Ad copy template generation method and device, equipment, medium and product

Abstract

The application relates to an advertisement template generation method and device, equipment, medium and product. The method comprises the following steps: obtaining an advertisement text of a commodity; decomposing the advertisement text into a label sequence according to a preset label structure, wherein the label structure comprises a plurality of label groups, and each label group comprises at least one word label; extracting material text corresponding to each label group from the advertisement text based on the word label in the label sequence; and constructing an advertisement template according to the label sequence, wherein part of the label groups are represented as corresponding replaceable group labels in the advertisement template, and other label groups are kept as their corresponding material texts. The advertisement template generated from the advertisement text under the guidance of the hierarchy and granularity of the label structure is more accurate and effective, can be used for batch production of advertisement texts, improves the generation efficiency of the required advertisement texts of a merchant, and improves the online advertisement technical service capability of an e-commerce platform.

Description

Technical Field

[0001] This application relates to the field of e-commerce information technology, and in particular to a method for generating copywriting templates and the corresponding apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Technology

[0002] In e-commerce, product advertising copy is a crucial component of advertising materials. Merchants need to fill in text information such as titles and body text when creating ads, which requires manpower. Excellent advertising copy itself has a certain writing threshold and skill requirement, especially for non-professionals writing copy in unfamiliar languages. The quality of advertising copy directly impacts the effectiveness of advertising campaigns; therefore, providing advertising copy generation services to merchants is a fundamental requirement for e-commerce platforms.

[0003] Traditional automatic generation technology for product advertising copy often uses template matching. However, the most common way to generate the corresponding templates is to manually create multiple copy templates in advance and publish them to a template library for merchants to use. Obviously, this method is quite inefficient.

[0004] There are also solutions for automatically generating templates in the industry, but due to the different algorithms and processing logic used by each company, the results vary. In actual testing, the existing solutions have not achieved good results, mainly because the expression of the final generated copy is not good, and further exploration is needed. Summary of the Invention

[0005] The purpose of this application is to solve the above-mentioned problems by providing a method for generating copywriting templates, as well as corresponding apparatus, computer equipment, computer-readable storage media, and computer program products.

[0006] To suit the various purposes of this application, the following technical solution is adopted:

[0007] In one respect, a method for generating a copy template is provided to suit one of the purposes of this application, comprising the following steps:

[0008] Obtain the product's advertising text;

[0009] The advertising text is deconstructed into a tag sequence according to a preset tag structure. The tag structure contains multiple tag groups, and each tag group contains at least one word tag.

[0010] Based on the word tags in the tag sequence, extract the material text corresponding to each tag group from the advertisement text;

[0011] A copywriting template is constructed based on the tag sequence. In the copywriting template, some tag groups are represented as corresponding replaceable group tags, while other tag groups are retained and represented as their corresponding material text.

[0012] Based on any of the above embodiments, the advertising text is deconstructed into a tag sequence according to a preset tag structure, including the following steps:

[0013] Obtain the semantic vector of each word segment in the advertisement text;

[0014] Obtain the part-of-speech vectors of each word segment in the advertisement text;

[0015] The semantic vector and part-of-speech vector of each word segment are used to construct a comprehensive vector;

[0016] Based on the comprehensive vector of each word segment, the corresponding word tag of each word segment in the preset tag structure is determined, and the tag sequence corresponding to the advertising text is obtained.

[0017] Based on any of the above embodiments, obtaining the part-of-speech vectors of each word segment in the advertising text includes the following steps:

[0018] The advertising text is input into a part-of-speech extractor to obtain the part-of-speech tags for each word segment.

[0019] The part-of-speech (POS) vectors corresponding to each word segment are determined by querying a preset part-of-speech encoding table. The POS encoding table is arranged in an ordered manner according to the statistical frequency of different parts of speech in the reference advertising text set. The one-hot encoding algorithm is applied to determine the POS vectors corresponding to each part of speech.

[0020] Based on any of the above embodiments, the tag structure includes tag groups for indicating the following types of information: advertising phrases, core words, reserved words, and punctuation marks, wherein,

[0021] The tag groups for the advertising sentence structure include tag groups corresponding to the opening sentence structure, discount information sentence structure, and closing sentence structure, wherein each tag group includes word tags used to indicate whether the corresponding word belongs to the starting position or the non-starting position.

[0022] The tag groups for the core words include tag groups corresponding to product words and brand words, wherein each tag group includes word tags used to indicate whether the corresponding word belongs to the starting position or a non-starting position;

[0023] The tag groups corresponding to the reserved words and punctuation marks include a single corresponding word tag.

[0024] Based on any of the above embodiments, constructing a copy template according to the tag sequence includes the following steps:

[0025] The word segments corresponding to the word tags in the tag groups that belong to the retention type in the advertisement text are retained;

[0026] The word segmentation set corresponding to each consecutively occurring word tag in the tag group belonging to the replaceable type in the advertisement text is replaced and represented as its corresponding group tag.

[0027] Based on any of the above embodiments, after the step of extracting the material text corresponding to each tag group from the advertising text based on the word tags in the tag sequence, the following steps are included:

[0028] The mapping relationship data between the group tags of the tag group and their material text is stored in the material library corresponding to the category of the product described in the advertising text. The group tags include group tags for indicating that the material text corresponding to the tag group is an advertising sentence and group tags for indicating that the material text corresponding to the tag group is a core word. The core word is the product word and / or brand word of the product.

[0029] Based on any of the above embodiments, after the step of constructing the copy template according to the tag sequence, the following steps are included:

[0030] In response to an advertising copy generation request submitted by a terminal device, the product core keywords specified in the request are obtained. The product core keywords include the product keywords and / or brand keywords of the target product corresponding to the request.

[0031] Replace the group tags in the copywriting template that indicate the core words with the product core words accordingly;

[0032] Replace the group tags in the copy template that indicate the advertising phrases with any material text in the material library that corresponds to the group tag in the category of the target product;

[0033] The advertising copy obtained after the copy template is replaced is pushed to the terminal device.

[0034] On the other hand, to suit one of the purposes of this application, a copywriting template generation device is provided, including a text acquisition module, a tag deconstruction module, a material extraction module, and a template construction module. The text acquisition module is used to acquire advertising text for a product; the tag deconstruction module is used to deconstruct the advertising text into a tag sequence according to a preset tag structure, the tag structure containing multiple tag groups, each tag group containing at least one word tag; the material extraction module is used to extract material text corresponding to each tag group from the advertising text based on the word tags in the tag sequence; and the template construction module is used to construct a copywriting template according to the tag sequence, wherein some tag groups in the copywriting template are represented as corresponding replaceable group tags, and other tag groups are retained and represented as their corresponding material text.

[0035] Based on any of the above embodiments, the tag deconstruction module includes: a semantic processing submodule, used to obtain the semantic vector of each word segment in the advertising text; a part-of-speech processing submodule, used to obtain the part-of-speech vector of each word segment in the advertising text; a vector synthesis submodule, used to construct a synthesis vector from the semantic vector and part-of-speech vector of each word segment; and a part-of-speech determination submodule, used to determine the word tag corresponding to each word segment in the preset tag structure based on the synthesis vector of each word segment, thereby obtaining the tag sequence corresponding to the advertising text.

[0036] Based on any of the above embodiments, the part-of-speech processing submodule includes: a part-of-speech extraction unit, used to input the advertising text into a part-of-speech extractor to obtain the part-of-speech identifier of each word segment; and a vector embedding unit, used to query a preset part-of-speech encoding table to determine the part-of-speech vector corresponding to each word segment. The part-of-speech encoding table is arranged in an ordered manner according to the statistical frequency of different parts of speech in the reference advertising text set, and the one-hot encoding algorithm is applied to determine the part-of-speech vector corresponding to each part of speech.

[0037] Based on any of the above embodiments, the tag structure includes tag groups for indicating the following types of information: advertising phrases, core words, reserved words, and punctuation marks, wherein,

[0038] The tag groups for the advertising sentence structure include tag groups corresponding to the opening sentence structure, discount information sentence structure, and closing sentence structure, wherein each tag group includes word tags used to indicate whether the corresponding word belongs to the starting position or the non-starting position.

[0039] The tag groups for the core words include tag groups corresponding to product words and brand words, wherein each tag group includes word tags used to indicate whether the corresponding word belongs to the starting position or a non-starting position;

[0040] The tag groups corresponding to the reserved words and punctuation marks include a single corresponding word tag.

[0041] Based on any of the above embodiments, the template construction module includes: a retention processing submodule, used to retain the word segments corresponding to the word tags in the tag groups belonging to the retention type in the advertising text; and a replacement processing submodule, used to replace the word segments corresponding to each consecutively occurring word tag in the tag groups belonging to the replaceable type in the advertising text with their corresponding group tags.

[0042] Based on any of the above embodiments, the material extraction module further includes: a material archiving module, used to store the mapping relationship data between the group tags of the tag group and their material text in a material library corresponding to the category of the product described in the advertising text. The group tags include group tags used to indicate that the material text corresponding to the tag group is an advertising sentence and group tags used to indicate that the material text corresponding to the tag group is a core word. The core word is the product word and / or brand word of the product.

[0043] Based on any of the above embodiments, the template construction module further includes: a request response module, used to respond to an advertising copy generation request submitted by a terminal device, and obtain the product core keywords specified in the request, wherein the product core keywords include product keywords and / or brand keywords of the target product corresponding to the request; a core keyword replacement module, used to replace the group tags in the copy template that indicate the core keywords with the product core keywords; an advertisement replacement module, used to replace the group tags in the copy template that indicate the advertisement sentence structure with any material text in the material library corresponding to the group tags in the material library corresponding to the category of the target product; and a copy push module, used to push the advertising copy obtained after the copy template has been replaced to the terminal device.

[0044] In another aspect, a computer device provided for one of the purposes of this application includes a central processing unit and a memory, the central processing unit being used to invoke and run a computer program stored in the memory to perform the steps of the document template generation method described in this application.

[0045] In another aspect, a computer-readable storage medium is provided to suit another purpose of this application, which stores, in the form of computer-readable instructions, a computer program implemented according to the described document template generation method, which, when invoked by a computer, performs the steps included in the method.

[0046] In another aspect, a computer program product provided for another purpose of this application includes a computer program / instructions that, when executed by a processor, implement the steps of the method described in any embodiment of this application.

[0047] Compared to existing technologies, this application has several advantages, including at least the following: In constructing a text template based on given advertising text, this application utilizes a preset tag structure. By reflecting the hierarchical and granular relationships between tag groups and their corresponding word tags within that tag structure, the application accurately extracts the material text corresponding to each tag group from the advertising text. Then, in constructing the corresponding text template, some tag groups are treated as replaceable tag groups, represented as group tags, acting as masks that can be used to replace other variable content. For the remaining tag groups, the original content of the advertising text is preserved by representing them as their corresponding material text. This allows for the optimal selection of the text content from the advertising text. The resulting copy template not only includes the compelling expressions from the advertising text but also allows for easy replacement of the group tags of replaceable tag groups with information corresponding to the products for which advertising copy needs to be generated. Guided by the hierarchy and granularity of the tag structure, the copy template is more accurate and effective, enabling mass production of advertising copy, improving the efficiency of generating advertising copy for merchants, and enhancing the online advertising technology service capabilities of e-commerce platforms. Attached Figure Description

[0048] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0049] Figure 1 This is a flowchart illustrating a typical embodiment of the document template generation method of this application.

[0050] Figure 2 This is a schematic diagram of an exemplary label structure for this application.

[0051] Figure 3 This is a schematic diagram of the network structure of an exemplary deep learning model in this application.

[0052] Figure 4 This is a flowchart illustrating the process of deconstructing advertising text into a sequence of tags in an embodiment of this application.

[0053] Figure 5 This is a flowchart illustrating the process of obtaining part-of-speech vectors in an embodiment of this application.

[0054] Figure 6 This is a flowchart illustrating the process of converting advertising text into a text template, as described in an embodiment of this application.

[0055] Figure 7 This is a flowchart illustrating the process of generating an advertising scheme based on a text template in an embodiment of this application.

[0056] Figure 8This is a schematic diagram of the document template generation device for this application;

[0057] Figure 9 This is a schematic diagram of the structure of a computer device used in this application. Detailed Implementation

[0058] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0059] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this application means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.

[0060] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0061] Those skilled in the art will understand that the terms "client," "terminal," and "terminal device" as used herein include both devices that receive wireless signals, devices that only possess wireless signal receiver capabilities without transmission capabilities, and devices with receiving and transmitting hardware, devices that have receiving and transmitting hardware capable of bidirectional communication over a bidirectional communication link. Such devices may include: cellular or other communication devices such as personal computers or tablets, having single-line displays, multi-line displays, or cellular or other communication devices without multi-line displays; PCS (Personal Communications Service) that can combine voice, data processing, fax, and / or data communication capabilities; PDA (Personal Digital Assistant) that may include a radio frequency receiver, pager, internet / intranet access, web browser, notepad, calendar, and / or GPS (Global Positioning System) receiver; and conventional laptops and / or handheld computers or other devices that have and / or include radio frequency receivers. As used herein, "client," "terminal," and "terminal device" can be portable, transportable, installed in a means of transportation (air, sea, and / or land), or suitable and / or configured to operate locally and / or in a distributed manner, operating in any other location on Earth and / or in space. "Client," "terminal," and "terminal device" as used herein can also be a communication terminal, an internet access terminal, or a music / video playback terminal, such as a PDA, a MID (Mobile Internet Device), and / or a mobile phone with music / video playback capabilities, or a smart TV, set-top box, etc.

[0062] The hardware referred to by the names "server," "client," and "service node" in this application is essentially an electronic device with the equivalent capabilities of a personal computer. It is a hardware device with the necessary components revealed by the von Neumann architecture, such as a central processing unit (including an arithmetic logic unit and a control unit), memory, input devices, and output devices. The computer program is stored in its memory, and the central processing unit loads the program stored in the secondary storage into the main memory to run it, execute the instructions in the program, and interact with the input and output devices to complete specific functions.

[0063] It should be noted that the concept of "server" used in this application can also be extended to the case of server clusters. Based on the network deployment principles understood by those skilled in the art, the servers should be logically divided. Physically, these servers can be independent of each other but accessible through interfaces, or they can be integrated into a single physical computer or a computer cluster. Those skilled in the art should understand this flexibility and should not use it to constrain the implementation of the network deployment method in this application.

[0064] One or more of the technical features of this application, unless explicitly specified herein, can be deployed on a server and accessed by a client remotely calling the online service interface provided by the server, or can be directly deployed and run on a client for access.

[0065] Unless otherwise specified, the neural network models referenced or potentially referenced in this application may be deployed on a remote server and invoked remotely on the client, or deployed on a client with the capability to invoke directly. In some embodiments, when running on the client, the corresponding intelligence may be acquired through transfer learning in order to reduce the requirements on the client's hardware resources and avoid excessive consumption of the client's hardware resources.

[0066] Unless otherwise specified, all data involved in this application may be stored remotely on a server or on a local terminal device, as long as it is suitable for use by the technical solution of this application.

[0067] Those skilled in the art will understand that although the various methods in this application are described based on the same concept and thus present commonality among them, they can be performed independently unless otherwise specified. Similarly, the various embodiments disclosed in this application are all based on the same inventive concept; therefore, concepts expressed in the same way, as well as concepts that are appropriately changed for convenience but are expressed differently, should be understood equivalently.

[0068] Unless otherwise expressly stated, the various embodiments disclosed in this application can be combined in a cross-cutting manner to flexibly construct new embodiments, as long as such combination does not depart from the inventive spirit of this application and can meet the needs of the prior art or solve a certain deficiency in the prior art. Those skilled in the art should be aware of such modifications.

[0069] The copy template generation method of this application can be programmed into a computer program product and deployed on a client or server to run. For example, in an exemplary application scenario of this application, it can be deployed on the server of an e-commerce platform. In this way, the method can be executed by human-computer interaction with the process of the computer program product through a graphical user interface by accessing the interface opened after the computer program product is run.

[0070] Please see Figure 1 The text template generation method of this application, in its typical embodiment, includes the following steps:

[0071] Step S1100: Obtain the product's advertising text;

[0072] An exemplary application scenario is that the computer program implemented according to this application can be run on an e-commerce platform to create copywriting templates.

[0073] The advertising text can be relatively long and may include content corresponding to any one or more of the following statements: opening statement, discount information statement, description statement, and closing statement. The description statement may contain the core words of the product corresponding to the advertising text. The core words may be product words representing the product name, brand words representing the brand to which the product belongs, or both.

[0074] The advertising text can be product promotional content collected from historical product advertisements. It can be pre-collected manually, obtained by using a text recognition model to identify the text content in product advertisement images containing the product promotional content, or it can be independently written by humans.

[0075] In one embodiment, historically placed advertising copy can be directly retrieved from the advertising copy database of the e-commerce platform's advertising delivery system, and the advertising text can be extracted for use. In a further embodiment, the advertising copy can be selected based on corresponding advertising performance indicator data to identify one or more high-quality advertising texts, such as those with advertising performance indicator data exceeding a preset threshold, for use in creating copy templates. The advertising performance indicator data can be the conversion rate of the corresponding product after the advertisement is placed, or other similar metrics.

[0076] In another embodiment, based on one or more keywords provided by the user / merchant, a similarity match can be performed between the semantic vector of the keywords and the semantic vector of the advertising text in historical advertising copy. One or more similar advertising texts with the highest similarity are selected to create a text template. Using semantic similarity to determine the advertising text matching the user's keywords for text template creation provides reference content that better matches the user's subjective intent, making the final generated text template more suitable for user needs.

[0077] In another embodiment, one or more historical advertising copy examples of similar products can be obtained based on a product provided by a merchant user. The advertising text in these historical advertising copy examples can be used to create a text template. Specifically, after the merchant user selects a product for which advertising text needs to be generated, based on the category to which the product belongs, one or more historical advertising copy examples corresponding to that category can be selected from the historical advertising copy examples, and the advertising text in them can be used to create the aforementioned text template.

[0078] In this way, those skilled in the art can obtain one or more existing product-related advertising texts for creating text templates.

[0079] Step S1200: Deconstruct the advertising text into a tag sequence according to a preset tag structure. The tag structure contains multiple tag groups, and each tag group contains at least one word tag.

[0080] For any one of the aforementioned advertising texts, one or more corresponding text templates can be generated. First, based on a preset tag structure and combined with part-of-speech and semantic information, the advertising text can be deconstructed into a tag sequence.

[0081] The label structure, in one embodiment, is as follows: Figure 2 As shown, it includes tag groups for indicating the following types of information: advertising phrases, core words, reserved words, and punctuation marks. Each tag group may include one or more word tags.

[0082] The advertising sentence structure, primarily used to convey advertising messages to readers, can include one or more tag groups as needed. The advertising sentence structure can be subdivided into multiple specific sentence structures based on the chapter structure of the advertising text, such as any one or more of the following: opening sentence structure, discount information sentence structure, and closing sentence structure, as well as other sentence structures not listed but which can be flexibly added by those skilled in the art. Each specific sentence structure has a corresponding tag group. Thus, in the multiple tag groups of the advertising sentence structure, each tag group can correspond to a specific sentence structure; for example, tag groups corresponding to the opening sentence structure, discount information sentence structure, and closing sentence structure can be set separately. For example, the tag group of the advertising sentence structure can include two word tags, used to indicate whether the corresponding word segment in the advertising sentence structure belongs to the starting position or a non-starting position.

[0083] For example, the tag group for the opening sentence structure may include the opening start word tag {B-Startwords} and the opening non-start word tag {I-Startwords}; the tag group for the closing sentence structure may include the closing start word tag {B-Endwords} and the closing non-start word tag {I-Endwords}; the tag group for the discount information sentence structure may include the information start word tag {B-Sale} and the information non-start word tag {I-Sale}. It is easy to understand that in a tag sequence, when a word tag starting with B is followed by multiple similar word tags starting with I (with the same suffixes Startwords, Endwords, or Sale), these word tags with the same suffix can be considered as text corresponding to the same tag group and processed collectively.

[0084] Regarding the function of advertising sentence structures, the opening sentence generally serves a guiding role in promotion, aiming to attract user attention with sensational text; the discount information sentence generally serves to promote product offers, aiming to be captivating; and the closing sentence generally serves to strengthen user confidence in purchasing the product. The descriptions of the functions of advertising sentence structures are for reference only. Those skilled in the art can set their specific content and functions separately, as long as the corresponding word tags are set according to the tag group structure of advertising sentence structures in this application.

[0085] The core words are used to indicate the brand and / or product name of the goods. Therefore, only one tag group can be defined to correspond to the brand words or product words. Alternatively, the core words can be subdivided into brand words and product words, and tag groups can be set for each product word and brand word. Each tag group includes word tags that indicate whether the corresponding word belongs to the starting position or the non-starting position.

[0086] Similarly, for the brand term tag group, it may include the brand term starting position tag {B-Product} and the brand term non-starting position tag {I-Product}; for the product term tag group, it may include the product term starting position tag {B-Brand} and the product term non-starting position tag {I-Brand}. It is easy to understand that in a tag sequence, when a word tag starting with B is followed by multiple similar word tags starting with I (with the same suffix "Product" or "Brand"), these word tags with the same suffix can be considered as text corresponding to the same tag group and processed collectively.

[0087] In addition to the tag groups for the above words and their corresponding word tags, the tag groups corresponding to the retained words and punctuation marks may include a single word tag. For example, a single word tag included in the tag group corresponding to a retained word may be represented as {0}, indicating that the word segment indicated by the word tag is unrelated to other content and can be retained in the text template; a single word tag included in the tag group corresponding to a tag symbol may be represented as {Punc}, indicating that the corresponding single character is a punctuation mark and can be retained in the text template.

[0088] Based on the above tag structure, the advertising text can be deconstructed and converted into a corresponding tag sequence. Through deconstruction, the advertising text is mapped to word segments as units, forming a tag sequence composed of word tags corresponding to each word segment, arranged in an orderly manner according to the order of the word segments in the advertising text.

[0089] The advertising text can be deconstructed and transformed in various ways, for example:

[0090] In one embodiment, the advertising text is matched according to a preset dictionary, and the corresponding word tags are determined based on the part of speech of the matched words in the dictionary, thereby obtaining the tag sequence.

[0091] In another embodiment, a deep learning model suitable for performing sequence labeling tasks and pre-trained to a convergent state according to the label structure is used to perform sequence labeling on the advertising text, directly obtaining the label sequence. This label sequence contains word tags that annotate each word segment in the advertising text at the corresponding position according to the label structure. The deep learning model can be a neural network model implemented using methods such as LSTM+CRF, or a neural network model implemented using methods such as BERT. Those skilled in the art can flexibly choose the appropriate model based on the principles disclosed in this embodiment, as long as sufficient training samples are used to train it to a convergent state, making it suitable for generating the corresponding label sequence for an advertising text.

[0092] It should be noted that the word segmentation indicated by each of the aforementioned word tags may vary depending on the language of the advertising text. For example, for Latin languages ​​such as English, the word segmentation may be a single word; for languages ​​such as Chinese, the word segmentation may be a single character or a multi-character word obtained after segmenting the advertising text.

[0093] Step S1300: Based on the word tags in the tag sequence, extract the material text corresponding to each tag group from the advertisement text;

[0094] After obtaining the tag sequence, each word segment in the advertising text has a corresponding word tag in the tag sequence. Therefore, by comparing and mapping the tag sequence with the advertising text, the tag group corresponding to the preset tag structure can be obtained, and the advertising text can be decomposed into multiple material texts.

[0095] For example, for the advertising text: “NEW LISTING IN OUR SHOP. Choosing the High-quality T-shirt will make you more comfortable! Get yours here!”, the parsed label sequence eg_label_sequence expression is as follows:

[0096] {B-Startwords;I-Startwords;I-Startwords;I-Startwords;I-Startwords;Punc;O;O;O;B-Product;O;O;O;O;O;Punc;B-Endwords;I-Endwords;I-Endwords;Punc}

[0097] By comparing and mapping the tag sequence with the advertising text, the correspondence between the word tags corresponding to each word segment in the advertising text can be obtained, where:

[0098] The tag corresponding to the opening sentence "NEW LISTING IN OUR SHOP." is:

[0099] “NEW”{B-Startwords}

[0100] “LISTING” {I-Startwords}

[0101] “IN”{I-Startwords}

[0102] “OUR”{I-Startwords}

[0103] “SHOP”{I-Startwords}

[0104] "."{Punc}

[0105] The sentence containing the retained words, "Choosing The High-quality T-shirt Will Make You More Comfortable," is tagged with the following words:

[0106] “Choosing”{O}

[0107] “The”{O}

[0108] “High-quality”{O}

[0109] “T-shirt”{B-Product}

[0110] “Will”{0}

[0111] “Make”{0}

[0112] “You”{0}

[0113] "More"{0}

[0114] "Comfortable"{0}

[0115] "!"{Punc}

[0116] The tag corresponding to the closing phrase "Get yours here!" is:

[0117] "Get"{B-Endwords}

[0118] "yours" {I-Endwords}

[0119] "here" {I-Endwords}

[0120] "!"{Punc}

[0121] Based on the above comparison and mapping results, it can be seen that by combining the word segments corresponding to the tag groups {B-Startwords} and {I-Startwords} belonging to the tag group corresponding to the opening sentence, the corresponding opening sentence material text "NEW LISTING IN OUR SHOP" is obtained. Similarly, by combining the word segments corresponding to the tag groups {B-Endwords} and {I-Endwords} belonging to the tag group corresponding to the closing sentence, the corresponding closing sentence material text "Get yours here" is obtained. Similarly, the word segments corresponding to {B-Product} can also obtain the corresponding material text, and if necessary, the word segments with the tag {0} can also be constructed as corresponding material text. It is easy to understand that each material text essentially corresponds to the tag group to which the word tags of its various word segments belong.

[0122] In an improved embodiment based on the above, after determining the material text, the mapping relationship data between the group tags of the tag group and its material text can be further stored in the material library corresponding to the category of the product described by the advertising text. The group tags include group tags for indicating that the material text corresponding to the tag group is an advertising sentence and group tags for indicating that the material text corresponding to the tag group is a core word. The core word is the product word and / or brand word of the product.

[0123] For example, corresponding group tags can be preset for each tag group in the tag structure, such as the group tag {Startwords} corresponding to the opening sentence, the group tag {Sale} corresponding to the discount information sentence, the group tag {Endwords} corresponding to the closing sentence, the group tag {Brand} corresponding to the brand word, the group tag {Product} corresponding to the product word, and other group tags can be set according to actual needs.

[0124] Based on the preset group tags, for the above example, a mapping relationship can be established between each material text and its corresponding group tag to obtain the corresponding mapping relationship data. For example, at least the corresponding mapping relationship data can be constructed for the opening sentence and the closing sentence as follows:

[0125] {Startwords}"NEW LISTING IN OUR SHOP"

[0126] {Endwords}"Get yours here"

[0127] Then, these mapping relationship data are stored in a preset material library so that they can be used as the base material for generating new advertising copy later.

[0128] The aforementioned material library can be constructed according to the product category corresponding to the advertising text. That is, a corresponding material library is set up for each category in the e-commerce platform's product category system. When the material text is extracted from the advertising text of a product and the corresponding mapping relationship data is obtained, the mapping relationship data can be stored in the material library corresponding to the product's category. Similarly, the material text for generating advertising copy is retrieved from the corresponding material library based on the product's category. By establishing material libraries categorized according to product categories, it is convenient to obtain semantically more suitable material text based on the product's category. For the automatic generation of advertising copy, the finer granularity of the material text results in a more precise semantic match in the generated advertising copy.

[0129] Step S1400: Construct a copy template based on the tag sequence. In the copy template, some tag groups are represented as corresponding replaceable group tags, while other tag groups are retained and represented as their corresponding material text.

[0130] Furthermore, a copy template can be constructed based on the tag sequence. The copy template is divided into two parts according to whether its content can be replaced. The first part acts as a mask and is represented by group tags belonging to the replaceable tag group. The second part directly quotes the original text of the advertisement text, and therefore can be represented by the material text of the tag group that needs to be retained.

[0131] Following the above example, for the opening sentence structure, closing sentence structure, discount / offer sentence structure, core keywords (including product keywords and / or brand keywords), etc., their corresponding tag groups can be pre-divided into replaceable tag groups. Therefore, in the copy template, at the location determined by referring to the advertising text, the corresponding group tags are used for representation. For the retained words, punctuation marks, etc., their corresponding tag groups can be pre-divided into tag groups to be retained. Therefore, in the copy template, at the location determined by referring to the advertising text, the corresponding original text from the advertising text is retained. Thus, based on the specific example of the above advertising text, a copy template expressing the following can be obtained:

[0132] {Startwords}.Choosing The High-quality{Product}Will Make You MoreComfortable! {Endwords}!

[0133] It should be noted that during the generation of the copy template, the punctuation marks can also be represented by the corresponding group tags in the copy template as replaceable group tags, so that the punctuation marks can be flexibly adjusted according to the actual tone later.

[0134] Once the copy template is obtained, it can be applied. In the case of a location indicated by a group tag, the corresponding group tag material text can be called from the material library to replace the group tag, thereby generating the corresponding advertising copy.

[0135] As can be seen from the above embodiments, this application has many advantages, including at least the following aspects: In the process of constructing a text template based on a given advertising text, this application utilizes a preset tag structure. By reflecting the hierarchical and granular relationship between tag groups and word tags under the tag groups in the tag structure, the material text corresponding to each tag group is accurately extracted from the advertising text. Then, in the process of constructing the corresponding text template, some tag groups are treated as replaceable tag groups and represented as group tags, which act as masks and can be used to replace other variable content. For the remaining tag groups, the original content in the advertising text is retained by representing them as their corresponding material text, thereby achieving the optimal use of the text content of the advertising text. The resulting copy template not only contains the wonderful expressions in the advertising text, but also facilitates the replacement of the group tags of the replaceable tag groups with the information corresponding to the products for which advertising copy needs to be generated. Under the guidance of the hierarchy and granularity of the tag structure, the copy template is more accurate and effective, and can be used for batch production of advertising copy, improving the generation efficiency of advertising copy required by merchants and enhancing the online advertising technology service capabilities of e-commerce platforms.

[0136] Based on any of the above embodiments, a deep learning model can be built to deconstruct the advertising text into corresponding tag sequences. An exemplary deep learning model network architecture is as follows: Figure 3 As shown, it employs a text feature extraction model to extract semantic vector vector1 for each word segment in the advertising text, and a part-of-speech extractor to extract part-of-speech vector vector2 for the advertising text. Then, through a concatenation layer, the semantic vector and part-of-speech vector of each word segment are constructed into a comprehensive vector [vector1, vector2] to integrate the semantic and part-of-speech information of each word segment. Then, a classification network is fully connected to achieve classification mapping, and the classification probability of each word segment being mapped to each word tag in each of the preset tag structures is calculated. The word tag corresponding to the maximum classification probability of each word segment is used as the word tag of that word segment.

[0137] The text feature extraction model described can be implemented using an RNN-based neural network model, including but not limited to mature models such as LSTM, Transformer, and BERT. The part-of-speech tagger can be implemented using generalized Markov model member algorithms, including HMM (Hidden Markov Model), Maximum Entropy Markov Model (MEMM), and Conditional Random Fields (CRFs), as well as deep learning algorithms represented by Recurrent Neural Networks (RNNs). Furthermore, some conventional machine learning classifiers, such as Support Vector Machines (SVMs), can also be used for part-of-speech tagging after improvements.

[0138] In a practically optimized combination of options, the text feature extraction model can be implemented using the BERT model, and the part-of-speech extractor can be implemented using the TextBlob interface provided by the Python development framework. TextBlob can be used to perform many natural language processing tasks, such as part-of-speech tagging, noun phrase extraction, sentiment analysis, text translation, and so on.

[0139] The deep learning model described above can be pre-trained to convergence by those skilled in the art using sufficient training samples, enabling it to learn the ability to generate corresponding tag sequences based on given advertising text. Since it is trained through a classification network, the loss function of the deep learning model can be constructed using the following formula:

[0140]

[0141]

[0142] Where c is the tag number, corresponding to each specific word tag in the preset tag structure, and its value range can be represented as [0, 11]. "." represents the dot product operation. The label with the highest predicted classification probability by the model. For the corresponding weights.

[0143] Based on the above exemplary deep learning models, please refer to... Figure 4 Step S1200, which involves deconstructing the advertising text into a tag sequence according to a preset tag structure, includes the following steps:

[0144] Step S1210: Obtain the semantic vector of each word segment in the advertisement text;

[0145] The advertising text is input into the text feature extraction model. The text feature extraction model performs vector embedding on each word to obtain the corresponding embedding vector, and then performs semantic extraction to obtain the semantic vector vector1 corresponding to each word. The semantic vector is a vector representation of the deep semantics of the corresponding word.

[0146] Step S1220: Obtain the part-of-speech vectors of each word segment in the advertisement text;

[0147] The advertising text is input into the part-of-speech extractor, which then labels the part of speech of each word in the tag structure according to the word tags, thereby obtaining the part-of-speech identifiers corresponding to each word in the advertising text.

[0148] Based on the part-of-speech tag corresponding to each word segment, the corresponding part-of-speech vector vector2 can be obtained by encoding each word segment of the advertising text according to its part-of-speech tag. The specific encoding method can be flexibly set; for example, the part-of-speech vector can also be obtained by using one-hot encoding.

[0149] Step S1230: Construct a comprehensive vector from the semantic vector and part-of-speech vector of each word segment;

[0150] To achieve the synthesis of semantic vectors and part-of-speech vectors for each word segment, such as Figure 3 As shown, by directly concatenating the two through a splicing layer, the corresponding comprehensive vector can be obtained. Thus, this comprehensive vector simultaneously achieves a comprehensive representation of the semantics and part-of-speech of the segmented words.

[0151] Step S1240: Based on the comprehensive vector of each word segment, determine the corresponding word tag of each word segment in the preset tag structure, and obtain the tag sequence corresponding to the advertising text.

[0152] The comprehensive vector of each word segment enters the classification network, undergoes deep feature interaction through one or more fully connected layers, and is then output to the activation layer constructed with the Softmax function. The classification probability of each word segment mapped to each word tag in the preset tag structure is calculated. For each word segment, the word tag with the highest classification probability is the word tag corresponding to that word segment. By sorting the word tags corresponding to each word segment in the entire advertisement text according to the natural order of the occurrence position of each word in the advertisement text, the tag sequence corresponding to the advertisement text can be obtained.

[0153] In one embodiment, an activation layer or a Dropout layer can be added to the classification network to improve convergence speed and avoid overfitting, so that the deep learning model can be trained to a convergent state more easily during the training phase, thereby improving its training efficiency.

[0154] As can be seen from the above embodiments, when performing sequence annotation on the advertising text according to the tag structure, this application refers to both the semantic information of each word segment in the advertising text and the part-of-speech information of each word segment. Classification mapping is performed based on the comprehensive vectors corresponding to these two types of information to determine the corresponding word tag for each word segment in the tag structure. This approach combines the semantics and part-of-speech of the word segments, avoiding being limited to single information for sequence annotation. The advantage is that the annotation result, i.e., the tag sequence, is more accurate. Furthermore, with the assistance of the sequence annotation capabilities of deep learning models, the corresponding tag sequence for the advertising text can be obtained more quickly, making it suitable for batch processing.

[0155] Based on any of the above embodiments, please refer to Figure 5 Step S1220, obtaining the part-of-speech vectors of each word segment in the advertisement text, includes the following steps:

[0156] Step S1221: Input the advertising text into the part-of-speech extractor to obtain the part-of-speech tags of each word segment;

[0157] The parts of speech can be pre-classified into multiple categories based on the grammatical structure of natural language and the textual habits used in e-commerce to represent product information. For example, the parts of speech can be defined as including the following categories: nouns, verbs, adjectives, adverbs, prepositions, conjunctions, classifiers, and others. Corresponding part-of-speech tags are set for each of these parts of speech so that the part-of-speech extractor can output according to the part-of-speech tags to indicate the corresponding part of speech.

[0158] Step S1222: Query the preset part-of-speech encoding table to determine the part-of-speech vector corresponding to each word segment. The part-of-speech encoding table is arranged in order according to the statistical frequency of different parts of speech in the reference advertising text set. The one-hot encoding algorithm is applied to determine the part-of-speech vector corresponding to each part of speech.

[0159] In order to obtain the part-of-speech vector corresponding to each word segment based on the part-of-speech identifier determined by the part-of-speech extractor, a part-of-speech encoding table can be constructed by applying the one-hot encoding algorithm, so that each part of speech has its corresponding one-hot encoding vector, which can be used as the part-of-speech vector of the corresponding part of speech segment.

[0160] When compiling the part-of-speech encoding table, the word frequency statistics of each advertising text in a reference advertising text set can be performed first to obtain the frequency data corresponding to each part of speech and determine the statistical word frequency corresponding to each part of speech. Then, each part of speech is arranged in order according to the statistical word frequency of each part of speech so as to effectively represent the statistical frequency of the part of speech through one-hot encoding.

[0161] Based on the above part-of-speech encoding principle, the dimension of the part-of-speech vector is determined to be 8, while the output vector of the exemplary text feature extraction model based on BERT is 768. Therefore, very few additional training parameters are needed, yet they can effectively supplement the features extracted by BERT.

[0162] Accordingly, the applicant provides an exemplary part-of-speech coding table as follows:

[0163]

[0164]

[0165] As can be seen from this exemplary part-of-speech encoding table, each part of speech has its corresponding part-of-speech vector. For example, the part-of-speech vector corresponding to nouns is 10000000, and the vector corresponding to adjectives is 00100000.

[0166] As can be seen from the above embodiments, one-hot encoding is used to encode parts of speech, providing a corresponding part-of-speech encoding table. After the part-of-speech extractor determines the part-of-speech identifier of each word segment, the corresponding part-of-speech vector can be quickly determined by looking up the table. The part-of-speech vector is obtained by encoding based on the statistical frequency of different parts of speech in the reference advertising text set. Therefore, it provides effective part-of-speech reference information for subsequently determining the comprehensive vector of word segmentation, and can guide the deep learning model to obtain an accurately labeled sequence.

[0167] Based on any of the above embodiments, please refer to Figure 6 Step S1400, constructing a copy template based on the tag sequence, includes the following steps:

[0168] Step S1410: Retain the word segments corresponding to the word tags in the tag groups that belong to the retention type in the advertisement text;

[0169] In the process of constructing the copy template, for the tag groups in the tag sequence that belong to the types that need to be retained in the preset tag structure, such as the tag groups corresponding to the retained words and punctuation marks mentioned in the previous example, the corresponding word segments are retained in the original text based on the advertising text, that is, there is no need to replace them.

[0170] Step S1420: Replace the word segmentation set corresponding to each consecutively occurring word tag in the tag group of the replaceable type in the advertisement text with its corresponding group tag.

[0171] In constructing the copy template, for the tag groups belonging to replaceable types in the preset tag structure within the tag sequence, such as the opening sentence, closing sentence, discount information sentence, and tag groups corresponding to core words (including product words and / or brand words) as described in the previous example, based on the advertising license, their corresponding word segments are replaced with the group tags of each corresponding tag group. Since the material text corresponding to each tag group has already been processed in the previous steps, it can be understood that here, the material text corresponding to each tag group is simply replaced with the corresponding group tags. When two identical tag groups appear consecutively, the two tags can be combined into a single tag group, thus represented by a single group tag. It is easy to understand that the expression obtained in this way constructs the copy template, and the group tags in the copy template actually act as masks, indicating that they can be replaced.

[0172] For punctuation marks in advertising text, as described above in this application, in some embodiments, when processing word tags corresponding to punctuation marks, they can also be regarded as replaceable types and represented as their group tags.

[0173] As can be seen from the above embodiments, in the process of generating copywriting templates, by corresponding to different types of tag groups, the original text is retained or represented as group tags, thus completing the construction of copywriting templates, realizing the standardization of copywriting template structure, and enabling copywriting templates to be flexibly transplanted and applied.

[0174] Based on any of the above embodiments, please refer to Figure 7 After step S1400, which involves constructing a copy template based on the tag sequence, the following steps are included:

[0175] Step S1500: Respond to the advertising copy generation request submitted by the terminal device, and obtain the product core keywords specified in the request. The product core keywords include the product keywords and / or brand keywords of the target product corresponding to the request.

[0176] When any user of an e-commerce platform, such as a merchant, needs to use the copywriting template generated by this application to automatically generate advertising copy, they can call the corresponding interface and submit an advertising copy generation request through their terminal device. This request can include the target product for which the merchant needs to generate advertising copy. The target product can be a product listed or about to be listed in the merchant's online store, and its product information can be pre-stored in the corresponding product database.

[0177] The target product has its corresponding core keywords predetermined, or they can be extracted from the product information of the target product by calling a preset core keyword extraction model. The core keywords may include only a single product keyword or a single brand keyword of the target product, or they may include both product keywords and brand keywords of the target product.

[0178] In another embodiment, the user can provide the core keywords of the product directly in text form, which can be directly included in the request for parsing and invocation.

[0179] Step S1600: Replace the group tags in the copywriting template used to indicate the core words with the product core words accordingly;

[0180] As disclosed in the preceding embodiments of this application, the text template includes replaceable group tags, which generally contain group tags corresponding to product words and / or brand words. Therefore, the group tags corresponding to the target product can be replaced with the corresponding group tags in the text template to complete the first stage of processing.

[0181] Step S1700: Replace the group tags in the copy template that indicate the advertising sentence with any material text in the material library that corresponds to the group tag and belongs to the category of the target product;

[0182] Similarly, for the remaining replaceable group tags in the copywriting template, such as the group tags corresponding to the opening and closing sentences, some prepared material text can be replaced with them.

[0183] As mentioned above, the material text can be called from the material library. The material library is obtained in advance by sequentially labeling each advertising text according to the principle of this application, and is stored in the corresponding category material library by associating the advertising text with the category of the corresponding product. Therefore, each material library stores various material texts available for products in the same category. These material texts are all associated with corresponding group tags. Based on the group tags, any one or more material texts carrying the same group tags can be called to fit into the copy template.

[0184] For generating a single advertising copy, based on each group tag belonging to the advertising sentence type in the copy template, such as the group tags corresponding to the opening sentence, closing sentence, and discount information sentence, a material text carrying the group tag is called from the corresponding material library, and the corresponding group tag in the text template is replaced.

[0185] When multiple advertising copy needs to be generated based on a copy template, for each group tag, multiple material texts can be called from the corresponding material library and respectively fitted into the copy template. Those skilled in the art should be able to adapt this flexibly.

[0186] It is easy to understand that once the group tags of the aforementioned advertising phrases are replaced with their corresponding material text, the corresponding advertising copy can be obtained.

[0187] For the resource library corresponding to the target product, since the resource library is categorized according to product type, the corresponding resource library can be determined based on the category pre-marked in the product information of the target product. Determining the corresponding resource library by product category allows access to resource texts of similar products to the target product, which is more suitable for generating advertising copy for the target product.

[0188] Taking the example of the text template generated above as an example, the text template is as follows:

[0189] {Startwords}.Choosing The High-quality{Product}Will Make You MoreComfortable! {Endwords}!

[0190] Assuming that the clothing database contains the text "HOTSUMMER IS COMING" corresponding to {Strartwords} and "Try it" corresponding to {Endwords}, and the user-specified product keyword {Product} is "Chiffon FabricShirt", then the resulting advertising copy would be:

[0191] HOT SUMMER IS COMING.Choosing The High-quality Chiffon Shirt SkirtWill Make You More Comfortable! Try it!

[0192] (Suggested translation: Summer is here. High-quality chiffon blouses will keep you comfortable! Give them a try!)

[0193] Step S1440: Push the advertising copy obtained after the copy template has been replaced to the terminal device.

[0194] After obtaining one or more corresponding advertising copy based on the copy template, these advertising copy can be pushed to the terminal device for users to further process, such as selecting one for editing and publishing, thereby guiding end users to complete the automatic generation of advertising copy.

[0195] As can be understood from the above embodiments, the copywriting templates created in this application are universal and can help users quickly generate advertising copy. Users do not need to deeply understand the text content of the advertising copy automatically generated by the system. They only need to specify the corresponding target product and submit the corresponding advertising copy generation request to obtain a trustworthy and effective advertising copy for product advertising, thereby improving advertising efficiency and enhancing the service functions of the e-commerce platform.

[0196] Please see Figure 8 To meet one of the purposes of this application, a copywriting template generation device is provided, which is a functional embodiment of the copywriting template generation method of this application. The device includes a text acquisition module 1100, a tag deconstruction module 1200, a material extraction module 1300, and a template construction module 1400. The text acquisition module 1100 is used to acquire advertising text for a product. The tag deconstruction module 1200 is used to deconstruct the advertising text into a tag sequence according to a preset tag structure. The tag structure includes multiple tag groups, and each tag group includes at least one word tag. The material extraction module 1300 is used to extract material text corresponding to each tag group from the advertising text based on the word tags in the tag sequence. The template construction module 1400 is used to construct a copywriting template according to the tag sequence. In the copywriting template, some tag groups are represented as corresponding replaceable group tags, while other tag groups are retained and represented as their corresponding material text.

[0197] Based on any of the above embodiments, the tag deconstruction module 1200 includes: a semantic processing submodule, used to obtain the semantic vector of each word segment in the advertising text; a part-of-speech processing submodule, used to obtain the part-of-speech vector of each word segment in the advertising text; a vector synthesis submodule, used to construct a synthesis vector from the semantic vector and part-of-speech vector of each word segment; and a part-of-speech determination submodule, used to determine the word tag corresponding to each word segment in the preset tag structure based on the synthesis vector of each word segment, and obtain the tag sequence corresponding to the advertising text.

[0198] Based on any of the above embodiments, the part-of-speech processing submodule includes: a part-of-speech extraction unit, used to input the advertising text into a part-of-speech extractor to obtain the part-of-speech identifier of each word segment; and a vector embedding unit, used to query a preset part-of-speech encoding table to determine the part-of-speech vector corresponding to each word segment. The part-of-speech encoding table is arranged in an ordered manner according to the statistical frequency of different parts of speech in the reference advertising text set, and the one-hot encoding algorithm is applied to determine the part-of-speech vector corresponding to each part of speech.

[0199] Based on any of the above embodiments, the tag structure includes tag groups for indicating the following types of information: advertising phrases, core words, reserved words, and punctuation marks, wherein,

[0200] The tag groups for the advertising sentence structure include tag groups corresponding to the opening sentence structure, discount information sentence structure, and closing sentence structure, wherein each tag group includes word tags used to indicate whether the corresponding word belongs to the starting position or the non-starting position.

[0201] The tag groups for the core words include tag groups corresponding to product words and brand words, wherein each tag group includes word tags used to indicate whether the corresponding word belongs to the starting position or a non-starting position;

[0202] The tag groups corresponding to the reserved words and punctuation marks include a single corresponding word tag.

[0203] Based on any of the above embodiments, the template construction module 1400 includes: a retention processing submodule, used to retain the word segments corresponding to the word tags in the tag groups belonging to the retention type in the advertising text; and a replacement processing submodule, used to replace the word segments corresponding to each consecutively occurring word tag in the tag groups belonging to the replaceable type in the advertising text with their corresponding group tags.

[0204] Based on any of the above embodiments, the material extraction module 1300 further includes: a material archiving module, used to store the mapping relationship data between the group tags of the tag group and their material text in a material library corresponding to the category of the product described in the advertising text. The group tags include group tags used to indicate that the material text corresponding to the tag group is an advertising sentence and group tags used to indicate that the material text corresponding to the tag group is a core word. The core word is the product word and / or brand word of the product.

[0205] Based on any of the above embodiments, the template construction module 1400 further includes: a request response module, used to respond to an advertising copy generation request submitted by a terminal device, and obtain the product core keywords specified in the request, wherein the product core keywords include product keywords and / or brand keywords of the target product corresponding to the request; a core keyword replacement module, used to replace the group tags in the copy template that indicate the core keywords with the product core keywords; an advertising replacement module, used to replace the group tags in the copy template that indicate the advertising sentence structure with any material text in the material library corresponding to the group tags in the material library corresponding to the category of the target product; and a copy push module, used to push the advertising copy obtained after the copy template is replaced to the terminal device.

[0206] To address the aforementioned technical problems, embodiments of this application also provide computer equipment. For example... Figure 9 The diagram shows the internal structure of a computer device. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected via a system bus. The computer-readable storage medium stores an operating system, a database, and computer-readable instructions. The database may store a sequence of control information. When executed by the processor, the computer-readable instructions enable the processor to implement a product search category identification method. The processor provides computing and control capabilities to support the operation of the entire computer device. The memory stores computer-readable instructions, which, when executed by the processor, enable the processor to execute the document template generation method of this application. The network interface of the computer device is used for communication with a terminal. Those skilled in the art will understand that… Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0207] In this embodiment, the processor is used to execute... Figure 8 The specific functions of each module and its sub-modules are defined within the device. The memory stores the program code and various data required to execute these modules or sub-modules. The network interface is used for data transmission between the user terminal and the server. In this embodiment, the memory stores the program code and data required to execute all modules / sub-modules in the document template generation device of this application. The server can call the server's program code and data to execute the functions of all sub-modules.

[0208] This application also provides a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the document template generation method of any embodiment of this application.

[0209] This application also provides a computer program product, including a computer program / instructions that, when executed by one or more processors, implement the steps of the method described in any embodiment of this application.

[0210] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. This computer program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. The aforementioned storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0211] In summary, the copy template generated from the advertising text in this application not only includes the compelling content of the advertising text, but also allows for easy replacement of the group tags of replaceable tag sets with information corresponding to the products for which advertising copy needs to be generated. Guided by the hierarchy and granularity of the tag structure, the copy template is more accurate and effective, and can be used for batch production of advertising copy, improving the efficiency of generating advertising copy for merchants and enhancing the online advertising technology service capabilities of e-commerce platforms.

[0212] Those skilled in the art will understand that the steps, measures, and solutions in the various operations, methods, and processes discussed in this application can be alternated, modified, combined, or deleted. Furthermore, other steps, measures, and solutions in the various operations, methods, and processes discussed in this application can also be alternated, modified, rearranged, decomposed, combined, or deleted. Furthermore, steps, measures, and solutions in the prior art that are similar to those disclosed in this application can also be alternated, modified, rearranged, decomposed, combined, or deleted.

[0213] The above description is only a partial embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for generating copywriting templates, characterized in that, Includes the following steps: Obtain the product's advertising text; The advertising text is deconstructed into a tag sequence according to a preset tag structure, including: obtaining the semantic vector of each word in the advertising text; obtaining the part-of-speech vector of each word in the advertising text; constructing a comprehensive vector from the semantic vector and part-of-speech vector of each word; determining the word tag corresponding to each word in the preset tag structure based on the comprehensive vector of each word, and obtaining the tag sequence corresponding to the advertising text; the tag structure includes multiple tag groups, each tag group containing at least one word tag; the tag structure includes tag groups for indicating advertising sentence structure, core words, reserved words, and punctuation marks; the tag groups for advertising sentence structure include tag groups corresponding to opening sentence structure, discount information sentence structure, and closing sentence structure, wherein each tag group includes a word tag for indicating whether the corresponding word belongs to the starting position or a non-starting position; Based on the word tags in the tag sequence, extract the material text corresponding to each tag group from the advertisement text; A copywriting template is constructed based on the tag sequence. In the copywriting template, some tag groups are represented as corresponding replaceable group tags, while other tag groups are retained and represented as their corresponding material text.

2. The copywriting template generation method according to claim 1, characterized in that, Obtaining the part-of-speech vectors of each word segment in the advertisement text includes the following steps: The advertising text is input into a part-of-speech extractor to obtain the part-of-speech tags for each word segment. The part-of-speech (POS) vectors corresponding to each word segment are determined by querying a preset part-of-speech encoding table. The POS encoding table is arranged in an ordered manner according to the statistical frequency of different parts of speech in the reference advertising text set. The one-hot encoding algorithm is applied to determine the POS vectors corresponding to each part of speech.

3. The copywriting template generation method according to claim 1, characterized in that: The tag groups for the core words include tag groups corresponding to product words and brand words, wherein each tag group includes word tags used to indicate whether the corresponding word belongs to the starting position or a non-starting position; The tag groups corresponding to the reserved words and punctuation marks include a single corresponding word tag.

4. The copywriting template generation method according to claim 1, characterized in that, Constructing a copy template based on the tag sequence includes the following steps: The word segments corresponding to the word tags in the tag groups that belong to the retention type in the advertisement text are retained; The word segmentation set corresponding to each consecutively occurring word tag in the tag group belonging to the replaceable type in the advertisement text is replaced and represented as its corresponding group tag.

5. The copywriting template generation method according to claim 1, characterized in that, After the step of extracting the material text corresponding to each tag group from the advertising text based on the word tags in the tag sequence, the following steps are included: The mapping relationship data between the group tags of the tag group and their material text is stored in the material library corresponding to the category of the product described in the advertising text. The group tags include group tags for indicating that the material text corresponding to the tag group is an advertising sentence and group tags for indicating that the material text corresponding to the tag group is a core word. The core word is the product word and / or brand word of the product.

6. The copywriting template generation method according to claim 5, characterized in that, After the step of constructing the copy template based on the tag sequence, the following steps are included: In response to an advertising copy generation request submitted by a terminal device, the product core keywords specified in the request are obtained. The product core keywords include the product keywords and / or brand keywords of the target product corresponding to the request. Replace the group tags in the copywriting template that indicate the core words with the product core words accordingly; Replace the group tags in the copy template that indicate the advertising phrases with any material text in the material library that corresponds to the group tag in the category of the target product; The advertising copy obtained after the copy template is replaced is pushed to the terminal device.

7. A copywriting template generation device, characterized in that, include: The text acquisition module is used to acquire the advertising text for products; The tag deconstruction module is used to deconstruct the advertising text into a tag sequence according to a preset tag structure, including: obtaining the semantic vector of each word in the advertising text; obtaining the part-of-speech vector of each word in the advertising text; constructing a comprehensive vector from the semantic vector and part-of-speech vector of each word; determining the word tag corresponding to each word in the preset tag structure based on the comprehensive vector of each word, and obtaining the tag sequence corresponding to the advertising text; the tag structure includes multiple tag groups, each tag group containing at least one word tag; the tag structure includes tag groups for indicating advertising sentence structure, core words, reserved words, and punctuation marks; the tag groups for advertising sentence structure include tag groups corresponding to opening sentence structure, discount information sentence structure, and closing sentence structure, wherein each tag group includes a word tag for indicating whether the corresponding word belongs to the starting position or a non-starting position; The material extraction module is used to extract material text corresponding to each tag group from the advertisement text based on the word tags in the tag sequence; The template construction module is used to construct a copy template based on the tag sequence. In the copy template, some tag groups are represented as corresponding replaceable group tags, while other tag groups are retained and represented as their corresponding material text.

8. The copywriting template generation device according to claim 7, characterized in that, Following the template construction module, it includes: The request and response module is used to respond to the advertising copy generation request submitted by the terminal device and obtain the product core keywords specified in the request. The product core keywords include the product keywords and / or brand keywords of the target product corresponding to the request. The core word replacement module is used to replace the group tags in the copy template that indicate core words with the product core words. The ad replacement module is used to replace the group tags in the copy template that indicate the ad phrase with any material text in the material library that corresponds to the group tag in the category of the target product. The copywriting push module is used to push the advertising copy obtained after the copywriting template has been replaced to the terminal device.

9. A computer device comprising a central processing unit and a memory, characterized in that, The central processing unit is used to invoke and run a computer program stored in the memory to perform the steps of the method as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, It stores, in the form of computer-readable instructions, a computer program implemented according to any one of claims 1 to 6, which, when invoked by a computer, executes the steps included in the corresponding method.

Images