wherein each set of candidate materials includes a creative title and at least one multimedia data, and the creative title and the at least one multimedia data correspond to each other;

[0032] Step S102: Generate a first dataset based on the at least two sets of candidate materials, wherein the first dataset includes a first landing page formed by each creative title and its corresponding target multimedia data in the at least two sets of candidate materials; the target multimedia data corresponding to the first creative title includes: at least one multimedia data corresponding to the first creative title, and at least one multimedia data corresponding to the second creative title; the first creative title is a creative title in any one set of candidate materials in the at least two sets of candidate materials, and the second creative title is a creative title in the at least two sets of candidate materials whose semantic relevance to the first creative title is greater than a preset value;

[0033] Step S103: Generate the target landing page based on the first dataset.

[0034] Among them, the aforementioned candidate materials can be materials provided by users who need to deliver landing pages, for example, materials provided by advertisers.

[0035] The aforementioned landing page can be of various types, such as an application download landing page. Specifically, the download landing page can be pushed to a user while they are using an application. After the user clicks on the download landing page, they can be redirected to a download link for the target application. The user can then download the target application through this link, thereby increasing the user conversion rate for application downloads. The aforementioned landing page generation method can be applied to a landing page generation platform, or a landing page push platform, etc. The following explanation uses the application of the landing page generation method to a landing page generation platform as an example to further illustrate the landing page generation method provided in this embodiment.

[0036] Each of the above creative titles can include one or more multimedia data, such as images or videos. The correspondence between the creative title and the multimedia data means that the content of the multimedia data matches the content indicated by the creative title. For example, when the creative title of a candidate material is "How to Make Braised Pork," the multimedia data in that candidate material can include: a video tutorial on how to make braised pork and images of braised pork with the necessary ingredients, etc.

[0037] Before generating the first dataset based on the at least two sets of candidate materials, the semantics of the creative titles in each candidate material can be identified to determine the semantic relevance between different creative titles. For example, the creative titles can be converted into corresponding semantic vectors, and then the semantic relevance between different creative titles can be determined by calculating the vector distance between different semantic vectors. In this case, the magnitude of the semantic relevance is the magnitude of the vector distance. In this way, the associated creative titles corresponding to each creative title can be determined, wherein the associated creative titles are those whose semantic relevance to the creative title is greater than a preset value.

[0038] Then, the creative title is combined with each corresponding multimedia data to form a landing page. It is understood that the creative title can be combined with its corresponding multimedia content to form the first landing page; additionally, the creative title can also be combined with the multimedia content corresponding to an associated creative title to form the first landing page, thus enriching the format of the resulting landing page.

[0039] In one embodiment of this disclosure, the at least two candidate materials include a first candidate material and a second candidate material. The first candidate material includes: the aforementioned first creative title and two pieces of first multimedia data, wherein the first creative title is: "How to Make Braised Chicken," and the first multimedia data is an image of the braised chicken recipe. The second candidate material includes: the aforementioned second creative title and three pieces of second multimedia data, wherein the second creative title is: "How to Make Braised Duck," and the second multimedia data is a video tutorial on how to make braised duck. If, through the above identification, it is determined that the semantic relevance between the first creative title and the second creative title is greater than a preset value, the process of generating a first dataset based on the at least two sets of candidate materials may include: combining the first creative title with each of the two pieces of first multimedia data to form two first landing pages; simultaneously, combining the first creative title with each of the three pieces of second multimedia data to form three first landing pages; combining the second creative title with each of the two pieces of first multimedia data to form two first landing pages; simultaneously, combining the second creative title with each of the three pieces of second multimedia data to form three first landing pages. In this way, by generating a first landing page based on each creative title and its corresponding target multimedia data, all the generated first landing pages can be combined to form the first dataset.

[0040] The number of associated creative titles corresponding to the creative title can be N, where N is an integer greater than or equal to 0. The specific process of forming the first landing page from the aforementioned creative title and its corresponding target multimedia data can be as follows: Obtain a pre-created landing page template, wherein the landing page template has a first replacement area and a second replacement area; then, integrate the creative title into the first replacement area and integrate the target multimedia data into the second replacement area, thereby completing the generation process of the first landing page.

[0041] Specifically, generating a target landing page based on the first dataset can mean selecting a landing page with a high conversion rate from the first dataset as the target landing page to improve the conversion rate of the generated landing page. Alternatively, different first landing pages from the first dataset can be pushed to different users. This allows for landing page recommendations tailored to the specific characteristics of each user, further improving the effectiveness of landing page recommendations.

[0042] In this embodiment, when generating the first dataset based on the at least two sets of candidate materials, since the creative title can not only be combined with its corresponding multimedia data to form a landing page, but also with the multimedia data of other creative titles with high semantic relevance to form a landing page, this helps to enrich the diversity of the generated landing pages and thus improve the quality of the generated landing pages.

[0043] Optionally, generating the target landing page based on the first dataset includes:

[0044] Obtain the user profile information and historical behavior log information of the first user, wherein the historical behavior log information is log information generated based on the first user's behavior of clicking on the landing page within a preset time period;

[0045] The user profile information, the historical behavior log information, and the target feature information of the landing page in the first dataset are input into the prediction model for prediction to obtain the target landing page corresponding to the first user. The target landing page is the landing page in the first dataset that matches the first user.

[0046] The first user can be a user of a specific application platform. User conversion can be achieved by pushing a target landing page to the first user. It is understood that the first user can be one of the user accounts on the application platform, and the user can access the application platform based on that user account. The aforementioned target landing page can be a landing page published by advertisers on other application platforms besides the stated application platform, in order to attract users to download applications from those other application platforms.

[0047] The aforementioned user profile information may include the identity attribute information of the first user, such as the first user's gender, age, and other identity attribute information.

[0048] The aforementioned historical behavior log information can be log information generated by the first user's click behavior on recommended landing pages during access to the application platform within a preset time period. The click behavior refers to the user clicking to view the detailed content of a landing page when the application platform pushes it to the user. The preset time period can be a historical period such as the past week or the past month. The historical behavior log information can include information related to the landing page clicked by the first user, the time information of the first user clicking the landing page, and whether the first user downloaded the corresponding application based on the landing page.

[0049] It is understood that the aforementioned prediction model can predict the target landing page that the first user in the first dataset might click on based on the user profile information and historical behavior log information. Thus, the determined target landing page can be subsequently pushed to the first user to improve advertising conversion rates. The first user can be any user on the application platform; the user profile information and historical behavior data obtained for different users on the application platform are usually different. The following explanation uses the example of pushing a landing page to a specific user (the first user) on the application platform to further illustrate the landing page generation method provided in this embodiment.

[0050] In one embodiment of this disclosure, an initial model can be pre-trained based on a large amount of first training data to obtain the prediction model. The first training data may include user profile features, historical behavior logs, and target feature information of the landing page clicked by the user. Thus, the prediction model can learn the correlation between user profile features, historical behavior logs, and the target feature information of the clicked landing page. Subsequently, when the prediction model receives a set of data to be predicted, it can determine whether the user profile features, historical behavior logs, and target feature information in the data to be predicted match, thereby determining the probability that the user clicked the landing page corresponding to that target feature information. The initial model can be built based on a large-scale deep neural network (DNN) model, and then distributed training can be performed using a multi-machine, multi-GPU approach to model hundreds of millions of discrete features.

[0051] Please see Figure 2In this embodiment, user profile features and historical behavior log features can be extracted through feature engineering on the offline side, and the extracted features are transmitted to a feature library. Simultaneously, the online test transmits the target features of the landing page from the first dataset to the feature library. Then, the connection layer inputs the user profile features, historical behavior logs, and target feature information from the feature library into the prediction model for prediction. The online landing page selection service of the online test determines the target landing page corresponding to the first user based on the prediction result of the prediction model. The online test can transmit the target landing page to the offline side for storage, so that the target landing page can be pushed to the first user when the first user accesses the application platform.

[0052] Among them, big data processing tools such as Hadoop and Spark can be used to process the data in the feature library, and a streaming feature extraction module can be built to complete the transformation of massive data into discrete features. Feature engineering can be used to refine the features.

[0053] In this implementation, the probability of a first user clicking on each landing page in the first dataset is predicted using user profile information and historical behavior log information. The landing page with the highest probability is then determined as the target landing page for the first user. Thus, compared to uniformly pushing the same landing page to all users, this embodiment of the disclosure allows for individual prediction of the target landing page for each user, taking into greater consideration each user's preferences and pushing a corresponding target landing page to each user. This helps to further improve the matching degree between the generated landing page and the user, thereby improving the conversion rate of the landing page.

[0054] Optionally, the target landing page includes at least two landing pages, each corresponding to at least two creative titles.

[0055] Specifically, since the first dataset includes landing pages with multiple creative titles, when generating the target landing page for the first user, a landing page matching the first user can be generated for each creative title. This allows for pushing landing pages with different themes to users in different scenarios. For example, when a user is watching a food-related short video, a landing page with the creative title "Food Recipe" can be pushed to the user; when a user is watching a travel-related short video, a landing page with the creative title "Travel Related" can be pushed to the user.

[0056] It is understandable that generating a landing page that matches the first user for each creative title means that the first user is most likely to click on the landing page for the corresponding creative title.

[0057] In this implementation, a landing page matching the first user is generated for each creative title. In this way, different landing pages can be pushed to the first user in different scenarios, which helps to further improve the user conversion rate.

[0058] Optionally, the at least two sets of candidate materials are materials associated with the second user, and the first dataset also includes a pre-created second landing page matched with the second user.

[0059] The second user can be a user who needs to submit a landing page, such as an advertiser. It is understood that each advertiser on the landing page generation platform can generate a landing page using the method described in this embodiment.

[0060] In this embodiment, the second user can customize the materials to be delivered, thus generating corresponding landing pages according to the advertiser's own requirements, thereby meeting the specific needs of different advertisers. Furthermore, the second user can also create a second landing page. Based on the second landing page created by the second user and the first landing page generated from the candidate materials provided by the second user, the first dataset is generated. Then, the target landing page that the first user might click on is predicted from the first dataset.

[0061] Furthermore, since the materials provided by different users are often of varying quality, the landing page generation platform can, after receiving materials uploaded by the second user, first filter the materials based on historical conversion behavior log data to obtain at least two candidate materials. The historical conversion behavior log data can be obtained by analyzing the conversion behavior of historical landing pages to determine the types of landing pages with better conversion performance, and then filtering the materials uploaded by the second user to obtain at least two candidate materials.

[0062] In this implementation, since the second user can customize the materials used to generate the first landing page, and can also pre-create the second landing page, it is beneficial to generate the corresponding landing page according to their own needs, thereby meeting the specific needs of different advertisers.

[0063] Optionally, the target feature information includes feature information of common features of the first landing page and the second landing page.

[0064] Wherein, when both the first landing page and the second landing page include: creative title, creative image, and landing page template, the template feature information may include the feature information of the creative title, the feature information of the creative image, and the feature information of the landing page template.

[0065] In this embodiment, since the first landing page is automatically generated by the landing page generation platform, and the second landing page is pre-created by the second user, when making predictions, the feature information of the common features of the first landing page and the second landing page can be taken as the target feature information. In this way, it can be ensured that the prediction model can avoid prediction deviations due to the differences in features between the first and second landing pages, which is conducive to achieving unbiased prediction of the model.

[0066] Optionally, generating the first dataset based on the at least two sets of candidate materials includes:

[0067] The creative titles from the at least two groups of candidate materials are input into the semantic transformation model for semantic transformation to obtain at least two semantic vectors, wherein one creative title corresponds to one semantic vector;

[0068] Based on the vector distance between the semantic vectors, determine the semantic relevance between any two creative titles in the creative titles of the at least two sets of candidate materials;

[0069] The first dataset is obtained by combining each creative title with its corresponding target multimedia data.

[0070] The aforementioned semantic transformation model can be any model in related technologies that can convert text into semantic vectors.

[0071] Please see Figure 3 In one embodiment of this disclosure, to improve the semantic conversion effect of the semantic conversion model, so that texts with similar semantics have a smaller vector distance after being converted into semantic vectors, this embodiment further improves the training process of the semantic conversion model. Specifically, the training process of the semantic conversion model includes the following steps:

[0072] First, an initial semantic transformation model is constructed, which includes an encoding module (ERNIE) and a graph aggregation module (TransformerSage). The encoding module is used to encode the received text and to receive the encoding from the encoding module to generate a semantic vector corresponding to the text.

[0073] In the pre-training phase, the initial semantic conversion model can be pre-trained based on a large amount of second training data to obtain a first intermediate model. The second training data can be training data generated based on search terms and their corresponding random walk sequences. The random walk sequence is a random walk sequence constructed based on user clicks. The random walk sequence is used to indicate the order in which a user browses pages. For example, in one embodiment of this disclosure, in a search scenario, if the generated random walk sequence is ABCD during a query based on search term 'a', it indicates that the user first browsed webpage A, then triggered webpage B in webpage A, then triggered webpage C in webpage B, and finally triggered webpage D in webpage C. Since webpages A, B, C, and D are all results obtained based on search term 'a', there is usually a semantic association between search term 'a' and the creative title of each webpage. Therefore, search term 'a' can be used to form a second training data set with the creative titles of webpages A, B, C, and D respectively, resulting in four sets of second training data. Meanwhile, since the next webpage is triggered from the previous webpage in any two adjacent webpages in ABCD, there is usually a semantic relationship between the creative titles of two adjacent webpages. Therefore, a second set of training data can be formed by combining the creative titles of any two adjacent webpages in ABCD.

[0074] Since each second training data includes two semantically similar texts, the two semantically similar texts can be encoded separately using ERNIE to obtain two initial semantic vectors. Then, TransformerSage uses a pre-training task (Mask Language Mode Pretrain) to narrow the vector distance between the two initial semantic vectors, so that when the first intermediate model transforms semantically similar texts, it can transform them into similar semantic spaces.

[0075] In the fine-tuning phase, to further improve the model's semantic conversion performance on landing page creative titles, a large amount of third training data can be pre-constructed. Based on this third training data, the first intermediate model is further trained to obtain the aforementioned semantic conversion model. Specifically, a large number of landing page creative titles can be acquired. Then, semantically similar creative titles are identified, and edges are set between them. Finally, pairs of creative titles with edges are grouped into third training data. Since the two creative titles in the second training data are semantically similar, the first intermediate model can further learn the relationships between semantically similar creative titles and convert the semantic vectors of the creative titles into a similar semantic space. This helps to further improve the accuracy of the semantic conversion model's semantic conversion of creative titles, thus completing the training process of the semantic conversion model.

[0076] After inputting the creative titles from the at least two sets of candidate materials into the semantic conversion model for semantic conversion to obtain at least two semantic vectors, the vector distance between any two vectors in the at least two semantic vectors can be calculated, and the creative titles corresponding to the two semantic vectors with smaller vector distances can be determined as creative titles with semantic relevance less than a preset value. For example, the creative titles corresponding to the two semantic vectors with semantic vectors less than 0.5 can be determined as creative titles with semantic relevance less than a preset value.

[0077] Specifically, the above vector calculation process can be completed based on a pre-configured task. In one embodiment of this disclosure, an Artificial Neural Network (ANN) is used to perform database retrieval on the at least two semantic vectors and recall the associated creative titles corresponding to each creative title. That is, the ANN determines the vector distance between any two adjacent semantic vectors and outputs creative titles with semantic relationships.

[0078] In this embodiment, the creative titles from at least two sets of candidate materials are input into a semantic transformation model for semantic transformation to obtain at least two semantic vectors. Then, based on the vector distance between the semantic vectors, the semantic relevance between any two creative titles in the at least two sets of candidate materials is determined. Each creative title is then combined with its corresponding target multimedia data to obtain the first dataset. This enriches the diversity of the generated landing pages and thus helps to improve the quality of the generated landing pages.

[0079] Optionally, the step of combining each creative title with its corresponding target multimedia data to obtain the first dataset includes:

[0080] The second dataset is obtained by combining each creative title with its corresponding target multimedia data.

[0081] Filter the third landing page in the second dataset to obtain the first dataset, wherein the third landing page is the first landing page whose creative title and multimedia data do not match.

[0082] In this implementation, since some of the numerous first landing pages generated by the landing page generation platform may have poor conversion rates, the first landing pages in the second dataset can be filtered based on historical conversion behavior data to remove those with poor conversion rates. This ensures that the first landing pages stored in the first dataset are those with good conversion rates. This reduces the computational load in subsequent prediction processes.

[0083] Please see Figure 4 This is a schematic diagram of a landing page generation system provided in an embodiment of the present disclosure. The system includes an offline material construction module, a material recall module, and a landing page download generation module. In the offline material construction module: the material library is used to store a large number of materials uploaded by users. The material library can be used to determine the materials with better conversion effects based on the conversion logs, and the materials with better conversion effects in the material library can be stored in the database. The data block is used to store the at least two candidate materials in the above embodiment.

[0084] The aforementioned material retrieval module may include a template generation submodule and a material retrieval module, used to pre-generate a landing page template based on the template generation submodule. Simultaneously, the material retrieval module may include the aforementioned semantic transformation model and ANN. The first dataset can be generated using at least two candidate data points from the data block and the landing page template based on the material retrieval module.

[0085] The aforementioned download landing page generation module is used to receive the first dataset recalled by the material recall module based on the prediction model, and generate the target landing page.

[0086] Please see Figure 5 This is a schematic diagram of the structure of a landing page generation device 500 provided in an embodiment of the present disclosure. The landing page generation device 500 includes:

[0087] The acquisition module 501 is used to acquire at least two sets of candidate materials, wherein each set of candidate materials includes a creative title and at least one multimedia data, and the creative title and the at least one multimedia data correspond to each other;

[0088] The first generation module 502 is configured to generate a first dataset based on the at least two sets of candidate materials, wherein the first dataset includes a first landing page formed by each creative title and its corresponding target multimedia data in the at least two sets of candidate materials; the target multimedia data corresponding to the first creative title includes at least one multimedia data corresponding to the first creative title, and at least one multimedia data corresponding to the second creative title; the first creative title is a creative title in any one set of candidate materials in the at least two sets of candidate materials, and the second creative title is a creative title in the at least two sets of candidate materials whose semantic relevance to the first creative title is greater than a preset value;

[0089] The second generation module 503 is used to generate a target landing page based on the first dataset.

[0090] Optionally, please see Figure 6 The second generation module 503 includes:

[0091] The acquisition submodule 5031 is used to acquire the user profile information and historical behavior log information of the first user. The historical behavior log information is log information generated based on the behavior of the first user clicking on the landing page within a preset time period.

[0092] The prediction submodule 5032 is used to input the user profile information, the historical behavior log information and the target feature information of the landing page in the first dataset into the prediction model for prediction, so as to obtain the target landing page corresponding to the first user, wherein the target landing page is the landing page in the first dataset that matches the first user.

[0093] Optionally, the target landing page includes at least two landing pages, each corresponding to at least two creative titles.

[0094] Optionally, the at least two sets of candidate materials are materials associated with the second user, and the first dataset also includes a pre-created second landing page matched with the second user.

[0095] Optionally, the target feature information includes feature information of common features of the first landing page and the second landing page.

[0096] Optionally, please see Figure 7 The first generation module 502 includes:

[0097] The conversion submodule 5021 is used to input the creative titles in the at least two groups of candidate materials into the semantic conversion model for semantic conversion to obtain at least two semantic vectors, wherein one creative title corresponds to one semantic vector;

[0098] The determination submodule 5022 is used to determine the semantic relevance between any two creative titles in the at least two sets of candidate materials based on the vector distance between the semantic vectors.

[0099] The combination submodule 5023 is used to combine each creative title with each corresponding target multimedia data to obtain the first dataset.

[0100] Optionally, please see Figure 8 The combined submodule includes:

[0101] Combination unit 50231 is used to combine each creative title with each corresponding target multimedia data to obtain a second dataset;

[0102] The filtering unit 50232 is used to filter the third landing page in the second dataset to obtain the first dataset, wherein the third landing page is the first landing page whose creative title and multimedia data do not match.

[0103] It should be noted that the landing page generation device 500 provided in this embodiment can realize all the technical solutions of the above landing page generation embodiment, and therefore can at least realize all the above technical effects, which will not be repeated here.

[0104] The acquisition, storage, and application of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0105] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0106] Figure 9 A schematic block diagram of an example electronic device 900 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0107] like Figure 9 As shown, the electronic device 900 includes a computing unit 901, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 902 or a computer program loaded from a storage unit 908 into a random access memory (RAM) 903. The RAM 903 may also store various programs and data required for the operation of the device 900. The computing unit 901, ROM 902, and RAM 903 are interconnected via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0108] Multiple components in electronic device 900 are connected to I / O interface 905, including: input unit 906, such as keyboard, mouse, etc.; output unit 907, such as various types of displays, speakers, etc.; storage unit 908, such as disk, optical disk, etc.; and communication unit 909, such as network card, modem, wireless transceiver, etc. Communication unit 909 allows device 900 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0109] The computing unit 901 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 901 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 901 performs the various methods and processes described above, such as the landing page generation method. For example, in some embodiments, the landing page generation method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 908. In some embodiments, part or all of the computer program may be loaded and / or installed on device 900 via ROM 902 and / or communication unit 909. When the computer program is loaded into RAM 903 and executed by the computing unit 901, one or more steps of the landing page generation method described above are performed. Alternatively, in other embodiments, the computing unit 901 may be configured to perform the landing page generation method by any other suitable means (e.g., by means of firmware).

[0110] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0111] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0112] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0113] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0114] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0115] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0116] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0117] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for generating a landing page, comprising: Obtain at least two sets of candidate materials, wherein each set of candidate materials includes a creative title and at least one multimedia data, and the creative title and the at least one multimedia data correspond to each other; A first dataset is generated based on the at least two sets of candidate materials, wherein the first dataset includes a first landing page formed by each creative title and its corresponding target multimedia data in the at least two sets of candidate materials; the target multimedia data corresponding to the first creative title includes at least one multimedia data corresponding to the first creative title, and at least one multimedia data corresponding to the second creative title; the first creative title is a creative title in any one set of candidate materials in the at least two sets of candidate materials, and the second creative title is a creative title in the at least two sets of candidate materials whose semantic relevance to the first creative title is greater than a preset value; Generate the target landing page based on the first dataset; The step of generating the target landing page based on the first dataset includes: Obtain the user profile information and historical behavior log information of the first user, wherein the historical behavior log information is log information generated based on the first user's behavior of clicking on the landing page within a preset time period; The user profile information, the historical behavior log information, and the target feature information of the landing page in the first dataset are input into the prediction model for prediction to obtain the target landing page corresponding to the first user. The target landing page is the landing page in the first dataset that matches the first user.

2. The method according to claim 1, wherein, The target landing pages include at least two landing pages, each corresponding to at least two creative titles.

3. The method according to claim 1, wherein, The at least two sets of candidate materials are materials associated with the second user, and the first dataset also includes a pre-created second landing page matched with the second user.

4. The method according to claim 3, wherein, The target feature information includes feature information of common features of the first landing page and the second landing page.

5. The method according to claim 1, wherein, The generation of the first dataset based on the at least two sets of candidate materials includes: The creative titles from the at least two groups of candidate materials are input into the semantic transformation model for semantic transformation to obtain at least two semantic vectors, wherein one creative title corresponds to one semantic vector; Based on the vector distance between the semantic vectors, determine the semantic relevance between any two creative titles in the creative titles of the at least two sets of candidate materials; The first dataset is obtained by combining each creative title with its corresponding target multimedia data.

6. The method according to claim 5, wherein, The first dataset is obtained by combining each creative title with its corresponding target multimedia data, including: The second dataset is obtained by combining each creative title with its corresponding target multimedia data. Filter the third landing page in the second dataset to obtain the first dataset, wherein the third landing page is the first landing page whose creative title and multimedia data do not match.

7. A landing page generation device, comprising: The acquisition module is used to acquire at least two sets of candidate materials, wherein each set of candidate materials includes a creative title and at least one multimedia data, and the creative title and the at least one multimedia data correspond to each other; A first generation module is configured to generate a first dataset based on the at least two sets of candidate materials, wherein the first dataset includes a first landing page formed by each creative title and its corresponding target multimedia data in the at least two sets of candidate materials; the target multimedia data corresponding to the first creative title includes at least one multimedia data corresponding to the first creative title, and at least one multimedia data corresponding to the second creative title; the first creative title is a creative title in any one set of candidate materials in the at least two sets of candidate materials, and the second creative title is a creative title in the at least two sets of candidate materials whose semantic relevance to the first creative title is greater than a preset value; The second generation module is used to generate a target landing page based on the first dataset; The second generation module includes: The acquisition submodule is used to acquire the user profile information and historical behavior log information of the first user. The historical behavior log information is log information generated based on the behavior of the first user clicking on the landing page within a preset time period. The prediction submodule is used to input the user profile information, the historical behavior log information, and the target feature information of the landing page in the first dataset into the prediction model for prediction, so as to obtain the target landing page corresponding to the first user. The target landing page is the landing page in the first dataset that matches the first user.

8. The apparatus according to claim 7, wherein, The target landing pages include at least two landing pages, each corresponding to at least two creative titles.

9. The apparatus according to claim 7, wherein, The at least two sets of candidate materials are materials associated with the second user, and the first dataset also includes a pre-created second landing page matched with the second user.

10. The apparatus according to claim 9, wherein, The target feature information includes feature information of common features of the first landing page and the second landing page.

11. The apparatus according to claim 7, wherein, The first generation module includes: The conversion submodule is used to input the creative titles from the at least two groups of candidate materials into the semantic conversion model for semantic conversion to obtain at least two semantic vectors, wherein one creative title corresponds to one semantic vector; The determination submodule is used to determine the semantic relevance between any two creative titles in the at least two sets of candidate materials based on the vector distance between the semantic vectors. The combination submodule is used to combine each creative title with each corresponding target multimedia data to obtain the first dataset.

12. The apparatus according to claim 11, wherein, The combined submodule includes: The combination unit is used to combine each creative title with each corresponding target multimedia data to obtain a second dataset; A filtering unit is used to filter the third landing page in the second dataset to obtain the first dataset, wherein the third landing page is the first landing page whose creative title and multimedia data do not match.

13. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the landing page generation method according to any one of claims 1-6.

14. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to execute the landing page generation method according to any one of claims 1-6.

15. A computer program product comprising a computer program that, when executed by a processor, implements the landing page generation method according to any one of claims 1-6.

Images