Virtual character copywriting generation method and device, storage medium and electronic equipment

Abstract

The invention relates to a virtual character copywriting generation method and device, a storage medium and electronic equipment, and relates to the technical field of big data processing.The method comprises the steps that in response to a selection operation for a virtual character in a copywriting generation interface, a target virtual character is determined, and target character attribute information of the target virtual character is obtained; in response to a selection operation for an application scene in the copywriting generation interface, determining a target copywriting generation scene, and obtaining target copywriting sample information corresponding to the target copywriting generation scene; inputting the target role attribute information and the target copywriting sample information into a preset copywriting generation model to obtain an original role copywriting; and performing copywriting auditing on the original role copywriting to obtain a target role copywriting corresponding to the target virtual role. The document generation efficiency is improved.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the technical field of big data processing. Specifically, the present disclosure relates to a method for generating virtual character copywriting, a device for generating virtual character copywriting, a computer-readable storage medium, and an electronic device. Background Art

[0002] In the related methods for generating virtual character copywriting, the copywriting is manually written by developers. However, this method has the problem of low copywriting generation efficiency.

[0003] It should be noted that the information disclosed in the above background art is only used to enhance the understanding of the background of the present disclosure. Therefore, it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0004] The purpose of the present disclosure is to provide a method for generating virtual character copywriting, a device for generating virtual character copywriting, a computer-readable storage medium, and an electronic device, so as to at least overcome to some extent the problem of low copywriting generation efficiency caused by the limitations and defects of the related art.

[0005] According to one aspect of the present disclosure, a method for generating virtual character copywriting is provided, including:

[0006] In response to a selection operation on a virtual character in the copywriting generation interface, determine a target virtual character, and obtain target character attribute information of the target virtual character;

[0007] In response to a selection operation on the application scenario in the copywriting generation interface, determine a target copywriting generation scenario, and obtain target copywriting sample information corresponding to the target copywriting generation scenario;

[0008] Input the target character attribute information and the target copywriting sample information into a preset copywriting generation model to obtain an original character copywriting;

[0009] Conduct copywriting review on the original character copywriting to obtain a target character copywriting corresponding to the target virtual character.

[0010] According to one aspect of the present disclosure, a device for generating virtual character copywriting is provided, including:

[0011] A virtual character determination module, configured to determine a target virtual character in response to a selection operation on a virtual character in the copywriting generation interface, and obtain target character attribute information of the target virtual character;

[0012] A generation scenario confirmation module, configured to determine a target copywriting generation scenario in response to a selection operation for an application scenario in the copywriting generation interface, and obtain target copywriting sample information corresponding to the target copywriting generation scenario;

[0013] An original character copywriting determination module, configured to input the target character attribute information and the target copywriting sample information into a preset copywriting generation model to obtain original character copywriting;

[0014] A target character copywriting determination module, configured to perform copywriting review on the original character copywriting to obtain target character copywriting corresponding to the target virtual character.

[0015] According to one aspect of the present disclosure, there is provided a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for generating virtual character copywriting described in any one of the above is implemented.

[0016] According to one aspect of the present disclosure, there is provided an electronic device, including:

[0017] A processor; and

[0018] A memory for storing executable instructions of the processor;

[0019] Wherein, the processor is configured to execute the method for generating virtual character copywriting described in any one of the above by executing the executable instructions.

[0020] A method for generating virtual character copywriting provided by an embodiment of the present disclosure, on the one hand, by responding to a selection operation for a virtual character in a copywriting generation interface, determining a target virtual character, and obtaining target character attribute information of the target virtual character; then responding to a selection operation for an application scenario in the copywriting generation interface, determining a target copywriting generation scenario, and obtaining target copywriting sample information corresponding to the target virtual scenario; then inputting the target character attribute information and the target copywriting sample information into a preset copywriting generation model to obtain original character copywriting; finally, performing copywriting review on the original character copywriting to obtain target character copywriting corresponding to the target virtual character, realizing the automatic generation of character copywriting, improving the generation efficiency of character copywriting, and solving the problem of low copywriting generation efficiency caused by manually writing character copywriting in the prior art; on the other hand, since the corresponding virtual character and the corresponding copywriting generation scenario can be customized, the personalized generation of character copywriting is realized; on the other hand, since the obtained character copywriting can also be reviewed, the accuracy of the obtained target character copywriting is improved.

[0021] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure. Description of the Drawings

[0022] The drawings herein are incorporated into and constitute a part of this specification, showing embodiments in accordance with the present disclosure, and are used together with the specification to explain the principles of the present disclosure. Obviously, the drawings in the following description are only some embodiments of the present disclosure, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0023] Figure 1 Schematically shows a flowchart of a method for generating a virtual character copywriting according to an exemplary embodiment of the present disclosure.

[0024] Figure 2 Schematically shows a structural example diagram of a preset copywriting generation model according to an exemplary embodiment of the present disclosure.

[0025] Figure 3 Schematically shows a structural example diagram of a preset risk detection model according to an exemplary embodiment of the present disclosure.

[0026] Figure 4 Schematically shows a structural example diagram of a preset repetition detection model according to an exemplary embodiment of the present disclosure.

[0027] Figure 5 Schematically shows an example diagram of a copywriting generation interface according to an exemplary embodiment of the present disclosure.

[0028] Figure 6 Schematically shows an example diagram of a scene of a selected target virtual character according to an exemplary embodiment of the present disclosure.

[0029] Figure 7 Schematically shows an example diagram of a scene of basic character information of a target virtual character according to an exemplary embodiment of the present disclosure.

[0030] Figure 8 Schematically shows an example diagram of a scene of personality tag information of a target virtual character according to an exemplary embodiment of the present disclosure.

[0031] Figure 9 Schematically shows an example diagram of a scene of role preference information of a target virtual character according to an exemplary embodiment of the present disclosure.

[0032] Figure 10 Schematically shows an example diagram of a scene of a determined target copywriting generation scene according to an exemplary embodiment of the present disclosure.

[0033] Figure 11 Schematically shows an example diagram of a scene of displayed historical copywriting sample information according to an exemplary embodiment of the present disclosure.

[0034] Figure 12 A schematic diagram showing a scene example of a displayed target character copy according to an exemplary embodiment of the present disclosure.

[0035] Figure 13 A block diagram schematically showing a generating device for virtual character copy according to an exemplary embodiment of the present disclosure.

[0036] Figure 14 An electronic device schematically showing a method for implementing the generation of virtual character copy according to an exemplary embodiment of the present disclosure. Detailed implementation manners

[0037] Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be more thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art. The features, structures, or characteristics described may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of the embodiments of the present disclosure. However, those skilled in the art will realize that the technical solutions of the present disclosure can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, well-known technical solutions are not shown or described in detail to avoid obscuring the various aspects of the present disclosure.

[0038] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus their repeated description will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0039] In traditional role-playing game scenarios, in order to make the game more story-driven, a large amount of copy needs to be generated for story-telling, character dialogues, and game interaction scenarios; therefore, in order to improve the user experience of users in role-playing games, a large number of copywriters need to be invested in the traditional production pipeline for the output and review of copy.

[0040] Meanwhile, in the copywriting production pipeline of traditional role-playing games, it is usually the copywriting planners responsible for corresponding characters who manually write copy such as daily activities, story plots, and character dialogues. Further, in the actual copywriting production process, generally, the copywriting requirements are first confirmed. For example, if a character's gratitude copy needs to be written, the tasks are divided according to the familiarity of each copywriting planner with the character. After the copywriting planners have each written the copy for their corresponding characters, other personnel will conduct a review, and only after the review is passed can it be configured into the corresponding copy list and applied in the game. Therefore, the above solution has the following defects:

[0041] On the one hand, in the traditional copywriting production pipeline, each copywriting planner corresponding to a character needs to be familiar with the character's character setting, speaking tone, the story plots that have occurred in the worldview, and the relationship and interaction between this character and other characters. On this premise, since not every copywriting planner is familiar with every character, it is relatively difficult to match a suitable copywriting planner for this character, and it greatly increases the entry threshold for copywriting planners. On the other hand, in some role-playing games with strong characters, the control of copywriting details is more rigorous, and there are more content scope and boundary requirements that need to be synchronized manually before creation. However, there is a large information communication cost in this process. If some information synchronization is lost, it may lead to later rework and many modifications, which not only reduces the copywriting efficiency but also reduces the accuracy of the obtained copy. On the other hand, since copywriting production also belongs to a kind of creation and innovation, writing wonderful, high-quality copy that conforms to the character set requires copywriting planners to invest a lot of man-days to possibly obtain better copywriting effects, which is difficult to accept for large-scale role-playing games or large-scale activities in long-term operation games. Therefore, how to improve the accuracy of the obtained character copy while improving the generation efficiency of character copy has become an urgent problem to be solved.

[0042] Based on this, in this exemplary embodiment, a method for generating virtual character copy is first provided. This method can run on terminal devices, servers, server clusters, cloud servers, etc. Of course, those skilled in the art can also run the method of the present disclosure on other platforms according to needs, and no special limitation is made in this exemplary embodiment. Specifically, referring to Figure 1 as shown, the method for generating virtual character copy may include the following steps:

[0043] Step S110. In response to a selection operation on a virtual character in the copywriting generation interface, determine a target virtual character, and obtain target character attribute information of the target virtual character;

[0044] Step S120. In response to a selection operation for the application scenario in the copywriting generation interface, determine the target copywriting generation scenario, and obtain target copywriting sample information corresponding to the target copywriting generation scenario;

[0045] Step S130. Input the target character attribute information and the target copywriting sample information into a preset copywriting generation model to obtain the original character copywriting;

[0046] Step S140. Conduct copywriting review on the original character copywriting to obtain the target character copywriting corresponding to the target virtual character.

[0047] In the above-described method for generating virtual character copywriting, on the one hand, by responding to a selection operation for the virtual character in the copywriting generation interface, determine the target virtual character, and obtain the target character attribute information of the target virtual character; then, in response to a selection operation for the application scenario in the copywriting generation interface, determine the target copywriting generation scenario, and obtain the target copywriting sample information corresponding to the target virtual scenario; then input the target character attribute information and the target copywriting sample information into a preset copywriting generation model to obtain the original character copywriting; finally, conduct copywriting review on the original character copywriting to obtain the target character copywriting corresponding to the target virtual character, which realizes the automatic generation of character copywriting, improves the generation efficiency of character copywriting, and solves the problem of low copywriting generation efficiency caused by manually writing character copywriting in the prior art; on the other hand, since the corresponding virtual character and the corresponding copywriting generation scenario can be customized, the personalized generation of character copywriting is realized; on the further hand, since the obtained character copywriting can also be reviewed, the accuracy of the obtained target character copywriting is improved.

[0048] Hereinafter, the method for generating virtual character copywriting recorded in the exemplary embodiments of the present disclosure will be further explained and described with reference to the accompanying drawings.

[0049] First, the technical implementation principle of the exemplary embodiments of the present disclosure will be explained and described. Specifically, in the process of actually generating the virtual character copywriting, it is first necessary to build and maintain a database including the background information of the virtual characters, and then develop a copywriting generation tool based on the large language model, which can solve the deficiencies of the above-mentioned manual creation and writing of copywriting. Further, when facing the copywriting generation tasks in different scenarios, the large language model can learn the fictional worldviews, all past background stories, and detailed information and details of each virtual character from the background database, and give copywriting creation hints and boundary explicitations in different scenarios, and perform vivid role-playing according to the set tasks to complete the specified copywriting generation tasks, solving the problems of unfamiliarity with the characters or incomplete information communication. Moreover, the copywriting generation model described in the exemplary embodiments of the present disclosure has excellent natural language understanding and creation capabilities, so it can bring the copywriting creation effect with ultra-short time cost, and the quality of the copywriting generated by it has reached or even exceeded the level of human planning. Only simple review and selection are required to obtain the target character copywriting, so that on the basis of improving the generation efficiency of the character copywriting, the accuracy of the obtained character copywriting can be improved.

[0050] Secondly, the preset copywriting generation model involved in the exemplary embodiments of the present disclosure will be explained and described. Specifically, as shown in Figure 2 the preset copywriting generation model may include a first input layer 201, a first embedding mapping layer 202, a first encoding layer 203, a first mixture-of-experts model layer 204, and a first output layer 205; among them, the specific functions played by each model layer in the process of copywriting generation will be detailed one by one later, and will not be further elaborated here. Further, the first encoding layer described here may be a multi-layer bidirectional Transformer or a Deep Interest Network (DIN). In the actual application process, it can be determined according to actual needs, and this example does not make special restrictions on this; moreover, the mixture-of-experts model layer described here can be implemented based on Expert or based on a Factor Machine (FM). In the actual application process, it can be determined according to actual needs, and this example does not make special restrictions on this.

[0051] Next, the specific training process of the preset copywriting generation model will be explained and described. Specifically, in the actual application process, first, all the original virtual characters included in a certain game scene and the original character attribute information of each original virtual character are obtained; second, the original copywriting generation scenes included in the game scene and the scene copywriting samples corresponding to each original copywriting generation scene are determined; then, a training data set is constructed based on the original character attribute information, the original copywriting generation scenes, and the scene copywriting samples of each original virtual character; finally, the low-rank adaptation model is trained based on the training data set to obtain the corresponding low-rank matrix parameters, and then the large model is fine-tuned based on the low-rank matrix parameters to obtain the preset copywriting generation model.

[0052] Next, the preset risk detection model involved in the exemplary embodiments of the present disclosure will be explained and described. Specifically, referring to Figure 3 as shown, the preset risk detection model may include a second input layer 301, multiple Transformer models (the first Transformer model, the second Transformer model,..., the Nth Transformer model) 302, and a second output layer 303. Among them, the specific functions played by each model layer in the risk detection process will be detailed later, and will not be further elaborated here. It should also be supplemented here that during the training of the preset risk detection model, a certain number of sensitive words can be extracted from the corresponding sensitive word library and added to the corresponding copywriting for training.

[0053] Next, the preset duplicate detection model involved in the exemplary embodiments of the present disclosure will be explained and described. Specifically, referring to Figure 4 as shown, the preset duplicate detection model may include a third input layer 401, a sentence embedding model 402, a cosine similarity calculation layer 403, and a third output layer 404; further, the sentence embedding model recorded here may include a second embedding mapping layer and an average pooling layer. Among them, the specific functions played by each model layer in the duplicate detection process will be detailed later, and will not be further elaborated here.

[0054] Next, in combination with Figures 2 - 4 to Figure 1 the generation method of the virtual character copywriting shown in will be further explained and described. Specifically:

[0055] In step S110, in response to a selection operation on the virtual character in the copywriting generation interface, a target virtual character is determined, and the target character attribute information of the target virtual character is obtained.

[0056] In this exemplary embodiment, first, determine the target virtual character from the copywriting generation interface; wherein, the copywriting generation interface described herein can be referred to Figure 5 as shown; specifically, as Figure 5 shown, the copywriting generation interface may include multiple different interactive controls. For example, copywriting generation scene (which can also be understood as the application scene) interactive control, data import interactive control, character rarity selection interactive control, character name search box (which supports both precise search and fuzzy search), and the corresponding character name list; at the same time, each character name included in the character name list is interactive; if a certain character name is clicked, relevant information about the virtual character corresponding to the character name can be displayed; on this premise, if it is necessary to determine the target virtual character, it can be directly selected from the character name list, or searched first and then selected based on the search results; at the same time, the number of selected target virtual characters can be one or more, and in the actual application process, it can be selected according to actual needs, and this example does not make special restrictions on this; further, the selected target virtual character can be referred to Figure 6 as shown.

[0057] Secondly, obtain the target character attribute information of the target virtual character; wherein, the target character attribute information described herein may include, but is not limited to, the basic character information, personality tag information, character preference information, story framework information, etc. of the target virtual character; specifically, taking a certain character in a certain game as an example, the basic character information described herein may include the character name, gender, rarity, height, weapon, character tag, voice-over (Character Voice, CV), person, real name, catchphrase, and place of residence, etc. of the target virtual character, and the specific example diagram can be referred to Figure 7 as shown; the personality tag information described herein may include, but is not limited to, the impression color, character advantages, character disadvantages, hobbies, small actions / habits, special skill information of the target virtual character, personality evaluation, and bond character information associated with the target virtual character, etc., and the specific example diagram can be referred to Figure 8 as shown; the character preference information described herein may include, but is not limited to, the action motivation, things good at, things not good at, things liked, things hated, weaknesses, advantages, etc. of the target virtual character, and the specific example diagram can be referred to Figure 9As shown; the story framework information recorded herein may include the background world view of the target game where the target virtual character is located, the location where the game story involved in the target game takes place, and the basic game story framework; further, the background world view recorded herein may be used to represent the era background of the target game, such as ancient Tang Dynasty, ancient Qing Dynasty, ancient Egypt; or modern times, future, end times, interstellar, etc.; the location where the game story takes place recorded herein may be used to represent the location corresponding to the game map corresponding to the target game, such as desert, city, countryside, garden, China or foreign countries, etc.; of course, the background world view may also include the purpose of the game, such as making friends, raising cute pets, leveling up by killing monsters, etc.

[0058] In step S120, in response to a selection operation for the application scenario in the copywriting generation interface, determine the target copywriting generation scenario, and obtain target copywriting sample information corresponding to the target copywriting generation scenario.

[0059] In this exemplary embodiment, first, determine the target copywriting generation scenario; among them, the determined target copywriting generation scenario may be as Figure 10 shown; in the actual application process, one target copywriting generation scenario can be selected, or multiple target copywriting generation scenarios can be selected, and this exemplary embodiment does not make special restrictions on this; however, if multiple copywriting generation scenarios are selected for the same target virtual character, an association between the target virtual character and the target copywriting generation scenario needs to be established, and target copywriting sample information under different target copywriting generation scenarios needs to be configured for the target virtual character respectively.

[0060] Secondly, obtain target copywriting sample information corresponding to the target copywriting generation scenario; specifically, it can be achieved in the following way: in response to an interaction operation for the sample import control in the copywriting generation interface, import historical copywriting sample information corresponding to the target virtual character, and display the historical copywriting sample information; in response to a selection operation for the displayed historical copywriting sample information, and determine whether the selected historical copywriting sample information needs to be modified; if it needs to be modified, modify the historical copywriting sample information, and determine the target copywriting sample information according to the modified historical copywriting sample information; if it does not need to be modified, determine the target copywriting sample information according to the selected historical copywriting sample information. That is, in the actual application process, historical copywriting sample information of the same virtual character under different copywriting generation scenarios and historical copywriting sample information of different virtual characters under the same copywriting generation scenario can be imported through external import, etc.; at the same time, after the historical copywriting sample information is imported, the historical copywriting sample information can be displayed; among them, the displayed historical copywriting sample information can be referred to Figure 11As shown; in the actual application process, it can be determined whether it is necessary to modify the historical copy sample information. If modification is required, click on the Figure 11 modification control shown in it for modification. The specific modification methods can include but are not limited to deletion, modification, addition, etc. This example does not make special restrictions on this; if no modification is required, the selected historical copy sample information can be directly used as the target copy sample information.

[0061] It should be further noted here that in the actual application process, in addition to determining the target role attribute information and the target copy sample information, other information of the target virtual role can also be supplemented, such as the biography information of the target virtual role, etc. This example does not make special restrictions on this.

[0062] In step S130, the target role attribute information and the target copy sample information are input into a preset copy generation model to obtain the original role copy.

[0063] Specifically, the specific generation process of the original role copy can be shown in the following way: generate the basic role information to be predicted according to the target role attribute information, and generate the context information to be predicted according to the target copy sample information and the preset parameter prompt information; perform embedding mapping processing on the basic role information to be predicted based on the first embedding mapping layer to obtain the first virtual role feature, and perform embedding mapping processing on the context information to be predicted based on the first embedding mapping layer to obtain the first context flag sequence; perform encoding processing on the first virtual role feature and the first context flag sequence based on the first encoding layer to obtain the first context overall representation; perform copy generation on the first context flag sequence and the first context overall representation based on the first mixture-of-experts model layer to obtain the original role copy. Among them, the first embedding mapping layer recorded here can include an Embedding embedding mapping layer and a Bert embedding mapping layer; in the specific embedding mapping process, the basic role information to be predicted can be subjected to embedding mapping processing based on the Embedding embedding mapping layer to obtain the first virtual role feature, and the context information to be predicted can be subjected to embedding mapping processing based on the Bert embedding mapping layer to obtain the first context flag sequence; further, the preset parameter prompt information recorded here can be set according to actual needs; for example, if your goal is to generate copy for an xx role in an xx scenario, you need to follow the following rules: ①xxxx; ②xxxx; ③xxx; …, etc. In the process of generating copy, attention should be paid to xxxx matters; the generated copy needs to be presented in an xx way; for the same virtual role in the same scenario, x copies need to be generated (the number of copies can be set through the copy generation interface, such as 3 copies or 5 copies, etc.).

[0064] In an exemplary embodiment, the first mixture-of-experts model layer described above may include a first gating network model and a plurality of first expert neural network models; on this premise, generating the original character copy based on the first mixture-of-experts model layer for the first context token sequence and the first overall context representation can be achieved in the following manner: based on the first gating network model, determine the model weights of the first expert neural network models according to the context token sequence, and determine the target neural network model required for performing the copy generation task from the first expert neural networks according to the model weights; input the first context token sequence and the first overall context representation into the target neural network model for copy prediction to obtain the original character copy.

[0065] In an exemplary embodiment, the target neural network model required for performing the copy generation task can be determined in the following manner: based on the first gating network model, determine the first model weight of the first expert neural network models in the personality label dimension, the second model weight in the character preference dimension, and the third model weight in the story framework dimension according to the first context token sequence; based on the first model weight, the second model weight, and the third model weight, determine the first target neural network model required for performing the copy generation task in the personality label dimension, the second target neural network model required for performing the copy generation task in the character preference dimension, and the third target neural network model required for performing the copy generation task in the story framework dimension from the plurality of first expert neural network models.

[0066] In an exemplary embodiment, the first context flag sequence and the first overall context representation are input into a target neural network model for copywriting prediction to obtain the original character copywriting, which can be achieved in the following manner: The first context flag sequence and the first overall context representation are respectively input into a first target neural network model, a second target neural network model, and a third target neural network model to obtain a first copywriting prediction result in the personality label dimension, a second copywriting prediction result in the character preference dimension, and a third copywriting prediction result in the story framework dimension; according to the first copywriting prediction result, the second copywriting prediction result, and the third copywriting prediction result, the original character copywriting is generated. Specifically, in the process of generating the original character copywriting based on the first copywriting prediction result, the second copywriting prediction result, and the third copywriting prediction result, the original character copywriting can be directly obtained by splicing the first copywriting prediction result, the second copywriting prediction result, and the third copywriting prediction result, or the first weight value, the second weight value, and the third weight value can be respectively configured for the first copywriting prediction result, the second copywriting prediction result, and the third copywriting prediction result, and obtained by weighted summation. This example does not make special restrictions on this. At the same time, the first weight value, the second weight value, and the third weight value recorded here can be configured depending on expert experience or determined based on the corresponding weight value prediction model. This example does not make special restrictions on this.

[0067] In step S140, the original character copywriting is reviewed to obtain the target character copywriting corresponding to the target virtual character.

[0068] Specifically, the copywriting review process described herein can be implemented from two dimensions: the legality of the copywriting and the repeatability of the copywriting. Of course, it can also be considered whether the copywriting includes foreign characters, etc. This example does not impose special restrictions on this. Further, taking the legality and repeatability of the copywriting as an example, the specific review process can be implemented in the following manner: Based on a preset risk detection model, perform a legality review on the original character copywriting to obtain a sensitive word risk detection result, and filter the original character copywriting according to the sensitive word risk detection result to obtain a target character copywriting corresponding to the target virtual character; and / or based on a preset repeatability detection model, perform a repeatability review on the original character copywriting to obtain a repeatability detection result, and filter the original character copywriting based on the repeatability detection result to obtain a target character copywriting corresponding to the target virtual character. Further, in the process of filtering the original character copywriting based on the sensitive word risk detection result, if the sensitive word risk detection result is not empty, then filter out the original character copywriting; otherwise, retain the original character copywriting; furthermore, in the process of filtering the original character copywriting based on the repeatability detection result, if the vector similarity in the repeatability detection result is greater than a preset similarity threshold (such as 0.5 or 0.6), then filter out the original character copywriting.

[0069] In an exemplary embodiment, performing a legality review on the original character copywriting based on a preset risk detection model to obtain a sensitive word risk detection result can be implemented in the following manner: Perform word embedding on the original character copywriting to obtain a word embedding vector, a word embedding matrix, and a position embedding matrix of the original character copywriting; according to the word embedding vector, the word embedding matrix, and the position embedding matrix, generate an embedding vector, and input the embedding vector into the first Transformer model to generate a first text semantic vector; input the first text semantic vector into other Transformer models to obtain text semantic vectors corresponding to the other Transformer models; wherein, in the other Transformer models, the output of the previous Transformer model is the input of the corresponding next Transformer model; according to the embedding vector and each of the text semantic vectors, obtain a first current encoding vector of the original character copywriting, and perform a legality review on the original character copywriting according to the first current encoding vector to obtain a sensitive word risk detection result. Specifically, in the actual application process, first, encode the original character copywriting X into a word embedding matrix W t and a position embedding matrix W p ; wherein, X = (x1, x2,..., x n)。Then, add the word embedding matrix and the position embedding matrix as two vectors to obtain the total input embedding representation h0 (i.e., the word embedding vector), and then pass the input word vector representation h0 through an N-layer Transformer network to obtain the text semantic representation vector h l , which can be specifically shown as the following formula (1) and formula (2):

[0070] h0 = XW t + W p ; Formula (1)

[0071] h l = Transformer(h l-1 ), l ∈ [1, N]; Formula (2)

[0072] Among them, h l is the hidden layer vector, that is, the output of the l-th layer Transformer network.

[0073] Then, splice the results of all Transformers in BERT to obtain the first current encoding vector of the original character copywriting, which can be specifically shown as the following formula (3):

[0074] h L = concatenate([h1, h2,..., h l ); Formula (3)

[0075] Finally, based on this first current encoding vector, it can be determined whether the original character copywriting includes sensitive words.

[0076] In an exemplary embodiment, the original character copywriting is subjected to repeatability review based on a preset repeatability detection model to obtain a repeatability detection result, which can be achieved in the following manner: the original character copywriting is embedded based on the sentence embedding model to obtain an overall copywriting vector, and the cosine similarity between the overall copywriting vectors is determined based on the cosine similarity calculation layer; the original character copywriting is subjected to repeatability review based on the cosine similarity to obtain a repeatability detection result. Further, in the actual application process, during the embedding process, the original character copywriting can be first embedded based on the second embedding mapping layer in the sentence embedding model, and then the embedding result is subjected to average pooling processing through an average pooling layer (meanpooling) to obtain an overall copywriting vector.

[0077] Finally, after obtaining the target character copywriting, the target character copywriting can be displayed, and the user can select the copywriting they need according to the displayed target character copywriting and export it for use; among them, a specific scenario example diagram can be referred to Figure 12 as shown.

[0078] So far, the method for generating virtual character copywriting recorded in the exemplary embodiments of the present disclosure has been fully implemented. Based on the foregoing content, it can be known that the method for generating virtual character copywriting recorded in the exemplary embodiments of the present disclosure can greatly improve the generation efficiency of character copywriting and the accuracy of the obtained character copywriting, achieving an output of more than 200 virtual characters and a total of 20,000 sentences of copywriting per day (each sentence is about 20-50 characters), which is an efficiency unimaginable for traditional production pipelines; and because of efficient production, all characters of different importance levels can obtain high-quality copywriting coverage, forming a scale effect in gameplay, thereby greatly enhancing the user experience of players during the game process.

[0079] The following is an embodiment of the present disclosure device, which can be used to execute the method embodiment of the present disclosure. For details not disclosed in the embodiment of the present disclosure device, please refer to the method embodiment of the present disclosure.

[0080] The exemplary embodiments of the present disclosure also provide a device for generating virtual character copywriting. Specifically, referring to Figure 13 as shown, the device for generating virtual character copywriting may include a virtual character determination module 1310, a generation scenario confirmation module 1320, an original character copywriting determination module 1330, and a target character copywriting determination module 1340. Among them:

[0081] The virtual character determination module 1310 can be used to determine a target virtual character in response to a selection operation on the virtual character in the copywriting generation interface, and obtain target character attribute information of the target virtual character;

[0082] The generation scenario confirmation module 1320 can be used to determine a target copywriting generation scenario in response to a selection operation on the application scenario in the copywriting generation interface, and obtain target copywriting sample information corresponding to the target copywriting generation scenario;

[0083] The original character copywriting determination module 1330 can be used to input the target character attribute information and the target copywriting sample information into a preset copywriting generation model to obtain original character copywriting;

[0084] The target character copywriting determination module 1340 can be used to perform copywriting review on the original character copywriting to obtain target character copywriting corresponding to the target virtual character.

[0085] In an exemplary embodiment of the present disclosure, the target character attribute information includes at least one of basic character information, personality label information, character preference information, and story framework information of the target virtual character.

[0086] In an exemplary embodiment of the present disclosure, obtaining target copywriting sample information corresponding to the target copywriting generation scenario includes: in response to an interaction operation on the sample import control in the copywriting generation interface, importing historical copywriting sample information corresponding to the target virtual character, and displaying the historical copywriting sample information; in response to a selection operation on the displayed historical copywriting sample information, and determining whether the selected historical copywriting sample information needs to be modified; if modification is required, modifying the historical copywriting sample information, and determining the target copywriting sample information according to the modified historical copywriting sample information; if modification is not required, determining the target copywriting sample information according to the selected historical copywriting sample information.

[0087] In an exemplary embodiment of the present disclosure, the preset copywriting generation model includes a first embedding mapping layer, a first encoding layer, and a first mixture-of-experts model layer; wherein, inputting the target character attribute information and the target copywriting sample information into the preset copywriting generation model to obtain the original character copywriting, including: generating the to-be-predicted basic character information according to the target character attribute information, and generating the to-be-predicted context information according to the target copywriting sample information and the preset parameter prompt information; performing an embedding mapping process on the to-be-predicted basic character information based on the first embedding mapping layer to obtain a first virtual character feature, and performing an embedding mapping process on the to-be-predicted context information based on the first embedding mapping layer to obtain a first context token sequence; performing an encoding process on the first virtual character feature and the first context token sequence based on the first encoding layer to obtain a first context overall representation; performing copywriting generation on the first context token sequence and the first context overall representation based on the first mixture-of-experts model layer to obtain the original character copywriting.

[0088] In an exemplary embodiment of the present disclosure, the first mixture-of-experts model layer includes a first gating network model and a plurality of first expert neural network models; wherein, performing copywriting generation on the first context token sequence and the first context overall representation based on the first mixture-of-experts model layer to obtain the original character copywriting, including: determining the model weights of the first expert neural network models based on the first gating network model according to the context token sequence, and determining the target neural network model required for performing the copywriting generation task from the first expert neural networks according to the model weights; inputting the first context token sequence and the first context overall representation into the target neural network model for copywriting prediction to obtain the original character copywriting.

[0089] In an exemplary embodiment of the present disclosure, based on the first gating network model, determining the model weights of the first expert neural network model according to the context flag sequence, and determining the target neural network model required for performing the copywriting generation task from the first expert neural network according to the model weights, includes: based on the first gating network model, determining the first model weight of the first expert neural network model in the personality tag dimension, the second model weight in the role preference dimension, and the third model weight in the story framework dimension according to the first context flag sequence; based on the first model weight, the second model weight, and the third model weight, determining the first target neural network model required for performing the copywriting generation task in the personality tag dimension, the second target neural network model required for performing the copywriting generation task in the role preference dimension, and the third target neural network model required for performing the copywriting generation task in the story framework dimension from multiple first expert neural network models.

[0090] In an exemplary embodiment of the present disclosure, inputting the first context flag sequence and the first overall context representation into the target neural network model for copywriting prediction to obtain the original character copywriting, includes: inputting the first context flag sequence and the first overall context representation into the first target neural network model, the second target neural network model, and the third target neural network model respectively to obtain the first copywriting prediction result in the personality tag dimension, the second copywriting prediction result in the role preference dimension, and the third copywriting prediction result in the story framework dimension; generating the original character copywriting according to the first copywriting prediction result, the second copywriting prediction result, and the third copywriting prediction result.

[0091] In an exemplary embodiment of the present disclosure, performing copywriting review on the original character copywriting to obtain the target character copywriting corresponding to the target virtual character, includes: performing legality review on the original character copywriting based on a preset risk detection model to obtain a sensitive word risk detection result, and filtering the original character copywriting according to the sensitive word risk detection result to obtain the target character copywriting corresponding to the target virtual character; and / or performing repeatability review on the original character copywriting based on a preset repeatability detection model to obtain a repeatability detection result, and filtering the original character copywriting based on the repeatability detection result to obtain the target character copywriting corresponding to the target virtual character.

[0092] In an exemplary embodiment of the present disclosure, the preset risk detection model includes multiple Transformer models; wherein, based on the preset risk detection model, a legality review is performed on the original character copywriting, and a sensitive word risk detection result is obtained, including: performing word embedding on the original character copywriting to obtain a character embedding vector, a character embedding matrix, and a position embedding matrix of the original character copywriting; generating an embedding vector according to the character embedding vector, the character embedding matrix, and the position embedding matrix, and inputting the embedding vector into the first Transformer model to generate a first text semantic vector; inputting the first text semantic vector into other Transformer models to obtain text semantic vectors corresponding to the other Transformer models; wherein, in the other Transformer models, the output of the previous Transformer model is the input of the corresponding next Transformer model; obtaining a first current encoding vector of the original character copywriting according to the embedding vector and each of the text semantic vectors, and performing a legality review on the original character copywriting according to the first current encoding vector to obtain a sensitive word risk detection result.

[0093] In an exemplary embodiment of the present disclosure, the preset repeatability detection model includes a sentence embedding model and a cosine similarity calculation layer; wherein, based on the preset repeatability detection model, a repeatability review is performed on the original character copywriting, and a repeatability detection result is obtained, including: performing embedding processing on the original character copywriting based on the sentence embedding model to obtain an overall copywriting vector, and determining the cosine similarity between the overall copywriting vectors based on the cosine similarity calculation layer; performing a repeatability review on the original character copywriting based on the cosine similarity to obtain a repeatability detection result.

[0094] The specific details of each module in the above virtual character copywriting generation device have been described in detail in the corresponding virtual character copywriting generation method, so they will not be repeated here.

[0095] It should be noted that although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of the two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0096] In addition, although the various steps of the methods in the present disclosure are described in a specific order in the drawings, this does not require or imply that these steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step for execution, and / or one step may be decomposed into multiple steps for execution, etc.

[0097] In an exemplary embodiment of the present disclosure, there is also provided an electronic device capable of implementing the above method. Those skilled in the art can understand that various aspects of the present disclosure can be implemented as a system, method, or program product. Therefore, various aspects of the present disclosure can be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or an implementation combining hardware and software aspects, which can be collectively referred to as a circuit, module, or system here.

[0098] The following refers to Figure 14 to describe the electronic device 1400 according to this embodiment of the present disclosure. Figure 14 The shown electronic device 1400 is merely an example and should not impose any limitations on the functions and usage scope of the embodiments of the present disclosure.

[0099] As Figure 14 shown, the electronic device 1400 is presented in the form of a general-purpose computing device. The components of the electronic device 1400 may include, but are not limited to: at least one of the above-mentioned processing units 1410, at least one of the above-mentioned storage units 1420, a bus 1430 connecting different system components (including the storage unit 1420 and the processing unit 1410), and a display unit 1440.

[0100] Among them, the storage unit stores program code, and the program code can be executed by the processing unit 1410, so that the processing unit 1410 executes the steps according to various exemplary embodiments of the present disclosure described in the above "Exemplary Method" section of this specification. For example, the processing unit 1410 can execute steps S110 as shown in Figure 1 : In response to a selection operation for a virtual character in the copywriting generation interface, determine a target virtual character and obtain target character attribute information of the target virtual character; step S120: In response to a selection operation for an application scenario in the copywriting generation interface, determine a target copywriting generation scenario and obtain target copywriting sample information corresponding to the target copywriting generation scenario; step S130: Input the target character attribute information and the target copywriting sample information into a preset copywriting generation model to obtain an original character copywriting; step S140: Perform copywriting review on the original character copywriting to obtain a target character copywriting corresponding to the target virtual character.

[0101] The storage unit 1420 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 14201 and / or a cache storage unit 14202, and may further include a read-only storage unit (ROM) 14203.

[0102] The storage unit 1420 may also include a program / utilities 14204 having a set (at least one) of program modules 14205. Such program modules 14205 include, but are not limited to: an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment.

[0103] The bus 1430 may represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus structures.

[0104] The electronic device 1400 may also communicate with one or more external devices 1500 (such as a keyboard, a pointing device, a Bluetooth device, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 1400, and / or may communicate with any device that enables the electronic device 1400 to communicate with one or more other computing devices (such as a router, a modem, etc.). Such communication may be carried out through an input / output (I / O) interface 1450. Moreover, the electronic device 1400 may also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 1460. As shown in the figure, the network adapter 1460 communicates with other modules of the electronic device 1400 through the bus 1430. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 1400, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0105] Through the description of the above embodiments, those skilled in the art can easily understand that the exemplary embodiments described herein can be implemented by software, or can be implemented by a combination of software and necessary hardware. Therefore, the technical solutions according to the embodiments of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiments of the present disclosure.

[0106] In an exemplary embodiment of the present disclosure, there is also provided a computer-readable storage medium, on which a program product capable of implementing the above methods in this specification is stored. In some possible implementation manners, various aspects of the present disclosure can also be implemented in the form of a program product, which includes program code. When the program product runs on a terminal device, the program code is used to cause the terminal device to execute the steps according to various exemplary embodiments of the present disclosure described in the above "Exemplary Method" section of this specification.

[0107] A program product for implementing the above method according to an embodiment of the present disclosure is described. It may be in the form of a portable compact disc read-only memory (CD-ROM) and include program code, and can run on a terminal device, such as a personal computer. However, the program product of the present disclosure is not limited thereto. In this document, a readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0108] The program product may adopt any combination of one or more readable media. The readable media may be a readable signal medium or a readable storage medium. The readable storage medium may, for example, but not be limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0109] The computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries the readable program code. Such a propagated data signal may take various forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The readable signal medium may also be any readable medium other than the readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0110] The program code contained on the readable medium can be transmitted by any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above.

[0111] Program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, C++, etc., and also including conventional procedural programming languages such as the "C" language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user's computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0112] In addition, the above-mentioned drawings are only schematic illustrations of the processes included in the method according to the exemplary embodiments of the present disclosure, rather than for limiting purposes. It is easy to understand that the processes shown in the above-mentioned drawings do not indicate or limit the chronological order of these processes. Additionally, it is also easy to understand that these processes may be executed synchronously or asynchronously in, for example, multiple modules.

[0113] Other embodiments of the present disclosure will be readily apparent to those skilled in the art after considering the specification and practicing the invention herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include known common knowledge or conventional technical means in the technical field not invented by the present disclosure. The specification and examples are only to be considered exemplary, and the true scope and spirit of the present disclosure are pointed out by the claims.

Claims

1. A method for generating virtual character copywriting, characterized in that, Including: In response to a selection operation for a virtual character in the copywriting generation interface, determine a target virtual character, and obtain target character attribute information of the target virtual character; In response to a selection operation for an application scenario in the copywriting generation interface, determine a target copywriting generation scenario, and obtain target copywriting sample information corresponding to the target copywriting generation scenario; Input the target character attribute information and the target copywriting sample information into a preset copywriting generation model to obtain an original character copy; Conduct copywriting review on the original character copy to obtain a target character copy corresponding to the target virtual character.

2. The method for generating the virtual character copywriting according to claim 1, wherein The target character attribute information includes at least one of basic character information, personality label information, character preference information, and story framework information of the target virtual character.

3. The method for generating virtual character copywriting according to claim 1, wherein Obtaining target copywriting sample information corresponding to the target copywriting generation scenario includes: In response to an interaction operation for a sample import control in the copywriting generation interface, import historical copywriting sample information corresponding to the target virtual character, and display the historical copywriting sample information; In response to a selection operation for the displayed historical copywriting sample information, and determine whether the selected historical copywriting sample information needs to be modified; If modification is required, modify the historical copywriting sample information, and determine the target copywriting sample information according to the modified historical copywriting sample information; If modification is not required, determine the target copywriting sample information according to the selected historical copywriting sample information.

4. The method for generating virtual character copywriting according to claim 1, wherein The preset copywriting generation model includes a first embedding mapping layer, a first encoding layer, and a first mixture-of-experts model layer; Among them, inputting the target character attribute information and the target copywriting sample information into the preset copywriting generation model to obtain an original character copy includes: Generate predicted basic character information according to the target character attribute information, and generate predicted context information according to the target copywriting sample information and preset parameter prompt information; Based on the first embedding mapping layer, perform embedding mapping processing on the predicted basic character information to obtain a first virtual character feature, and based on the first embedding mapping layer, perform embedding mapping processing on the predicted context information to obtain a first context token sequence; Based on the first encoding layer, perform encoding processing on the first virtual character feature and the first context token sequence to obtain a first context overall representation; Based on the first mixture-of-experts model layer, perform copywriting generation on the first context token sequence and the first context overall representation to obtain an original character copy.

5. The method for generating the virtual character copywriting according to claim 4, wherein The first mixture-of-experts model layer includes a first gating network model and multiple first expert neural network models; Among them, based on the first mixture-of-experts model layer, perform copywriting generation on the first context token sequence and the first context overall representation to obtain an original character copy, including: Based on the first gating network model, determine the model weights of the first expert neural network model according to the context flag sequence, and determine the target neural network model required for performing the copywriting generation task from the first expert neural network according to the model weights; Input the first context flag sequence and the first overall context representation into the target neural network model for copywriting prediction to obtain the original character copywriting.

6. The method for generating virtual character copywriting according to claim 5, wherein, Based on the first gating network model, determine the model weights of the first expert neural network model according to the context flag sequence, and determine the target neural network model required for performing the copywriting generation task from the first expert neural network according to the model weights, including: Based on the first gating network model, determine the first model weight of the first expert neural network model in the personality label dimension, the second model weight in the character preference dimension, and the third model weight in the story framework dimension according to the first context flag sequence; Based on the first model weight, the second model weight, and the third model weight, determine the first target neural network model required for performing the copywriting generation task in the personality label dimension, the second target neural network model required for performing the copywriting generation task in the character preference dimension, and the third target neural network model required for performing the copywriting generation task in the story framework dimension from multiple first expert neural network models.

7. The method for generating a virtual character copywriting according to claim 5, characterized in that, Input the first context flag sequence and the first overall context representation into the target neural network model for copywriting prediction to obtain the original character copywriting, including: Input the first context flag sequence and the first overall context representation into the first target neural network model, the second target neural network model, and the third target neural network model respectively to obtain the first copywriting prediction result in the personality label dimension, the second copywriting prediction result in the character preference dimension, and the third copywriting prediction result in the story framework dimension; Generate the original character copywriting according to the first copywriting prediction result, the second copywriting prediction result, and the third copywriting prediction result.

8. The method for generating the virtual character copywriting according to claim 1, wherein Conduct copywriting review on the original character copywriting to obtain the target character copywriting corresponding to the target virtual character, including: Based on a preset risk detection model, conduct legality review on the original character copywriting to obtain a sensitive word risk detection result, and filter the original character copywriting according to the sensitive word risk detection result to obtain the target character copywriting corresponding to the target virtual character; and / or Based on a preset repeatability detection model, conduct repeatability review on the original character copywriting to obtain a repeatability detection result, and filter the original character copywriting based on the repeatability detection result to obtain the target character copywriting corresponding to the target virtual character.

9. The method for generating virtual character copywriting according to claim 8, wherein The preset risk detection model includes multiple Transformer models; Among them, based on the preset risk detection model, conduct legality review on the original character copywriting to obtain a sensitive word risk detection result, including: Perform word embedding on the original character copywriting to obtain the word embedding vector, word embedding matrix, and position embedding matrix of the original character copywriting; Generate an embedding vector based on the word embedding vector, word embedding matrix, and position embedding matrix, and input the embedding vector into the first Transformer model to generate a first text semantic vector; Input the first text semantic vector into other Transformer models to obtain text semantic vectors corresponding to the other Transformer models; wherein, in the other Transformer models, the output of the previous Transformer model is the input of the corresponding next Transformer model; Obtain the first current encoding vector of the original character copywriting based on the embedding vector and each of the text semantic vectors, and perform legality review on the original character copywriting based on the first current encoding vector to obtain the sensitive word risk detection result.

10. The method for generating virtual character copywriting according to claim 8, characterized in that, The preset repeatability detection model includes a sentence embedding model and a cosine similarity calculation layer; Among them, performing repeatability review on the original character copywriting based on the preset repeatability detection model to obtain the repeatability detection result, including: Performing embedding processing on the original character copywriting based on the sentence embedding model to obtain an overall copywriting vector, and determining the cosine similarity between the overall copywriting vectors based on the cosine similarity calculation layer; Performing repeatability review on the original character copywriting based on the cosine similarity to obtain the repeatability detection result.

11. A generating device for virtual character copywriting, characterized in that, Including: A virtual character determination module, configured to respond to a selection operation on a virtual character in the copywriting generation interface, determine a target virtual character, and obtain target character attribute information of the target virtual character; A generation scenario confirmation module, configured to respond to a selection operation on the application scenario in the copywriting generation interface, determine a target copywriting generation scenario, and obtain target copywriting sample information corresponding to the target copywriting generation scenario; An original character copywriting determination module, configured to input the target character attribute information and the target copywriting sample information into a preset copywriting generation model to obtain the original character copywriting; A target character copywriting determination module, configured to perform copywriting review on the original character copywriting to obtain the target character copywriting corresponding to the target virtual character.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the virtual character copywriting generation method according to any one of claims 1-10.

13. An electronic device, characterized in that, Including: A processor; And A memory, configured to store executable instructions of the processor; Wherein, the processor is configured to execute the virtual character copywriting generation method according to any one of claims 1-10 by executing the executable instructions.

Images