Advertisement copy template processing method and device

Abstract

The present invention discloses a method and device for processing an advertisement copy template, the method comprising: obtaining a plurality of advertisement copies; based on the similarity information between any two of the advertisement copies, screening out at least one similar advertisement copy group from the plurality of advertisement copies; performing character matching on all advertisement copies in the similar advertisement copy group, replacing the unmatched character segments in the advertisement copy in the group with slots, so as to generate a corresponding advertisement copy template. It can be seen that the present invention can determine similar copy groups based on the similarity between advertisement copies, and determine the advertisement copy template corresponding to each group of similar copies based on character matching, so as to determine the advertisement copy template based on the relevance between the copies, which is conducive to improving the reliability and accuracy of the generated template, and providing an accurate and reliable template for subsequent advertisement copy generation or prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of natural language processing, and in particular to a method and device for processing an advertisement copy template. Background Art

[0002] With the development of Internet technology, the forms of online advertisements have become more and more complex and diversified. In order to improve the efficiency of generating advertisement content, the existing technology has begun to introduce natural language processing technology and neural network technology into the field of algorithms for automatically generating advertisements. However, when determining the templates for generating advertisement copy, the existing technology generally only manually selects the iconic advertisement copy forms without considering the relevance of the advertisement copy itself. In addition, the manual method also has the problems of low efficiency and poor accuracy. It can be seen that the existing technology has defects that need to be solved urgently. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a method and device for processing and determining advertising copy templates, which can determine the advertising copy templates based on the relevance between the copies, which is beneficial to improving the reliability and accuracy of the generated templates, and providing accurate and reliable templates for subsequent advertising copy generation or prediction.

[0004] In order to solve the above technical problems, the first aspect of the present invention discloses a method for processing an advertisement copy template, the method comprising:

[0005] Get multiple ad copies;

[0006] Based on the similarity information between any two of the advertising copies, at least one similar advertising copy group is selected from the plurality of advertising copies;

[0007] Character matching is performed on all the intra-group advertisement copies in the similar advertisement copy group, and unmatched character segments in the intra-group advertisement copies are replaced with slots to generate corresponding advertisement copy templates.

[0008] As an optional implementation manner, in the first aspect of the present invention, the selecting at least one similar advertising copy group from the plurality of advertising copies based on the similarity information between any two of the advertising copies comprises:

[0009] For any two of the plurality of advertisement copies, determining similarity information between the two advertisement copies;

[0010] Determine whether the similarity information is greater than a first similarity threshold;

[0011] When the judgment result is yes, the two advertising copies are classified into the same similar advertising copy group.

[0012] As an optional embodiment, in the first aspect of the present invention, the similarity information includes at least one of cosine similarity information, Euclidean distance similarity information, Jaccard distance similarity information, edit distance similarity information, Chebyshev distance similarity information, Hamming distance similarity information, Mahalanobis distance similarity information, Manhattan distance similarity information and Minkowski distance similarity information.

[0013] As an optional implementation, in the first aspect of the present invention, performing character matching on all the intra-group advertisement copies in the similar advertisement copy group, and replacing unmatched character segments in the intra-group advertisement copies with slots to generate corresponding advertisement copy templates, includes:

[0014] Matching all characters of the advertisements in at least two groups of the similar advertisements to each other, and determining the distinguishing character segments in the advertisements in each group that cannot be matched;

[0015] The distinguishing character segments in the advertisement copy within the group are replaced with filling slots to generate an advertisement copy template corresponding to the similar advertisement copy group.

[0016] As an optional implementation, in the first aspect of the present invention, determining the distinguishing character segments that cannot be matched in the advertisement copy in each group includes:

[0017] Determine all character segments in the ad copy that cannot be matched within each group;

[0018] Among all the character segments that cannot be matched, character segments whose segment length is greater than a preset length threshold are determined as distinguishing character segments.

[0019] As an optional embodiment, in the first aspect of the present invention, the method further comprises:

[0020] Acquire all processed advertisement copies in the similar advertisement copy group; the processed advertisement copies are advertisement copies in the group in which unmatched character segments are replaced with slots;

[0021] Calculating the text vector corresponding to each processed advertising copy;

[0022] A group vector of the similar advertising copy group is determined according to the text vectors corresponding to all the processed advertising copies.

[0023] As an optional embodiment, in the first aspect of the present invention, the method further comprises:

[0024] Get the advertising copy to be processed;

[0025] Calculating the similarity between the advertisement copy to be processed and any of the similar advertisement copy groups;

[0026] Determining whether the similarity between the advertisement copy to be processed and all the similar advertisement copy groups is less than a second similarity threshold;

[0027] If the judgment result is yes, a new similar advertising copy group is determined based on the advertising copy to be processed.

[0028] A second aspect of the present invention discloses an advertising copy template processing device, the device comprising:

[0029] An acquisition module, used to acquire multiple advertising copies;

[0030] A screening module, configured to screen out at least one similar advertising copy group from the plurality of advertising copies based on similarity information between any two of the advertising copies;

[0031] The generating module is used to perform character matching on all the intra-group advertisement copies in the similar advertisement copy group, and replace the unmatched character segments in the intra-group advertisement copies with slots to generate corresponding advertisement copy templates.

[0032] As an optional implementation, in the second aspect of the present invention, the screening module includes:

[0033] a determination unit, configured to determine similarity information between any two of the plurality of advertisement copies;

[0034] A judging unit, configured to judge whether the similarity information is greater than a first similarity threshold;

[0035] The grouping unit is used for grouping the two advertisements into the same similar advertisement group when the judgment result of the judgment unit is yes.

[0036] As an optional embodiment, in the second aspect of the present invention, the similarity information includes at least one of cosine similarity information, Euclidean distance similarity information, Jaccard distance similarity information, edit distance similarity information, Chebyshev distance similarity information, Hamming distance similarity information, Mahalanobis distance similarity information, Manhattan distance similarity information and Minkowski distance similarity information.

[0037] As an optional implementation, in the second aspect of the present invention, the specific manner in which the generating module performs character matching on all intra-group advertisement copies in the similar advertisement copy group and replaces unmatched character segments in the intra-group advertisement copies with slots to generate a corresponding advertisement copy template includes:

[0038] Matching all characters of the advertisements in at least two groups of the similar advertisements to each other, and determining the distinguishing character segments in the advertisements in each group that cannot be matched;

[0039] The distinguishing character segments in the advertisement copy within the group are replaced with filling slots to generate an advertisement copy template corresponding to the similar advertisement copy group.

[0040] As an optional implementation, in the second aspect of the present invention, the specific manner in which the generating module determines the distinguishing character segments that cannot be matched in the advertisement copy in each group includes:

[0041] Determine all character segments in the ad copy that cannot be matched within each group;

[0042] Among all the character segments that cannot be matched, character segments whose segment length is greater than a preset length threshold are determined as distinguishing character segments.

[0043] As an optional implementation, in the second aspect of the present invention, the apparatus further includes a group vector determination module, configured to perform the following steps:

[0044] Acquire all processed advertisement copies in the similar advertisement copy group; the processed advertisement copies are advertisement copies in the group in which unmatched character segments are replaced with slots;

[0045] Calculating the text vector corresponding to each processed advertising copy;

[0046] A group vector of the similar advertising copy group is determined according to the text vectors corresponding to all the processed advertising copies.

[0047] As an optional implementation, in the second aspect of the present invention, the device further includes a document group determination module, which is used to perform the following steps:

[0048] Get the advertising copy to be processed;

[0049] Calculating the similarity between the advertisement copy to be processed and any of the similar advertisement copy groups;

[0050] Determining whether the similarity between the advertisement copy to be processed and all the similar advertisement copy groups is less than a second similarity threshold;

[0051] If the judgment result is yes, a new similar advertising copy group is determined based on the advertising copy to be processed.

[0052] The third aspect of the present invention discloses another advertising copy template processing device, the device comprising:

[0053] A memory storing executable program code;

[0054] a processor coupled to the memory;

[0055] The processor calls the executable program code stored in the memory to execute part or all of the steps in the advertising copy template processing method disclosed in the first aspect of the embodiment of the present invention.

[0056] A fourth aspect of an embodiment of the present invention discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute some or all of the steps in the advertising copy template processing method disclosed in the first aspect of the embodiment of the present invention.

[0057] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0058] In an embodiment of the present invention, a plurality of advertisements are obtained; based on the similarity information between any two of the advertisements, at least one similar advertisement group is selected from the plurality of advertisements; character matching is performed on all advertisements in the similar advertisement group, and unmatched character segments in the advertisements in the group are replaced with slots to generate corresponding advertisement templates. It can be seen that the present invention can determine similar copy groups based on the similarity between advertisements, and determine the advertisement templates corresponding to each group of similar copies based on character matching, so that the advertisement templates can be determined based on the relevance between the copies, which is conducive to improving the reliability and accuracy of the generated templates, and providing accurate and reliable templates for subsequent advertisement generation or prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0060] Figure 1 It is a flowchart of an advertisement copy template processing method disclosed in an embodiment of the present invention;

[0061] Figure 2 It is a flowchart of another method for processing an advertisement copy template disclosed in an embodiment of the present invention;

[0062] Figure 3 It is a structural schematic diagram of an advertisement copy template processing device disclosed in an embodiment of the present invention;

[0063] Figure 4It is a structural schematic diagram of another advertising copy template processing device disclosed in an embodiment of the present invention;

[0064] Figure 5 It is a structural schematic diagram of another advertising copy template processing device disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0065] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0066] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or end including a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units inherent to these processes, methods, products or ends.

[0067] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0068] The present invention discloses an advertising copy template processing method and device, which can determine similar copy groups based on the similarity between advertising copies, and determine the advertising copy template corresponding to each group of similar copies based on character matching, so that the advertising copy template can be determined based on the relevance between the copies, which is conducive to improving the reliability and accuracy of the generated template, and providing an accurate and reliable template for subsequent advertising copy generation or prediction. The following is a detailed description.

[0069] Embodiment 1

[0070] See also Figure 1 , Figure 1 1 is a flow chart of a method for processing an advertisement copy template disclosed in an embodiment of the present invention. Figure 1The described method is applied to a device for automatically generating text content, and the device can be a corresponding generating terminal, generating device or generating server, and the server can be a local server or a cloud server, which is not limited in the embodiment of the present invention. Figure 1 As shown, the advertising copy template processing method may include the following operations:

[0071] 101. Get multiple advertising copies.

[0072] Optionally, the method of obtaining the advertisement copy may be by crawling the text of a specific network address, by receiving advertisement copy input after manual screening, or by obtaining advertisement text through voice recognition or image recognition, and the present invention is not limited thereto.

[0073] Optionally, the advertising copy may be promotional copy for different categories of goods or services, such as promotional copy used in the introduction page of goods or services or in merchant activities, or may be promotional copy used to attract traffic to goods or services, such as article titles or story quotes written by specific media such as WeChat public accounts to attract users, and the present invention is not limited thereto.

[0074] Optionally, the language of the advertisement copy is not limited to Chinese, and may also be English, French or other languages ​​with clear grammar, which is not limited in the present invention.

[0075] 102. Based on similarity information between any two advertising copies, select at least one similar advertising copy group from multiple advertising copies.

[0076] Specifically, the similarity between all the copies in each similar advertising copy group is relatively high, which is used to represent a class of relatively similar advertising copies, for example, copies used to promote the same type of products, or copies with similar writing styles.

[0077] 103. Perform character matching on all the intra-group advertisement copies in the similar advertisement copy group, and replace the unmatched character segments in the intra-group advertisement copies with slots to generate corresponding advertisement copy templates.

[0078] Optionally, the generated advertising copy template can be subsequently used as training material for neural networks, for example, in a neural network for advertising copy style recognition or pattern recognition. This is because the slot representations in the advertising copy template can be replaced with different nouns of the same type, which are invalid information for style recognition or pattern recognition. Therefore, the advertising copy template can be used as training material that has been effectively cleaned to train the neural network.

[0079] Optionally, the generated advertising copy template can also be used as an auxiliary template for advertising copy generation later. For example, in some copy generation needs, a specific style of advertising is required for a specific category of products. In this case, an advertising copy template corresponding to the specific style can be selected, and the character fragments corresponding to the slots generated according to the specific category of products can be filled into the advertising copy template to generate the advertising copy.

[0080] It can be seen that the method described in the embodiment of the present invention can determine similar copy groups based on the similarity between advertising copies, and determine the advertising copy template corresponding to each group of similar copies based on character matching, so that the advertising copy template can be determined based on the correlation between the copies, which is beneficial to improving the reliability and accuracy of the generated template, and providing accurate and reliable templates for subsequent advertising copy generation or prediction.

[0081] As an optional implementation, in the above step 102, based on the similarity information between any two advertisement copies, selecting at least one similar advertisement copy group from a plurality of advertisement copies includes:

[0082] For any two advertisement copies among the plurality of advertisement copies, determining similarity information between the two advertisement copies;

[0083] Determine whether the similarity information is greater than a first similarity threshold;

[0084] When the judgment result is yes, the two advertisements are classified into the same similar advertisement group.

[0085] Optionally, the expression that the similarity information is greater than the first similarity threshold is intended to limit the degree of similarity between the two advertising copies to be greater than a preset degree, and is not limited to the comparison of numerical values. This is because in some cases, the higher the value of the similarity information, the lower the degree of similarity. In this case, the similarity information being greater than the first similarity threshold means that the value of the similarity information is less than the first similarity threshold, and in most cases, the higher the value of the similarity information, the higher the degree of similarity. In this case, the similarity information being greater than the first similarity threshold means that the value of the similarity information is higher than the first similarity threshold.

[0086] Optionally, the similarity information may include at least one of cosine similarity information, Euclidean distance similarity information, Jaccard distance similarity information, edit distance similarity information, Chebyshev distance similarity information, Hamming distance similarity information, Mahalanobis distance similarity information, Manhattan distance similarity information, and Minkowski distance similarity information. Optionally, when the similarity information includes a combination of at least two of the above multiple similarity information types, the similarity information may be a weighted combination of multiple similarity information types, wherein different similarity information types may have different weights.

[0087] In a preferred embodiment, the similarity information is edit distance similarity information. The edit distance is generally used to characterize the minimum number of editing operations required to convert one string into another. Generally speaking, the smaller the edit distance, the greater the similarity between the two texts. At this time, the edit distance between the character strings corresponding to the two advertising texts, that is, the Levenshtein distance, can be calculated, and the edit distance score between the two advertising texts can be obtained. At this time, the first similarity threshold can be set to the edit distance score threshold, such as 0.5. Since the edit distance score is generally between 0 and 1, two advertising texts with a score less than the edit distance score threshold can be classified as the same similar advertising text group.

[0088] Optionally, in the present embodiment, the setting of the first similarity threshold can be set according to an experimental value or an empirical value. Preferably, the first similarity threshold can be increased to make the similarity between all the advertising copies in the same similar advertising copy group higher, so that the lengths of all the advertising copies in the same similar advertising copy group are roughly the same, with only some character fragments different, to facilitate subsequent replacement processing.

[0089] It can be seen that by implementing this optional implementation method, the similarity information between any two advertising copies among multiple advertising copies can be judged, and when it is judged that the similarity information is greater than a threshold, the two advertising copies are classified into the same similar advertising copy group, so that two or more advertising copies that are relatively similar among multiple advertising copies can be classified according to their similarity, so as to determine at least one group of advertising copies that are relatively similar to each other, thereby providing a data basis for subsequent replacement operations.

[0090] As an optional implementation, in the above step 103, character matching is performed on all the advertisements in the similar advertisement group, and unmatched character segments in the advertisements in the group are replaced with slots to generate a corresponding advertisement template, including:

[0091] Matching all characters of at least two groups of advertisements in the group of similar advertisements to each other, and determining the distinguishing character segments in each group of advertisements that cannot be matched;

[0092] The distinguishing character segments in the advertising copy within the group are replaced with filler slots to generate an advertising copy template corresponding to the similar advertising copy group.

[0093] Optionally, it can be seen from the above implementation that the advertising copy within the same group of similar advertising copy preferably tends to have the same text character length but different partial character fragments. Therefore, the above-mentioned mutual matching of all characters of at least two advertising copy within the group can be performed by comparing all characters of at least two advertising copy within the group word by word while aligning their order, extracting different characters that are compared, and determining different characters obtained by continuous comparison as the same distinguishing character fragment.

[0094] Optionally, you can first determine at least two advertising copies with the same length in the same similar advertising copy group, then compare all the characters of the at least two advertising copies in the group word by word while aligning their order, extract the different characters that are compared, and determine the different characters obtained by continuous comparison as the same distinguishing character segment.

[0095] Optionally, the distinguishing character fragments in the advertising copy in any group are replaced with filling slots. The parts of speech of the distinguishing character fragments can be marked through an existing part-of-speech tagging library, and the distinguishing character fragments in the advertising copy in any group are replaced with corresponding part-of-speech representation vectors. For example, the Jieba part-of-speech tagging tool can be used to perform part-of-speech tagging on unmatched text fragments, and the noun fragments in the text fragments, such as place names, personal names, constellation names, etc., are replaced with filling slots. The format of the filling slots is {"part of speech corresponding to the distinguishing character fragment"} to generate a template.

[0096] It can be seen that the implementation of this optional implementation method can determine the distinguishing character segments that cannot be matched in each group of advertising copy, and replace the distinguishing character segments in the advertising copy in the group with filling slots to generate advertising copy templates corresponding to similar advertising copy groups, thereby being able to determine the advertising copy template based on the distinguishing characters between similar copies, which is conducive to improving the reliability and accuracy of the generated template, and providing accurate and reliable templates for subsequent advertising copy generation or prediction.

[0097] As an optional implementation, in the above step, determining the distinguishing character segments that cannot be matched in the advertisement copy in each group includes:

[0098] Determine all character segments in the ad copy that cannot be matched within each group;

[0099] Among all the character segments that cannot be matched, character segments whose segment length is greater than a preset length threshold are determined as distinguishing character segments.

[0100] Optionally, the length threshold may be 2.

[0101] It can be seen that implementing this optional implementation method can determine the character fragments whose fragment length is greater than the length threshold among all character fragments that cannot be matched as distinguishing character fragments, so as to filter out character fragments that are too short to form specific part-of-speech meanings, thereby improving the representativeness and reliability of the generated template.

[0102] Embodiment 2

[0103] See also Figure 2 , Figure 2 1 is a flow chart of another method for processing an advertisement copy template disclosed in an embodiment of the present invention. Figure 2 The described method is applied to a device for automatically generating text content, and the device can be a corresponding generating terminal, generating device or generating server, and the server can be a local server or a cloud server, which is not limited in the embodiment of the present invention. Figure 2 As shown, the advertising copy template processing method may include the following operations:

[0104] 201. Obtain multiple advertising copies.

[0105] 202. Based on similarity information between any two advertisement copies, select at least one similar advertisement copy group from a plurality of advertisement copies.

[0106] 203. Perform character matching on all intra-group advertisement copies in the similar advertisement copy group, and replace unmatched character segments in the intra-group advertisement copies with slots to generate corresponding advertisement copy templates.

[0107] The specific technical details and technical terminology of the above steps 201-203 can be referred to the description of steps 101-103 in the first implementation, and will not be repeated here.

[0108] 204. Obtain all processed advertising copies in the similar advertising copy group.

[0109] The processed advertisement copy is an advertisement copy in which unmatched character segments are replaced with slots.

[0110] 205. Calculate the text vector corresponding to each processed advertising copy.

[0111] Optionally, a word vector calculation model, such as Tencent's word vector model, may be used to calculate the text vector corresponding to each processed advertising copy.

[0112] 206. Determine a group vector of a similar advertising copy group according to the text vectors corresponding to all processed advertising copies.

[0113] Optionally, the sum of the text vectors corresponding to all processed advertising texts is determined as the group vector of the similar advertising text group. Optionally, the group vector of the similar advertising text group can be used for subsequent grouping and matching of new advertising texts, and can also be used to represent the characteristics of the same type of advertising texts in the similar advertising text group, and can further be used as a calculation parameter for the advertising texts of this type in neural network training or other algorithm processing.

[0114] It can be seen that the method described in the embodiment of the present invention can calculate the text vector corresponding to each processed advertising copy in all processed advertising copies in the similar advertising copy group, and determine the group vector of the similar advertising copy group based on the text vectors corresponding to all processed advertising copies, so that clothing gender recognition can be performed based on the features of more modal levels of clothing products, which can effectively improve the accuracy and efficiency of clothing product gender recognition, and further effectively solve the problem of lower accuracy caused by the prior art of using only single modal data for clothing gender recognition.

[0115] As an optional implementation, the method may further include:

[0116] Get the advertising copy to be processed;

[0117] Calculate the similarity between the advertisement copy to be processed and any similar advertisement copy group;

[0118] Determine whether the similarity between the advertisement copy to be processed and all similar advertisement copy groups is less than a second similarity threshold;

[0119] If the judgment result is yes, a new similar advertising copy group is determined based on the advertising copy to be processed.

[0120] Optionally, the similarity between the advertising copy to be processed and any group of similar advertising copies can be calculated by calculating the similarity between the text vector of the copy to be processed and the group vector of any group of similar advertising copies. Optionally, the expression that the similarity is less than the second similarity threshold is intended to limit the similarity between the advertising copy to be processed and any group of similar advertising copies to be less than a preset degree, and is not limited to the comparison of numerical values. This is because in some cases, the higher the value of the similarity, the lower the similarity. At this time, the meaning of the similarity being less than the second similarity threshold is that the value of the similarity is greater than the second similarity threshold, and in most cases, the higher the value of the similarity, the higher the similarity. At this time, the meaning of the similarity being less than the second similarity threshold is that the value of the similarity is less than the second similarity threshold.

[0121] Optionally, the similarity between the to-be-processed advertisement copy and any similar advertisement copy group may include at least one of cosine similarity, Euclidean distance similarity, Jaccard distance similarity, edit distance similarity, Chebyshev distance similarity, Hamming distance similarity, Mahalanobis distance similarity, Manhattan distance similarity and Minkowski distance similarity. Optionally, when the similarity includes a combination of at least two of the above-mentioned multiple similarity information types, the similarity may be a weighted combination of multiple similarity information types, wherein different similarity information types may have different weights.

[0122] Optionally, in this embodiment, the second similarity threshold can be set according to an experimental value or an empirical value. Optionally, if it is determined that the similarity between the advertisement copy to be processed and several similar advertisement copy groups is higher than the second similarity threshold, the advertisement copy to be processed can be classified into the similar advertisement copy group with the highest similarity, so that the advertisement copy to be processed can be processed in the similar advertisement copy group later. At the same time, this classification can also be used to characterize the type identification of the advertisement copy to be processed, for example, it can be used to characterize that the advertisement copy to be processed is classified as the advertisement style or advertisement type corresponding to the similar advertisement copy group.

[0123] It can be seen that by implementing this optional implementation method, when it is determined that the similarity between the advertising copy to be processed and all similar advertising copy groups is less than the second similarity threshold, a new similar advertising copy group can be determined based on the advertising copy to be processed, thereby determining a new advertising copy group that is dissimilar to other advertising copy groups, providing a data basis for subsequent replacement operations.

[0124] Embodiment 3

[0125] See also Figure 3 , Figure 3 Schematic diagram of the structure of an advertising copy template processing device disclosed in an embodiment of the present invention. Figure 3 The described device can be applied to a corresponding automatic copywriting content generation device, which can be a corresponding generation terminal, generation device or generation server, and the server can be a local server or a cloud server, which is not limited in the embodiment of the present invention. Figure 3 As shown, the device may include:

[0126] The acquisition module 301 is used to acquire a plurality of advertisement copies.

[0127] Optionally, the method of obtaining the advertisement copy may be by crawling the text of a specific network address, by receiving advertisement copy input after manual screening, or by obtaining advertisement text through voice recognition or image recognition, and the present invention is not limited thereto.

[0128] Optionally, the advertising copy may be promotional copy for different categories of goods or services, such as promotional copy used in the introduction page of goods or services or in merchant activities, or may be promotional copy used to attract traffic to goods or services, such as article titles or story quotes written by specific media such as WeChat public accounts to attract users, and the present invention is not limited thereto.

[0129] Optionally, the language of the advertisement copy is not limited to Chinese, and may also be English, French or other languages ​​with clear grammar, which is not limited in the present invention.

[0130] The screening module 302 is used to screen out at least one similar advertising copy group from a plurality of advertising copies based on similarity information between any two advertising copies.

[0131] Specifically, the similarity between all the copies in each similar advertising copy group is relatively high, which is used to represent a class of relatively similar advertising copies, for example, copies used to promote the same type of products, or copies with similar writing styles.

[0132] The generating module 303 is used to perform character matching on all the intra-group advertisement copies in the similar advertisement copy group, and replace the unmatched character segments in the intra-group advertisement copies with slots to generate corresponding advertisement copy templates.

[0133] Optionally, the generated advertising copy template can be subsequently used as training material for neural networks, for example, in a neural network for advertising copy style recognition or pattern recognition. This is because the slot representations in the advertising copy template can be replaced with different nouns of the same type, which are invalid information for style recognition or pattern recognition. Therefore, the advertising copy template can be used as training material that has been effectively cleaned to train the neural network.

[0134] Optionally, the generated advertising copy template can also be used as an auxiliary template for advertising copy generation later. For example, in some copy generation needs, a specific style of advertising is required for a specific category of products. In this case, an advertising copy template corresponding to the specific style can be selected, and the character fragments corresponding to the slots generated according to the specific category of products can be filled into the advertising copy template to generate the advertising copy.

[0135] It can be seen that the device described in the embodiment of the present invention can determine similar copy groups based on the similarity between the advertising copies, and determine the advertising copy template corresponding to each group of similar copies based on character matching, so that the advertising copy template can be determined based on the correlation between the copies, which is conducive to improving the reliability and accuracy of the generated template, and providing accurate and reliable templates for subsequent advertising copy generation or prediction.

[0136] As an optional implementation, Figure 4 As shown, the screening module 302 includes:

[0137] The determining unit 3021 is used to determine similarity information between any two advertisement copies among the plurality of advertisement copies;

[0138] A judging unit 3022, configured to judge whether the similarity information is greater than a first similarity threshold;

[0139] The grouping unit 3023 is configured to group the two advertisements into the same similar advertisement group when the judgment result of the judgment unit 3022 is yes.

[0140] Optionally, the expression that the similarity information is greater than the first similarity threshold is intended to limit the degree of similarity between the two advertising copies to be greater than a preset degree, and is not limited to the comparison of numerical values. This is because in some cases, the higher the value of the similarity information, the lower the degree of similarity. In this case, the similarity information being greater than the first similarity threshold means that the value of the similarity information is less than the first similarity threshold, and in most cases, the higher the value of the similarity information, the higher the degree of similarity. In this case, the similarity information being greater than the first similarity threshold means that the value of the similarity information is higher than the first similarity threshold.

[0141] Optionally, the similarity information may include at least one of cosine similarity information, Euclidean distance similarity information, Jaccard distance similarity information, edit distance similarity information, Chebyshev distance similarity information, Hamming distance similarity information, Mahalanobis distance similarity information, Manhattan distance similarity information, and Minkowski distance similarity information. Optionally, when the similarity information includes a combination of at least two of the above multiple similarity information types, the similarity information may be a weighted combination of multiple similarity information types, wherein different similarity information types may have different weights.

[0142] In a preferred embodiment, the similarity information is edit distance similarity information. The edit distance is generally used to characterize the minimum number of editing operations required to convert one string into another. Generally speaking, the smaller the edit distance, the greater the similarity between the two texts. At this time, the edit distance between the character strings corresponding to the two advertising texts, that is, the Levenshtein distance, can be calculated, and the edit distance score between the two advertising texts can be obtained. At this time, the first similarity threshold can be set to the edit distance score threshold, such as 0.5. Since the edit distance score is generally between 0 and 1, two advertising texts with a score less than the edit distance score threshold can be classified as the same similar advertising text group.

[0143] Optionally, in the present embodiment, the setting of the first similarity threshold can be set according to an experimental value or an empirical value. Preferably, the first similarity threshold can be increased to make the similarity between all the advertising copies in the same similar advertising copy group higher, so that the lengths of all the advertising copies in the same similar advertising copy group are roughly the same, with only some character fragments different, to facilitate subsequent replacement processing.

[0144] It can be seen that by implementing this optional implementation method, the similarity information between any two advertising copies among multiple advertising copies can be judged, and when it is judged that the similarity information is greater than a threshold, the two advertising copies are classified into the same similar advertising copy group, so that two or more advertising copies that are relatively similar among multiple advertising copies can be classified according to their similarity, so as to determine at least one group of advertising copies that are relatively similar to each other, thereby providing a data basis for subsequent replacement operations.

[0145] As an optional implementation, the generation module 303 performs character matching on all the intra-group advertisement copies in the similar advertisement copy group, and replaces the unmatched character segments in the intra-group advertisement copies with slots to generate the corresponding advertisement copy template, including:

[0146] Matching all characters of at least two groups of advertisements in the group of similar advertisements to each other, and determining the distinguishing character segments in each group of advertisements that cannot be matched;

[0147] The distinguishing character segments in the advertising copy within the group are replaced with filler slots to generate an advertising copy template corresponding to the similar advertising copy group.

[0148] Optionally, it can be seen from the above implementation that the advertising copy within the same group of similar advertising copy preferably tends to have the same text character length but different partial character fragments. Therefore, the above-mentioned mutual matching of all characters of at least two advertising copy within the group can be performed by comparing all characters of at least two advertising copy within the group word by word while aligning their order, extracting different characters that are compared, and determining different characters obtained by continuous comparison as the same distinguishing character fragment.

[0149] Optionally, you can first determine at least two advertising copies with the same length in the same similar advertising copy group, then compare all the characters of the at least two advertising copies in the group word by word while aligning their order, extract the different characters that are compared, and determine the different characters obtained by continuous comparison as the same distinguishing character segment.

[0150] Optionally, the distinguishing character fragments in the advertising copy in any group are replaced with filling slots. The parts of speech of the distinguishing character fragments can be marked through an existing part-of-speech tagging library, and the distinguishing character fragments in the advertising copy in any group are replaced with corresponding part-of-speech representation vectors. For example, the Jieba part-of-speech tagging tool can be used to perform part-of-speech tagging on unmatched text fragments, and the noun fragments in the text fragments, such as place names, personal names, constellation names, etc., are replaced with filling slots. The format of the filling slots is {"part of speech corresponding to the distinguishing character fragment"} to generate a template.

[0151] It can be seen that the implementation of this optional implementation method can determine the distinguishing character segments that cannot be matched in each group of advertising copy, and replace the distinguishing character segments in the advertising copy in the group with filling slots to generate advertising copy templates corresponding to similar advertising copy groups, thereby being able to determine the advertising copy template based on the distinguishing characters between similar copies, which is conducive to improving the reliability and accuracy of the generated template, and providing accurate and reliable templates for subsequent advertising copy generation or prediction.

[0152] As an optional implementation, the specific manner in which the generating module 303 determines the distinguishing character segments that cannot be matched in the advertisement copy in each group includes:

[0153] Determine all character segments in the ad copy that cannot be matched within each group;

[0154] Among all the character segments that cannot be matched, character segments whose segment length is greater than a preset length threshold are determined as distinguishing character segments.

[0155] Optionally, the length threshold may be 2.

[0156] It can be seen that implementing this optional implementation method can determine the character fragments whose fragment length is greater than the length threshold among all character fragments that cannot be matched as distinguishing character fragments, so as to filter out character fragments that are too short to form specific part-of-speech meanings, thereby improving the representativeness and reliability of the generated template.

[0157] As an optional implementation, Figure 4 As shown, the device also includes a group vector determination module 304, which is used to perform the following steps:

[0158] Obtain all processed advertising copy in a similar advertising copy group;

[0159] Calculate the text vector corresponding to each processed advertising copy;

[0160] According to the text vectors corresponding to all processed advertising copies, a group vector of a group of similar advertising copies is determined.

[0161] The processed advertisement copy is the advertisement copy in the group in which the unmatched character segments are replaced with slots. Optionally, a word vector calculation model, such as Tencent word vector model, may be used to calculate the text vector corresponding to each processed advertisement copy.

[0162] Optionally, the sum of the text vectors corresponding to all processed advertising texts is determined as the group vector of the similar advertising text group. Optionally, the group vector of the similar advertising text group can be used for subsequent grouping and matching of new advertising texts, and can also be used to represent the characteristics of the same type of advertising texts in the similar advertising text group, and can further be used as a calculation parameter for the advertising texts of this type in neural network training or other algorithm processing.

[0163] It can be seen that the implementation of this optional implementation method can calculate the text vector corresponding to each processed advertising copy in all processed advertising copies in the similar advertising copy group, and determine the group vector of the similar advertising copy group based on the text vectors corresponding to all processed advertising copies, so that clothing gender recognition can be performed based on the features of more modal levels of clothing products, which can effectively improve the accuracy and efficiency of clothing product gender recognition, and further effectively solve the problem of lower accuracy caused by the prior art of using only single modal data for clothing gender recognition.

[0164] As an optional implementation, Figure 4 As shown, the device further includes a document group determination module 305, which is used to perform the following steps:

[0165] Get the advertising copy to be processed;

[0166] Calculate the similarity between the advertisement copy to be processed and any similar advertisement copy group;

[0167] Determine whether the similarity between the advertisement copy to be processed and all similar advertisement copy groups is less than a second similarity threshold;

[0168] If the judgment result is yes, a new similar advertising copy group is determined based on the advertising copy to be processed.

[0169] Optionally, the similarity between the advertising copy to be processed and any group of similar advertising copies can be calculated by calculating the similarity between the text vector of the copy to be processed and the group vector of any group of similar advertising copies. Optionally, the expression that the similarity is less than the second similarity threshold is intended to limit the similarity between the advertising copy to be processed and any group of similar advertising copies to be less than a preset degree, and is not limited to the comparison of numerical values. This is because in some cases, the higher the value of the similarity, the lower the similarity. At this time, the meaning of the similarity being less than the second similarity threshold is that the value of the similarity is greater than the second similarity threshold, and in most cases, the higher the value of the similarity, the higher the similarity. At this time, the meaning of the similarity being less than the second similarity threshold is that the value of the similarity is less than the second similarity threshold.

[0170] Optionally, the similarity between the to-be-processed advertisement copy and any similar advertisement copy group may include at least one of cosine similarity, Euclidean distance similarity, Jaccard distance similarity, edit distance similarity, Chebyshev distance similarity, Hamming distance similarity, Mahalanobis distance similarity, Manhattan distance similarity and Minkowski distance similarity. Optionally, when the similarity includes a combination of at least two of the above-mentioned multiple similarity information types, the similarity may be a weighted combination of multiple similarity information types, wherein different similarity information types may have different weights.

[0171] Optionally, in this embodiment, the second similarity threshold can be set according to an experimental value or an empirical value. Optionally, if it is determined that the similarity between the advertisement copy to be processed and several similar advertisement copy groups is higher than the second similarity threshold, the advertisement copy to be processed can be classified into the similar advertisement copy group with the highest similarity, so that the advertisement copy to be processed can be processed in the similar advertisement copy group later. At the same time, this classification can also be used to characterize the type identification of the advertisement copy to be processed, for example, it can be used to characterize that the advertisement copy to be processed is classified as the advertisement style or advertisement type corresponding to the similar advertisement copy group.

[0172] It can be seen that by implementing this optional implementation method, when it is determined that the similarity between the advertising copy to be processed and all similar advertising copy groups is less than the second similarity threshold, a new similar advertising copy group can be determined based on the advertising copy to be processed, thereby determining a new advertising copy group that is dissimilar to other advertising copy groups, providing a data basis for subsequent replacement operations.

[0173] Embodiment 4

[0174] See also Figure 5 , Figure 5 FIG. 1 is a schematic diagram of the structure of another advertising copy template processing device disclosed in an embodiment of the present invention. Figure 5As shown, the device may include:

[0175] A memory 401 storing executable program codes;

[0176] a processor 402 coupled to the memory 401;

[0177] The processor 402 calls the executable program code stored in the memory 401 to execute part or all of the steps in the advertising copy template processing method disclosed in the first or second embodiment of the present invention.

[0178] Embodiment 5

[0179] An embodiment of the present invention discloses a computer storage medium storing computer instructions. When the computer instructions are called, they are used to execute part or all of the steps in the advertising copy template processing method disclosed in the first or second embodiment of the present invention.

[0180] The device embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative work.

[0181] Through the specific description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution can be essentially or partly contributed to the prior art in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0182] Finally, it should be noted that the advertising copy template processing method and device disclosed in the embodiments of the present invention only disclose the preferred embodiments of the present invention, which are only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for processing an advertisement copy template, characterized in that: The method comprises: Get multiple ad copies; Based on the similarity information between any two of the advertising copies, at least one similar advertising copy group is selected from the plurality of advertising copies; Performing character matching on all intra-group advertisement copies in the similar advertisement copy group, and replacing unmatched character segments in the intra-group advertisement copies with slots to generate corresponding advertisement copy templates; The performing character matching on all the advertisement texts in the group of similar advertisement texts and replacing the unmatched character segments in the advertisement texts in the group with slots to generate a corresponding advertisement text template includes: Matching all characters of the advertisements in at least two groups of the similar advertisements to each other, and determining the distinguishing character segments in the advertisements in each group that cannot be matched; Replacing the distinguishing character segments in the advertising copy in the group with filling slots to generate an advertising copy template corresponding to the group of similar advertising copies; The step of determining the distinguishing character segments that cannot be matched in the advertisement copy in each group includes: Determine all character segments in the ad copy that cannot be matched within each group; Determine, among all the character segments that cannot be matched, character segments whose segment length is greater than a preset length threshold as distinguishing character segments; And, the method further comprises: Get the advertising copy to be processed; Calculating the similarity between the advertisement copy to be processed and any of the similar advertisement copy groups; Determining whether the similarity between the advertisement copy to be processed and all the similar advertisement copy groups is less than a second similarity threshold; If the judgment result is yes, a new similar advertising copy group is determined based on the advertising copy to be processed.

2. The method for processing an advertisement copy template according to claim 1, characterized in that: The step of selecting at least one similar advertising copy group from the plurality of advertising copies based on the similarity information between any two of the advertising copies comprises: For any two of the plurality of advertisement copies, determining similarity information between the two advertisement copies; Determine whether the similarity information is greater than a first similarity threshold; When the judgment result is yes, the two advertising copies are classified into the same similar advertising copy group.

3. The method for processing an advertisement copy template according to claim 2, characterized in that: The similarity information includes at least one of cosine similarity information, Euclidean distance similarity information, Jaccard distance similarity information, edit distance similarity information, Chebyshev distance similarity information, Hamming distance similarity information, Mahalanobis distance similarity information, Manhattan distance similarity information and Minkowski distance similarity information.

4. The method for processing an advertisement copy template according to claim 1, characterized in that: The method further comprises: Acquire all processed advertisement copies in the similar advertisement copy group; the processed advertisement copies are advertisement copies in the group in which unmatched character segments are replaced with slots; Calculating the text vector corresponding to each processed advertising copy; A group vector of the similar advertising copy group is determined according to the text vectors corresponding to all the processed advertising copies.

5. An advertising copy template processing device, characterized in that: The device comprises: An acquisition module, used to acquire multiple advertising copies; A screening module, configured to screen out at least one similar advertising copy group from the plurality of advertising copies based on similarity information between any two of the advertising copies; A generating module, configured to perform character matching on all the intra-group advertisement copies in the similar advertisement copy group, and replace the unmatched character segments in the intra-group advertisement copies with slots to generate a corresponding advertisement copy template; The specific method in which the generating module performs character matching on all the intra-group advertisement texts in the similar advertisement text group and replaces the unmatched character segments in the intra-group advertisement texts with slots to generate the corresponding advertisement text template includes: Matching all characters of the advertisements in at least two groups of the similar advertisements to each other, and determining the distinguishing character segments in the advertisements in each group that cannot be matched; Replacing the distinguishing character segments in the advertising copy in the group with filling slots to generate an advertising copy template corresponding to the similar advertising copy group; The specific method in which the generating module determines the distinguishing character segments that cannot be matched in the advertisement copy in each group includes: Determine all character segments in the ad copy that cannot be matched within each group; Determine, among all the character segments that cannot be matched, character segments whose segment length is greater than a preset length threshold as distinguishing character segments; Furthermore, the device further comprises a document group determination module, which is used to perform the following steps: Get the advertising copy to be processed; Calculating the similarity between the advertisement copy to be processed and any of the similar advertisement copy groups; Determining whether the similarity between the advertisement copy to be processed and all the similar advertisement copy groups is less than a second similarity threshold; If the judgment result is yes, a new similar advertising copy group is determined according to the advertising copy to be processed.

6. An advertising copy template processing device, characterized in that: The device comprises: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the advertising copy template processing method according to any one of claims 1 to 4.

7. A computer storage medium, characterized in that: The computer storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the advertising copy template processing method according to any one of claims 1 to 4.

Images