Landing page copy adding method and device, model training method and device, and electronic equipment

Abstract

The present disclosure provides a landing page copy adding method and device, a model training method and device, and electronic equipment, relates to the technical field of artificial intelligence, in particular to the technical field of computer vision, deep learning, NLP and the like. The specific implementation scheme is: extracting first feature information of a landing page; performing a copy generation operation on the landing page based on the first feature information to obtain an operation result; in the case that the operation result indicates that no valid copy of the landing page is obtained, obtaining a pocket copy of the landing page based on at least one of second feature information and industry information of the landing page; and adding a copy on the landing page based on a candidate copy, wherein in the case that the operation result indicates that no valid copy of the landing page is obtained, the candidate copy includes the pocket copy. The present disclosure can improve the copy effect of the landing page.

Description

Technical Field

[0001] This disclosure relates to the fields of artificial intelligence technology, such as computer vision, deep learning, and natural language processing (NLP), and in particular to a method for adding landing page copy, a model training method, a device, and an electronic device. Background Technology

[0002] A landing page is a page that displays information to users. Currently, some landing pages can display some text, but the text on landing pages is usually manually edited before the landing page is generated. Specifically, the user edits the text on the computer and sends it to the landing page's server. Summary of the Invention

[0003] This disclosure provides a method for adding landing page copy, a model training method, a device, and an electronic device.

[0004] According to one aspect of this disclosure, a method for adding landing page copy is provided, including:

[0005] Extract the primary feature information of the landing page;

[0006] Based on the first feature information, a copywriting generation operation is performed on the landing page to obtain the operation result;

[0007] If the operation result indicates that no valid copy of the landing page has been obtained, the pocket copy of the landing page shall be obtained based on at least one of the second feature information of the landing page and industry information.

[0008] Based on the candidate copy, add copy to the landing page, wherein, if the operation result indicates that no valid copy for the landing page is obtained, the candidate copy includes the copy in the pocket.

[0009] According to one aspect of this disclosure, a model training method is provided, comprising:

[0010] Obtain source text and target text, wherein the source text includes at least one of the following: source text obtained by prepending business information related to the landing page of the first sample to the first source training sample, and source text obtained by prepending regional information related to the landing page of the second sample to the second source training sample; wherein the first source training sample and the second source training sample are pre-set training samples; and / or, the target text includes: target text obtained by filling the title of a multimedia creative with keywords that meet preset conditions, wherein the multimedia creative is a multimedia creative found through a search engine, and the title of the creative is the title of the multimedia creative;

[0011] A preset network model is trained based on the source text and the target text to obtain a text generation model; wherein, the text generation model is used to generate the text of the landing page based on the feature information of the landing page.

[0012] According to another aspect of this disclosure, a landing page copy adding device is provided, comprising:

[0013] The extraction module is used to extract the first feature information of the landing page;

[0014] The generation module is used to perform a copy generation operation on the landing page based on the first feature information and obtain the operation result;

[0015] The acquisition module is used to acquire the landing page's "pocket copy" based on at least one of the second feature information and industry information of the landing page when the operation result indicates that no valid copy of the landing page has been acquired.

[0016] An add module is used to add text to the landing page based on candidate text, wherein, if the operation result indicates that no valid text is obtained for the landing page, the candidate text includes the text in the pocket.

[0017] According to another aspect of this disclosure, a model training apparatus is provided, comprising:

[0018] The acquisition module is used to acquire source text and target text, wherein the source text includes at least one of the following: source text obtained by prepending business information related to the landing page of the first sample to the first source training sample, and source text obtained by prepending regional information related to the landing page of the second sample to the second source training sample; wherein the first source training sample and the second source training sample are pre-set training samples; and / or, the target text includes: target text obtained by filling the title of a multimedia creative with keywords that meet preset conditions, wherein the multimedia creative is a multimedia creative found through a search engine, and the title of the creative is the title of the multimedia creative;

[0019] The training module is used to train a preset network model based on the source text and the target text to obtain a text generation model; wherein, the text generation model is used to generate the text of the landing page based on the feature information of the landing page.

[0020] According to another aspect of this disclosure, an electronic device is provided, comprising:

[0021] At least one processor; and

[0022] A memory communicatively connected to the at least one processor; wherein,

[0023] The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the landing page copy addition method or the model training method provided in this disclosure.

[0024] According to another aspect of this disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause the computer to execute the landing page copy addition method or the model training method provided in this disclosure.

[0025] According to another aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the landing page copy addition method or model training method provided in this disclosure.

[0026] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0027] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0028] Figure 1 This is a flowchart of a publicly provided method for adding landing page copy;

[0029] Figure 2 This is a flowchart of a model training method provided in this publication;

[0030] Figure 3 This is a diagram illustrating a method for adding landing page copy provided in this public document;

[0031] Figures 4a to 4c This is a structural diagram of the landing page copy adding device provided in this public document;

[0032] Figure 5 This is a structural diagram of a model training device disclosed herein;

[0033] Figure 6 This is a block diagram of an electronic device used to implement embodiments of the present disclosure. Detailed Implementation

[0034] 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.

[0035] Please see Figure 1 , Figure 1 This is a flowchart of a publicly provided method for adding landing page copy, such as... Figure 1 As shown, it includes the following steps:

[0036] Step S101: Extract the first feature information of the landing page.

[0037] The aforementioned landing page can be an advertising landing page or a landing page displaying other multimedia information, such as a landing page displaying corporate promotional information.

[0038] The first feature information extracted from the landing page can be at least one feature information extracted from the text and image of the landing page.

[0039] Step S102: Perform a copy generation operation on the landing page based on the first feature information to obtain the operation result.

[0040] The aforementioned copy generation operation based on the first feature information for the landing page can be performed by using a pre-acquired text generation model to perform copy generation on the landing page based on the first feature information, that is, by using the text generation model to predict the copy for the landing page. Alternatively, the aforementioned copy generation operation based on the first feature information for the landing page can be performed by concatenating the text information in the first feature information into a single copy.

[0041] It should be noted that the results of the above operations may include any of the following:

[0042] The landing page copy was not generated. For example, the effective content of the first feature information was insufficient, which resulted in the inability to generate the landing page copy.

[0043] The generated copy for the above landing page is generated, but the generated copy does not meet the preset conditions, such as the core keywords in the generated copy not being in the above landing page, or the brand in the generated copy not matching the above landing page, etc.

[0044] The generated candidate text for the landing page is considered valid.

[0045] Step S103: If the operation result indicates that no valid copy of the landing page has been obtained, obtain the pocket copy of the landing page based on at least one of the second feature information and industry information of the landing page.

[0046] The above operation results may indicate that no valid copy was obtained for the landing page for several reasons: The copy does not meet preset conditions, such as the core keywords in the generated copy not being present on the landing page, or the brand in the generated copy not matching the landing page; or, there is no copy in the above operation results, meaning the copy generation operation did not generate any copy; or, the number of candidate copy in the above operation results does not reach the preset number; or, the generated copy does not cover the landing page.

[0047] The aforementioned valid copy can be copy that meets preset conditions, such as the core keywords being within the aforementioned landing page, or the brand in the copy matching the aforementioned landing page, etc. Alternatively, the aforementioned valid copy can be copy of a preset quantity, meaning that only copy that has reached a preset quantity is valid.

[0048] The second feature information may be at least one of the text and image of the landing page, and the second feature information may have an intersection with the first feature information, or the second feature information and the first feature information may be different feature information.

[0049] The industry information above indicates the industry to which the landing page belongs, such as: decoration industry, automotive industry, legal industry, etc.

[0050] The aforementioned method of obtaining the pocket copy of the landing page based on at least one of the second feature information and industry information of the landing page can be: generating the pocket copy of the landing page based on the second feature information of the landing page; obtaining copy from the industry to which the landing page belongs based on the industry information as the pocket copy; or the pocket copy includes copy generated based on the second feature information and copy obtained based on the industry information.

[0051] Step S104: Based on the candidate copy, add copy to the landing page, wherein, if the operation result indicates that no valid copy for the landing page is obtained, the candidate copy includes the copy in the pocket.

[0052] The aforementioned candidate copy, including the copy in the pocket, may be the copy in the pocket; or, in addition to the copy in the pocket, the aforementioned candidate copy may also include the copy in the aforementioned operation result. In this case, the number of copy in the aforementioned operation result does not reach the preset number corresponding to the aforementioned valid copy. That is to say, although the aforementioned operation result contains copy, it indicates that no valid copy for the landing page has been obtained because the number of copy does not meet the standard.

[0053] In some implementations, if the above operation result indicates that a valid landing page text has been obtained, the above candidate text includes the valid text text in the above operation result.

[0054] It should be noted that the above candidate text can be one or more texts, and the text added to the landing page can be one or more texts.

[0055] Adding text to the landing page can be done in the form of a pop-up or a guide overlay, where the guide overlay can also be called a guide mask.

[0056] In this disclosure, when a valid copy for the landing page is not obtained, a pocket copy for the landing page is obtained based on at least one of the second feature information and industry information of the landing page. This ensures that the obtained pocket copy matches the landing page, thereby avoiding the situation where matching copy cannot be added to the landing page when a valid copy is not obtained, and thus improving the copywriting effect of the landing page.

[0057] It should be noted that the landing page copy addition method disclosed herein is performed by an electronic device, that is, all steps included in the above method are performed by the electronic device, which can be a server, computer, mobile phone or other electronic device.

[0058] In one embodiment, Figure 1 In the illustrated embodiment, obtaining the landing page's copy based on at least one of the second feature information and industry information of the landing page includes at least one of the following:

[0059] Extract the second feature information of the landing page, and concatenate the second feature information to obtain the pocket copy of the landing page. The second feature information includes at least one of key fragment information and business information.

[0060] Obtain industry information of the landing page, and based on the industry information, obtain target copy within the industry to which the landing page belongs that meets preset conditions for click-through rate. Use the target copy as the pocket copy of the landing page, wherein the target copy does not contain business information.

[0061] The aforementioned key information segments can be keywords from the landing page. For example, multiple information segments can be identified from the landing page, each segment including words, and keywords can be identified from these words, such as "express delivery", "low price and high quality", "quality guaranteed", etc.

[0062] The industry information mentioned above can be industry phrases from the landing page, such as "XX Housekeeping Services" or "XX Home Renovation".

[0063] The above-mentioned splicing of the second feature information can be done by splicing key fragment information and business information. For example, if the business information is "XX Household Services" and the key fragment information is "Express On-Site Service", then the resulting pocket copy would be "XX Household Services, Express On-Site Service". Alternatively, the above-mentioned splicing of the second feature information can be done by splicing multiple key fragment information. For example, if the key fragment information includes "Express On-Site Service" and "Quality Guaranteed", then the resulting pocket copy would be "Express On-Site Service, Quality Guaranteed".

[0064] In this embodiment, since the pocket copy includes the second feature information, the matching degree between the pocket copy and the landing page can be improved, thereby further improving the copywriting effect of the landing page.

[0065] The target copy for the landing page's industry that meets preset click-through rates can be one or more of the highest-performing copy within that industry, or copy with a click-through rate reaching a preset threshold. Furthermore, the target copy can be user-written direct-to-conversion copy. This allows for the rapid acquisition of the landing page's core copy. Additionally, because the target copy does not contain business information, it possesses strong versatility, meeting the needs of different landing pages.

[0066] In some implementations, target copy for multiple industries can be obtained in advance. This allows for the rapid generation of copy for landing pages in different industries, thus avoiding the inability to generate copy for landing pages with direct conversion functionality due to cold start issues caused by data delays.

[0067] In one embodiment, the extraction of the second feature information of the landing page includes:

[0068] Extract multiple fragments of information from the landing page;

[0069] The multiple information segments are scored to obtain a score for each segment;

[0070] At least one segment of information is selected from the plurality of segment information as the key segment information, and the score of the at least one segment information is higher than the scores of the other segment information among the plurality of segment information.

[0071] The scoring of multiple information segments can be done according to pre-defined scoring rules, such as scoring each segment based on the rules of the segment information and the landing page theme, or scoring each segment based on a pre-acquired scoring model, such as scoring each segment based on an NLP text scoring model.

[0072] In this embodiment, the key segment information with the highest score can be selected. The key segment information with the highest score is often the advantageous segment information of the landing page. This can realize the generation of pocket copy based on the advantageous segment information of the landing page, so as to improve the copywriting effect of the pocket copy.

[0073] In one embodiment, scoring the plurality of fragment information to obtain a score for each fragment information includes:

[0074] Extract text feature information from target fragment information;

[0075] Calculate the overlap between the text feature information and the preset text feature information, and calculate the score of the target segment information based on the overlap.

[0076] The target segment information is any one of the multiple segment information.

[0077] The aforementioned preset text feature information can be a pre-defined target text, which is text with high conversion rate, high precision rate, or high recall rate, etc. The specific preset text feature information can be set according to actual needs.

[0078] In some implementations, the extraction of textual feature information and the calculation of scores can be achieved through an NLP text scoring model, which can improve the efficiency of scoring.

[0079] In this embodiment, since the target fragment information is any of the above fragment information, each fragment information can be scored based on the above rules, thereby selecting the key fragment information with the highest overlap with the above preset text feature information to improve the copywriting effect of the above pocket copy.

[0080] It should be noted that this disclosure is not limited to selecting the above key fragment information through the above scoring method. For example, in some embodiments, multiple fragment information is extracted as key fragment information that is the same as or related to the topic name of the landing page.

[0081] In one embodiment, Figure 1 In the illustrated embodiment, step S104 includes:

[0082] The candidate text is filtered, and the filtered text is added to the landing page. The filtering process includes at least one of the following:

[0083] Keyword filtering, brand filtering, and segment filtering.

[0084] The core word filtering includes: identifying core words in the candidate text, verifying the core words, deleting the first text if the candidate solution includes the first text, and ensuring that the landing page does not include the core words of the first text.

[0085] The aforementioned core words can be identified through a pre-acquired core word recognition model, for example, by using a customized Natural Language Processing (NLP Customization, NLPC) core word recognition model to identify core words in candidate texts.

[0086] The landing page does not include the core words of the first copy, which may be that the optical character recognition (OCR) text of the landing page does not include the core words of the first copy.

[0087] By using the core keyword filtering described above, candidate copy that does not belong to the landing page can be filtered out, thereby improving the relevance of the final added copy to the landing page.

[0088] The brand filtering includes: identifying brand information in the candidate copy, verifying the brand information, deleting the second copy if the candidate solution includes the second copy, and determining that the landing page does not match the brand information in the second copy.

[0089] The aforementioned brand information can be identified in candidate copy through a pre-acquired brand recognition model.

[0090] By using the brand filtering described above, candidate copy that does not belong to the landing page can be filtered out, thereby improving the matching degree between the final added copy and the landing page.

[0091] The aforementioned segment filtering includes: identifying segment content of a preset type in the candidate text, verifying the segment content, and deleting the third text if the candidate solution includes the third text, so that the landing page does not include the segment content of the third text.

[0092] The aforementioned preset content segments can be marketing content segments, i.e., marketing segments from the candidate copy. For example, a candidate copy might include marketing phrases such as "Enjoy x% off," "Down payment x yuan," or "Free." This verification process can filter out marketing phrases that don't appear on the landing page, thus improving the match between the final added copy and the landing page.

[0093] In some implementations, filtering the candidate text may involve filtering the text in the pocket, because in some implementations, the candidate text may include the text in the pocket, as well as the candidate text that has been filtered through the text generation operation.

[0094] In one embodiment, the text added to the landing page is target text, which includes multiple pieces of information, and the method further includes:

[0095] The multiple pieces of information are respectively input into the information recognition model to predict key point information, thereby obtaining the key point information of the target text. The information recognition model is a network model used to identify the key point information of the text.

[0096] The key information is highlighted on the landing page.

[0097] The aforementioned information recognition model is pre-acquired and used to identify key information in the target copy that has the highest accuracy, precision, or recall rate. This key information can also be referred to as the selling points of the target copy.

[0098] Highlighting key information on the landing page can be achieved by highlighting these key points in red to emphasize them. For example, the target text could be: "Positive and negative pressure forming and cutting integrated machine, quality assurance." Customization supported ",in" ”" The segments between “” (customizable) are highlighted in red.

[0099] In this embodiment, by highlighting the key information on the landing page, the display effect of the copywriting information can be improved, allowing users to quickly view the key information when browsing the landing page. For example, for conversion-direct copy, the conversion-direct copy can be further highlighted to improve advertising conversion efficiency.

[0100] In one embodiment, Figure 1 Step S102 in the illustrated embodiment includes:

[0101] Using a text generation model, a copywriting generation operation is performed on the landing page based on the first feature information to obtain the operation result;

[0102] The text generation model is a network model used to generate the text of the landing page based on the first feature information of the landing page.

[0103] The first feature information mentioned above is the feature information extracted by using the landing page as input to the text generation model.

[0104] The text generation model described above can be a Natural Language Processing (NLP) model, such as the ERNIE-GEN model, and there is no limitation thereto. In some implementations, the text generation model described above can also be an attention model. In some implementations, the text generation model described above can also be a Masked Sequence to Sequence Pre-training (MASS) model.

[0105] In some embodiments, the text generation model described above can generate approximately 10 pieces of text for the landing page. However, if the landing page lacks sufficient valid content, or the model's generated results fail to meet preset conditions, the above operation result indicates that no valid text for the landing page has been obtained.

[0106] In this embodiment, the efficiency of generating candidate text can be improved by performing the text generation operation through a text generation model.

[0107] In one embodiment, the text generation model is a text generation model trained based on source text and target text;

[0108] The source text includes at least one of the following: source text obtained by prepending business information related to the landing page of the first sample to the first source training sample, and source text obtained by prepending geographic information related to the landing page of the second sample to the second source training sample; wherein the first source training sample and the second source training sample are pre-set training samples; and / or,

[0109] The target text includes at least one of the following: a creative title, or target text obtained by filling the creative title with keywords from the multimedia creative that meet preset conditions, wherein the multimedia creative is a multimedia creative found through a search engine, and the creative title is the title of the multimedia creative.

[0110] The source and target texts mentioned above are the source and target texts used by the text generation model during training.

[0111] The landing page of the first sample and the landing page of the second sample can be the same or different landing pages, and there can be one or more landing pages.

[0112] The first and second source training samples mentioned above can be training texts from a corpus.

[0113] The aforementioned method of prepending the business information related to the landing page of the first sample to the first source training sample can be achieved by adding the business information related to the landing page of the first sample to each training sample in order to define a domain. For example, the 10 business phrases most relevant to the landing page (such as "multifunctional positive and negative pressure machine manufacturer") can be prepended to the source text of each of the 10 training samples. This prepending adds business information to the very beginning of the source training samples.

[0114] The source text obtained by placing the geographic information related to the second sample landing page before the second source training sample can be the geographic information (such as "Shanghai") from the second sample landing page in the second source training sample.

[0115] In some implementations, source text can be obtained by combining prior business information and geographic information from the same source training sample.

[0116] In this embodiment, by using prior business information, the relevance between the source text and the landing page can be strengthened and enriched, thereby improving the prediction accuracy of the text generation model. Furthermore, by using prior geographic information, the connections between regions can be explicitly indicated, preventing the text generation model from incorrectly learning geographic relationships, thus further improving the prediction accuracy of the text generation model.

[0117] The target text mentioned above can be the creative title from search engine search results (such as "How to choose {XXXX}, factory direct sales, lower price than competitors"), which can replace the conversion copy written by the user and serve as the target text for fine-tuning the text generation model.

[0118] Alternatively, the target text could be filled into the title of the multimedia creative (e.g., advertising creative) with the keyword that has the highest conversion rate (such as "multifunctional positive and negative pressure machine") to obtain a more complete target text.

[0119] In this embodiment, target text is generated by using keywords in creative titles and multimedia creatives. Since the searched creative titles are relatively high-quality and similar to the conversion direct copy in terms of copywriting style, copywriting, and relevance to the landing page, the amount of data for fine-tuning can be increased through the aforementioned target text.

[0120] In addition, the creative titles mentioned above can also be creative titles that have been displayed for a period of time after being searched (such as "How to choose a multi-functional positive and negative pressure machine? Factory direct sales, lower price than competitors"), which to a certain extent ensures the quality and conversion efficiency of the target text, thereby improving the prediction accuracy of the text generation model.

[0121] In some embodiments, the added text can be direct conversion text, which can improve the conversion rate of the landing page.

[0122] In this disclosure, when a valid copy for the landing page is not obtained, a pocket copy for the landing page is obtained based on at least one of the second feature information and industry information of the landing page. This ensures that the obtained pocket copy matches the landing page, thereby avoiding the situation where matching copy cannot be added to the landing page when a valid copy is not obtained, and thus improving the copywriting effect of the landing page.

[0123] Please see Figure 2 , Figure 2 This is a flowchart of a model training method provided in this disclosure, such as... Figure 2 As shown, it includes the following steps:

[0124] Step S201: Obtain source text and target text, wherein the source text includes at least one of the following: source text obtained by prepending the business information related to the landing page of the first sample to the first source training sample, and source text obtained by prepending the regional information related to the landing page of the second sample to the second source training sample; wherein the first source training sample and the second source training sample are pre-set training samples; and / or, the target text includes: target text obtained by filling the title of the multimedia creative with keywords that meet preset conditions, wherein the multimedia creative is a multimedia creative found through a search engine, and the title of the creative is the title of the multimedia creative;

[0125] Step S202: Train a preset network model based on the source text and the target text to obtain a text generation model; wherein, the text generation model is used to generate the text of the landing page based on the feature information of the landing page.

[0126] The source text, target text, and text generation model mentioned above can be found in the corresponding descriptions of the above-described landing page copywriting addition method embodiments, and will not be repeated here.

[0127] In this embodiment, since a text generation model is obtained by training a preset network model based on the source text and the target text, the prediction accuracy of the text generation model is improved.

[0128] It should be noted that the model training method disclosed herein is executed by an electronic device, that is, all steps included in the above method are executed by the electronic device, which may be a server, computer, mobile phone or other electronic device.

[0129] It should be noted that the various embodiments provided in this disclosure can be implemented individually or in combination with each other, for example: Figure 3As shown, one technical solution includes the following:

[0130] 301. Text generation model, wherein the text generation model can be trained based on the business information, region and OCR data of the landing page, as well as the search ad creative title;

[0131] 302. Pocket copy generation: Pocket copy can be generated by piecing together advantageous segments of the landing page, and can also include general copy with high click-through rates in the industry;

[0132] 303. Post-processing includes filtering and highlighting key information. Filtering includes core keyword filtering, brand filtering, and segment filtering.

[0133] Please see Figure 4a , Figure 4a This is a landing page copy addition device provided in this public document, such as... Figure 4a As shown, the landing page copy adding device 400 includes:

[0134] Extraction module 401 is used to extract the first feature information of the landing page;

[0135] The generation module 402 is used to perform a copy generation operation on the landing page based on the first feature information to obtain the operation result;

[0136] The acquisition module 403 is used to acquire the landing page's "pocket copy" based on at least one of the second feature information of the landing page and industry information when the operation result indicates that no valid copy of the landing page has been acquired.

[0137] Add module 404 is used to add text to the landing page based on candidate text, wherein, if the operation result indicates that no valid text is obtained for the landing page, the candidate text includes the text in the pocket.

[0138] In one embodiment, such as Figure 4b As shown, the acquisition module 403 includes at least one of the following:

[0139] The splicing unit 4031 is used to extract the second feature information of the landing page, splice the second feature information to obtain the pocket copy of the landing page, wherein the second feature information includes at least one of key fragment information and business information;

[0140] The acquisition unit 4032 is used to acquire industry information of the landing page, acquire target copy with a click-through rate that meets preset conditions within the industry to which the landing page belongs based on the industry information, and use the target copy as the pocket copy of the landing page, wherein the target copy does not contain business information.

[0141] In one embodiment, the splicing unit 4031 is used for:

[0142] Extract multiple fragments of information from the landing page;

[0143] The multiple information segments are scored to obtain a score for each segment;

[0144] At least one segment of information is selected from the plurality of segment information as the key segment information, and the score of the at least one segment information is higher than the scores of the other segment information among the plurality of segment information.

[0145] In one embodiment, the splicing unit 4031 is used for:

[0146] Extract text feature information from target fragment information;

[0147] Calculate the overlap between the text feature information and the preset text feature information, and calculate the score of the target segment information based on the overlap.

[0148] The target segment information is any one of the multiple segment information.

[0149] In one embodiment, the adding module 404 is used for:

[0150] The candidate text is filtered, and the filtered text is added to the landing page. The filtering process includes at least one of the following:

[0151] Keyword filtering, brand filtering, and segment filtering.

[0152] In one embodiment, the core word filtering includes: identifying core words in the candidate copy, verifying the core words, and, if the candidate solution includes the first copy, deleting the first copy, wherein the landing page does not include the core words of the first copy; and / or,

[0153] The brand filtering includes: identifying brand information in the candidate copy, verifying the brand information, and deleting the second copy if the candidate solution includes the second copy, wherein the landing page does not match the brand information of the second copy; and / or,

[0154] The segment filtering includes: identifying segment content of a preset type in the candidate copy, verifying the segment content, and deleting the third copy if the candidate solution includes the third copy, so that the landing page does not include the segment content of the third copy.

[0155] In one embodiment, the text added to the landing page is the target text, which includes multiple pieces of information, such as... Figure 4c As shown, the device further includes:

[0156] Prediction module 405 is used to input the multiple pieces of information into the information recognition model to predict key point information, thereby obtaining the key point information of the target text. The information recognition model is a network model used to identify the key point information of the text.

[0157] The display module 406 is used to highlight the key information on the landing page.

[0158] In one embodiment, the generation module 402 is used for:

[0159] Using a text generation model, a copywriting generation operation is performed on the landing page based on the first feature information to obtain the operation result;

[0160] The text generation model is a network model used to generate the text of the landing page based on the first feature information of the landing page.

[0161] In one embodiment, the text generation model is a text generation model trained based on source text and target text;

[0162] The source text includes at least one of the following: source text obtained by prepending business information related to the landing page of the first sample to the first source training sample, and source text obtained by prepending geographic information related to the landing page of the second sample to the second source training sample; wherein the first source training sample and the second source training sample are pre-set training samples; and / or,

[0163] The target text includes at least one of the following: a creative title, or target text obtained by filling the creative title with keywords from the multimedia creative that meet preset conditions, wherein the multimedia creative is a multimedia creative found through a search engine, and the creative title is the title of the multimedia creative.

[0164] The landing page text addition device provided in this disclosure can realize all the processes of the landing page text addition method provided in this disclosure and achieve the same technical effect. To avoid duplication, it will not be described in detail here.

[0165] Please see Figure 5 , Figure 5 This is a model training device provided in this disclosure, such as Figure 5 As shown, the model training device 500 includes:

[0166] The acquisition module 501 is used to acquire source text and target text, wherein the source text includes at least one of the following: source text obtained by prepending business information related to the landing page of the first sample to the first source training sample, and source text obtained by prepending regional information related to the landing page of the second sample to the second source training sample; wherein the first source training sample and the second source training sample are pre-set training samples; and / or, the target text includes: target text obtained by filling the title of the multimedia creative with keywords that meet preset conditions, wherein the multimedia creative is a multimedia creative found through a search engine, and the title of the creative is the title of the multimedia creative;

[0167] Training module 502 is used to train a preset network model based on the source text and the target text to obtain a text generation model; wherein, the text generation model is used to generate the text of the landing page based on the feature information of the landing page.

[0168] The model training apparatus provided in this disclosure can implement all the processes of the model training method provided in this disclosure and achieve the same technical effect. To avoid repetition, it will not be described in detail here.

[0169] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0170] The aforementioned electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute the landing page copywriting addition method or model training method provided in this disclosure.

[0171] The aforementioned readable storage medium stores computer instructions, wherein the computer instructions are used to cause the computer to execute the landing page copy addition method or the model training method provided in this disclosure.

[0172] The aforementioned computer program product includes a computer program that, when executed by a processor, implements the landing page copy addition method or model training method provided in this disclosure.

[0173] Figure 6A schematic block diagram of an example electronic device 600 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.

[0174] like Figure 6 As shown, device 600 includes a computing unit 601, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 602 or a computer program loaded from storage unit 608 into random access memory (RAM) 603. RAM 603 may also store various programs and data required for the operation of device 600. The computing unit 601, ROM 602, and RAM 603 are interconnected via bus 604. Input / output (I / O) interface 605 is also connected to bus 604.

[0175] Multiple components in device 600 are connected to I / O interface 605, including: input unit 606, such as keyboard, mouse, etc.; output unit 607, such as various types of monitors, speakers, etc.; storage unit 608, such as disk, optical disk, etc.; and communication unit 609, such as network card, modem, wireless transceiver, etc. Communication unit 609 allows device 600 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0176] The computing unit 601 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 601 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 601 performs the various methods and processes described above, such as the landing page copywriting method or the model training method. For example, in some embodiments, the landing page copywriting method or the model training method can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program can be loaded and / or installed on device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by the computing unit 601, one or more steps of the landing page copywriting method or the model training method described above can be performed. Alternatively, in other embodiments, the computing unit 601 may be configured in any other suitable manner (e.g., by means of firmware) to perform a landing page copy addition method or a model training method.

[0177] 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.

[0178] 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.

[0179] 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.

[0180] 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).

[0181] 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.

[0182] 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.

[0183] 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.

[0184] 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 adding landing page copy, including: Extract the primary feature information of the landing page; Based on the first feature information, a copywriting generation operation is performed on the landing page to obtain the operation result; If the operation result indicates that no valid copy of the landing page has been obtained, the pocket copy of the landing page is obtained based on the second feature information and industry information of the landing page. Based on the candidate copy, add copy to the landing page, wherein, if the operation result indicates that no valid copy for the landing page is obtained, the candidate copy includes the copy in the pocket; The step of obtaining the landing page's copy based on the second feature information and industry information includes: Extract the second feature information of the landing page, and concatenate the second feature information to obtain the pocket copy of the landing page. The second feature information includes at least one of key fragment information and business information. Obtain industry information of the landing page, obtain target copy with a click-through rate that meets preset conditions within the industry to which the landing page belongs based on the industry information, and use the target copy as the pocket copy of the landing page, wherein the target copy does not contain business information; The text added to the landing page is the target text, which includes multiple pieces of information. The method further includes: The multiple pieces of information are respectively input into the information recognition model to predict key point information, thereby obtaining the key point information of the target text. The information recognition model is a network model used to identify the key point information of the text. The key information is highlighted on the landing page.

2. The method according to claim 1, wherein, The extraction of the second feature information of the landing page includes: Extract multiple fragments of information from the landing page; The multiple information segments are scored to obtain a score for each segment; At least one segment of information is selected from the plurality of segment information as the key segment information, and the score of the at least one segment information is higher than the scores of the other segment information among the plurality of segment information.

3. The method according to claim 2, wherein, The scoring of the multiple information segments to obtain a score for each segment includes: Extract text feature information from target fragment information; Calculate the overlap between the text feature information and the preset text feature information, and calculate the score of the target segment information based on the overlap. The target segment information is any one of the multiple segment information.

4. The method according to any one of claims 1 to 3, wherein, The step of adding text to the landing page based on candidate text includes: The candidate text is filtered, and the filtered text is added to the landing page. The filtering process includes at least one of the following: Keyword filtering, brand filtering, and segment filtering.

5. The method according to claim 4, characterized in that, The core keyword filtering includes: identifying core keywords in the candidate texts, verifying the core keywords, and deleting the first text if it is included in the candidate texts, and the landing page does not include the core keywords of the first text; and / or, The brand filtering includes: identifying brand information in the candidate copy, verifying the brand information, and deleting the second copy if it is included in the candidate copy, wherein the landing page does not match the brand information in the second copy; and / or, The segment filtering includes: identifying segment content of a preset type in the candidate text, verifying the segment content, and deleting the third text if the candidate text includes a third text, so that the landing page does not include the segment content of the third text.

6. The method according to any one of claims 1 to 3, wherein, The step of performing copy generation on the landing page based on the first feature information to obtain the operation result includes: Using a text generation model, a copywriting generation operation is performed on the landing page based on the first feature information to obtain the operation result; The text generation model is a network model used to generate the text of the landing page based on the first feature information of the landing page.

7. The method according to claim 6, wherein, The text generation model is a text generation model trained based on source text and target text; The source text includes at least one of the following: source text obtained by prepending business information related to the landing page of the first sample to the first source training sample, and source text obtained by prepending geographic information related to the landing page of the second sample to the second source training sample; wherein the first source training sample and the second source training sample are pre-set training samples; and / or, The target text includes at least one of the following: a creative title, or target text obtained by filling the creative title with keywords from the multimedia creative that meet preset conditions, wherein the multimedia creative is a multimedia creative found through a search engine, and the creative title is the title of the multimedia creative.

8. A landing page copy adding device, comprising: The extraction module is used to extract the first feature information of the landing page; The generation module is used to perform a copy generation operation on the landing page based on the first feature information and obtain the operation result; The acquisition module is used to acquire the "pocket copy" of the landing page based on the second feature information and industry information of the landing page when the operation result indicates that no valid copy of the landing page has been acquired. An add module is used to add text to the landing page based on candidate text, wherein, if the operation result indicates that no valid text is obtained for the landing page, the candidate text includes the text in the pocket; The acquisition module includes: The splicing unit is used to extract the second feature information of the landing page, splice the second feature information to obtain the pocket copy of the landing page, wherein the second feature information includes at least one of key fragment information and business information; The acquisition unit is used to acquire industry information of the landing page, acquire target copy with a click-through rate that meets preset conditions within the industry to which the landing page belongs based on the industry information, and use the target copy as the pocket copy of the landing page, wherein the target copy does not contain business information; The text added to the landing page is the target text, which includes multiple pieces of information. The device also includes: The prediction module is used to input the multiple pieces of information into the information recognition model to predict key point information and obtain the key point information of the target text. The information recognition model is a network model used to recognize the key point information of the text. The display module is used to highlight the key information on the landing page.

9. The apparatus according to claim 8, wherein, The splicing unit is used for: Extract multiple fragments of information from the landing page; The multiple information segments are scored to obtain a score for each segment; At least one segment of information is selected from the plurality of segment information as the key segment information, and the score of the at least one segment information is higher than the scores of the other segment information among the plurality of segment information.

10. The apparatus according to claim 9, wherein, The splicing unit is used for: Extract text feature information from target fragment information; Calculate the overlap between the text feature information and the preset text feature information, and calculate the score of the target segment information based on the overlap. The target segment information is any one of the multiple segment information.

11. The apparatus according to any one of claims 8 to 10, wherein, The added module is used for: The candidate text is filtered, and the filtered text is added to the landing page. The filtering process includes at least one of the following: Keyword filtering, brand filtering, and segment filtering.

12. The apparatus according to claim 11, characterized in that, The core keyword filtering includes: identifying core keywords in the candidate texts, verifying the core keywords, and deleting the first text if it is included in the candidate texts, and the landing page does not include the core keywords of the first text; and / or, The brand filtering includes: identifying brand information in the candidate copy, verifying the brand information, and deleting the second copy if it is included in the candidate copy, wherein the landing page does not match the brand information in the second copy; and / or, The segment filtering includes: identifying segment content of a preset type in the candidate text, verifying the segment content, and deleting the third text if the candidate text includes a third text, so that the landing page does not include the segment content of the third text.

13. The apparatus according to any one of claims 8 to 10, wherein, The generation module is used for: Using a text generation model, a copywriting generation operation is performed on the landing page based on the first feature information to obtain the operation result; The text generation model is a network model used to generate the text of the landing page based on the first feature information of the landing page.

14. The apparatus according to claim 13, wherein, The text generation model is a text generation model trained based on source text and target text; The source text includes at least one of the following: source text obtained by prepending business information related to the landing page of the first sample to the first source training sample, and source text obtained by prepending geographic information related to the landing page of the second sample to the second source training sample; wherein the first source training sample and the second source training sample are pre-set training samples; and / or, The target text includes at least one of the following: a creative title, or target text obtained by filling the creative title with keywords from the multimedia creative that meet preset conditions, wherein the multimedia creative is a multimedia creative found through a search engine, and the creative title is the title of the multimedia creative.

15. 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 method of any one of claims 1-7.

16. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-7.

17. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-7.

Images