Stroma generation method, device and equipment of small theater intelligent machine, medium and product

Abstract

The invention discloses a plot generation method, device and equipment of a small theater intelligent machine, a medium and a product, and relates to the technical field of artificial intelligence, and the method comprises the following steps: performing word segmentation processing on a plot text at the current moment to generate a plurality of words; converting the plurality of words by using a pre-training language model of Transform, and generating a word vector corresponding to each word; taking the word vector corresponding to each word as a feature vector corresponding to the story text at the current moment; inputting the feature vector corresponding to the plot text at the current moment and the scene information at the current moment into a trained deep learning model, and determining the plot text at the next moment; and taking the story line text at the next moment as the story line text at the current moment, and returning to the step of performing word segmentation processing on the story line text at the current moment to generate a plurality of words. According to the method and the device, diversified and personalized interactive experience of the plot text can be realized, so that the interest, learning and participation experience feeling of children is improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and particularly relates to a method, device, equipment, medium and product for generating plot of a mini-theater intelligent machine. Background Art

[0002] Ordinary educational toys all focus on specific disciplines or skills, resulting in children only being able to exercise in specific fields, lacking diversity and comprehensiveness. At the same time, due to the single form, these toys lack sufficient long-term attraction for children, weakening the educational effect. At the same time, traditional story toys have the lowest attraction for children because their interactivity and creativity are relatively weak, and the determination and singleness of the stories may cause children to quickly lose interest in the same story content, so they cannot meet the need for long-term play.

[0003] In view of the above problems, there is an urgent need for a method for generating the plot of a mini-theater intelligent machine that can realize diverse and personalized interactive experiences of plot texts to improve children's interest, learning and participation experience. Summary of the Invention

[0004] The purpose of the present application is to provide a method, device, equipment, medium and product for generating the plot of a mini-theater intelligent machine, which can solve the problems of single generation of plot texts and poor interactivity of the mini-theater intelligent machine in related technologies.

[0005] To achieve the above purpose, the present application provides the following solutions:

[0006] In a first aspect, the present application provides a method for generating the plot of a mini-theater intelligent machine, including:

[0007] Performing word segmentation on the plot text at the current moment to generate multiple words; the plot text includes an educational story or a game plot, a plot category, and the corresponding character dialogue of the plot category;

[0008] Using the pre-trained language model of Transformer to transform the multiple words to generate a word vector corresponding to each word;

[0009] Taking the word vector corresponding to each word as the feature vector corresponding to the plot text at the current moment;

[0010] Inputting the feature vector corresponding to the plot text at the current moment and the scenario information at the current moment into a trained deep learning model to determine the plot text at the next moment; the scenario information at the current moment includes the historical selection of the user at the current moment, the emotional state of the character at the current moment, and the real-time operation instruction of the user at the current moment; the operation instruction includes a voice instruction and a touch instruction;

[0011] Use the next moment's plot text as the current moment's plot text, and return to the step of "performing word segmentation on the current moment's plot text to generate multiple words".

[0012] In a second aspect, a plot generation device for a mini-theater intelligent machine is provided, including:

[0013] A mini-theater intelligent machine, a far-infrared induction flashlight, a projector, and a power supply module;

[0014] An SD card slot is provided on the far-infrared induction flashlight; the SD card slot is used to insert a character NFC card, and the character NFC card is used to store character information; the character information includes character dialogues, scenes corresponding to the character dialogues, and character information;

[0015] The far-infrared induction flashlight is used to read the character NFC card to trigger corresponding plots, and irradiate the characters on the screen of the mini-theater intelligent machine through far-infrared rays to trigger the plot text corresponding to the characters;

[0016] The mini-theater intelligent machine includes a sensor and a speaker, the sensor is used to sense the user's operation instructions; the speaker is used to play the sound effects in the plot;

[0017] The mini-theater intelligent machine is used to execute the plot generation method of the mini-theater intelligent machine described in the first aspect;

[0018] The projector is used to project the plot screen corresponding to the next moment's plot text generated by the mini-theater intelligent machine onto the screen;

[0019] The power supply module is used to supply power to the mini-theater intelligent machine, the far-infrared induction flashlight, and the projector.

[0020] In a third aspect, the present application provides a computer device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor executes the computer program to implement the plot generation method of the mini-theater intelligent machine described above.

[0021] In a fourth aspect, the present application provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the plot generation method of the mini-theater intelligent machine described above is implemented.

[0022] In a fifth aspect, the present application provides a computer program product, including a computer program, and when the computer program is executed by a processor, the plot generation method of the mini-theater intelligent machine described above is implemented.

[0023] According to the specific embodiments provided by the present application, the following technical effects are disclosed in the present application:

[0024] The present application provides a method, apparatus, device, medium and product for generating a plot of a small theater intelligent machine. First, the present application performs word segmentation on the plot text at the current moment to generate multiple words; wherein, the plot text includes educational stories or game plots, plot categories and character dialogues corresponding to the plot categories. Then, a pre-trained language model of Transformer is used to convert the multiple words to generate word vectors corresponding to each word, so as to preprocess the plot text and lay a foundation for the subsequent deep learning model to accurately generate the plot text at the next moment. Further, the word vectors corresponding to each word are used as the feature vectors corresponding to the plot text at the current moment, all the selections of the user within the current moment, the emotional state of the character and the real-time operation instructions of the user to input into the trained deep learning model to determine the plot text at the next moment, and the plot text at the next moment is used as the plot text at the current moment to achieve the coherence and interest of the plot. The present application combines the voice instructions and touch instructions of the user, and conducts voice interaction with the user, especially users such as children, through the plot of the story or game, enhancing the sense of participation of the children; and enables the children to freely and real-time adjust the plot according to their own operation instructions during the play process to achieve multi-module human-computer interaction, create their own unique story plots, ensure interactivity and creativity, meet the personalized creation needs, and also learn the knowledge in educational stories. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0026] Figure 1 It is a schematic flowchart of a method for generating a plot of a small theater intelligent machine provided in an embodiment of the present application;

[0027] Figure 2 It is a flowchart of a device for generating a plot of a small theater intelligent machine provided in an embodiment of the present application;

[0028] Figure 3 It is a working operation schematic diagram of a device for generating a plot of a small theater intelligent machine provided in an embodiment of the present application;

[0029] Figure 4 It is a structural detail diagram of a device for generating a plot of a small theater intelligent machine provided in an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0030] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the protection scope of the present application.

[0031] To make the above objects, features, and advantages of the present application more obvious and understandable, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0032] As Figure 1 shown, the present application provides a method for generating a plot of a small theater intelligent machine, including:

[0033] Step 101: Perform word segmentation on the plot text at the current moment to generate multiple words; the plot text includes educational stories or game plots, plot categories, and character dialogues corresponding to the plot categories.

[0034] Step 102: Use the pre-trained language model of Transformer to transform the multiple words to generate word vectors corresponding to each word.

[0035] Step 103: Use the word vector corresponding to each word as the feature vector corresponding to the plot text at the current moment.

[0036] Specifically, for steps 101-103, which are used for preprocessing the plot text, the specific process is as follows.

[0037] Collect a large number of plot texts, including stories, plots, and character dialogues, etc. The preprocessing process includes steps such as removing noise (such as irrelevant characters, special symbols), performing normalization processing, word segmentation, part-of-speech tagging, and semantic parsing. After preprocessing, the text data is converted into word vectors or serialized forms.

[0038] In order to enable the model to understand and generate reasonable plots, the plot text may need to be annotated. For example, annotate the speaker for each dialogue, annotate the time sequence for the story or game plot, and annotate the emotional trend for the plot, etc.

[0039] Before inputting data into a deep learning model, it is also necessary to perform steps such as removing noise (such as irrelevant characters and special symbols), standardizing (such as uniform case handling), tokenization, part-of-speech tagging, and semantic parsing on the text data. For a recurrent neural network (RNN) or other deep learning models, the text usually needs to be converted into word vectors or a serialized form. This application uses the pre-trained language model of Transformer (Bidirectional Encoder Representations from Transformers, BERT) to convert the plot text into feature vectors, as follows.

[0040] Tokenization: Split the text into words or sub-word units. For example, for the sentence "The cat sat on the mat", the tokenization result is ["The","cat","sat","on","the","mat"].

[0041] Word embedding conversion: Use BERT to convert each word or sub-word unit into its corresponding word vector. For example, BERT will generate a fixed-length vector representation for each word, and these vectors can capture the semantic and context information of the words.

[0042] Feature vector representation: The final processed data is represented in the form of feature vectors. The word vectors of each word are used as the input of the model to form a serialized vector sequence. For example, after being processed by BERT, the word vectors of each word in the sentence "The cat sat on the mat" are represented as:

[0043] The: [0.1, 0.2, 0.3,..., 0.9];

[0044] cat: [0.4, 0.5, 0.6,..., 0.1];

[0045] sat: [0.7, 0.8, 0.9,..., 0.2];

[0046] on: [0.1, 0.1, 0.1,..., 0.3];

[0047] the: [0.2, 0.3, 0.4,..., 0.5];

[0048] mat: [0.3, 0.4, 0.5,..., 0.6];

[0049] Input the deep learning model: Input the converted sequence of word vectors into a recurrent neural network or other deep learning models for the task of plot generation. The model captures semantic and context information in the text through these feature vectors and generates coherent and logical plot content.

[0050] Step 104: Input the feature vector corresponding to the plot text at the current moment and the scenario information at the current moment into the trained deep learning model to determine the plot text at the next moment; the scenario information at the current moment includes the historical selections of the user at the current moment, the emotional state of the character at the current moment, and the real-time operation instructions of the user at the current moment; the operation instructions include voice instructions and touch instructions.

[0051] Step 105: Use the plot text at the next moment as the plot text at the current moment and return to the step of "performing word segmentation on the plot text at the current moment to generate multiple words".

[0052] In practical applications, through multi-modal human-computer interaction and artificial intelligence plot generation technology, the story plot is enriched to make each game experience unique. At the same time, with the help of exploration-style games and voice interaction modes, the participation of children is enhanced; game plot customization is supported, allowing children to create exclusive and unique stories during the play process to meet personalized creation needs.

[0053] In some embodiments, before step 104, it further includes: constructing a deep learning model; the training process of the deep learning model specifically includes: using all the selections of the historical user, the emotional state of the historical character, and the operation instructions of the historical user obtained at the starting moment as scenario information to generate the historical plot text at the starting moment; using the historical plot text at the starting moment as the plot text at the current moment; inputting the feature vector corresponding to the plot text at the current moment and the historical scenario information into the deep learning model to output the plot text at the next moment; training the deep learning model with the goal of minimizing the difference between the plot category of the historical plot text at the next moment and the real plot category corresponding to the user operation instructions to obtain the trained deep learning model.

[0054] Among them, the deep learning model selects a recurrent neural network. The recurrent neural network is one of the commonly used sequence models, especially the long short-term memory network or the gated recurrent unit as the basic model. For example, using the long short-term memory network is suitable for processing time series data or tasks that rely on context relationships. In plot generation, the long short-term memory network or the gated recurrent unit are common variants of the recurrent neural network, and they can better handle long-distance dependence relationships and avoid the problem of gradient disappearance.

[0055] During training, the input to a deep learning model is usually the previous context of a story or a segment of a conversation, and the output is the next sentence or event predicted by the model. To generate a continuous plot, the model can generate a part of the content and use it as the input for the next step to gradually generate a complete plot.

[0056] In the forward propagation stage, the input data passes through the various layers of the model (including the embedding layer, recurrent neural network layer, and output layer) to gradually calculate the output probability distribution. For each time step, the deep learning model predicts the output for the next time step based on the current input and the previous hidden state. In the plot text generation task, the cross-entropy loss function is usually used to measure the gap between the plot text generated by the deep learning model and the real plot text. The cross-entropy loss function can effectively measure the difference in the probability distributions between the predicted word and the target word. The model is trained using the forward propagation and backpropagation algorithms.

[0057] The plot generation process of the deep learning model is as follows.

[0058] 1. Generate a plot based on the scenario information:

[0059] When a story or game starts, the deep learning model receives the current scenario information as input, which can include the current state of the game, the user's selection history, the emotional state of the characters, etc., that is, all the user's choices, the emotional state of the characters, and the user's real-time operation instructions at the current moment. The deep learning model synthesizes this scenario information to generate an appropriate plot.

[0060] 2. Recursive generation:

[0061] Plot generation is usually a recursive process. After the model generates a line of dialogue or an event, it uses the output as the input for the next step to continue generating subsequent plot content. This method can ensure the coherence and logic of the plot. The specific implementation is as follows:

[0062] Initial input: When a story or game starts, the deep learning model receives the initial input, such as the background setting of the story or game, character information, or the user's selection, etc. This information constitutes the starting point of text generation.

[0063] Generation process: The deep learning model generates the first line of dialogue or the first event based on the initial input. The generated content is not only based on the preset plot direction but also dynamically adjusted according to the user's behavior. For example, if the user selects a specific option, the model will generate a dialogue or event related to that option.

[0064] Use the generated dialogue or event as the input for the next step and continue to generate subsequent plot content. Through recursive generation, the deep learning model can ensure the coherence and logic of the plot. The plot text generated at each step is based on the result of the previous step, making the entire plot form an organic whole. For example, if a character's dialogue is generated in the previous step, the actions of that character or the response of another character can be generated in the next step.

[0065] In addition, the deep learning model can adjust the generated plot text in real time according to the user's real-time operation instructions (such as touch screen selection or voice commands). The generated plot text not only depends on the preset plot plot trend, but also makes dynamic adjustments according to the user's behavior to achieve a personalized and interactive plot experience. The specific implementation is as follows:

[0066] Real-time input: During the game, users can interact with the intelligent machine by touching the screen, voice commands or selecting options. These inputs will be received and processed by the deep learning model in real time.

[0067] Condition judgment: The deep learning model makes condition judgments according to the user's real-time operation instructions to determine the next plot trend. For example, if the user selects a specific option, the model will judge whether the option triggers a specific plot branch.

[0068] Dynamic adjustment: According to the result of the condition judgment, the deep learning model will dynamically adjust the generated plot content. For example, if the user selects a positive option, the model may generate a positive plot development; if the user selects a negative option, the model may generate a negative plot development.

[0069] Personalized experience: Through real-time adjustment, the model can provide a personalized plot experience for each user. The choices and operations of each user will affect the development of the plot, making the game have higher replay value and attractiveness.

[0070] In some embodiments, with the goal of minimizing the loss between the plot text at the next moment and the real plot text corresponding to the user operation instruction, the deep learning model is trained. After obtaining the trained deep learning model, it further includes: obtaining the plot screen generated by the plot text at the next moment; optimizing the plot screen according to the structural similarity index or peak signal-to-noise ratio to determine the optimized plot screen; using the optimized plot screen as the plot screen.

[0071] In some embodiments, the plot screen is optimized according to the structural similarity index or the peak signal-to-noise ratio to determine the optimized plot screen, which specifically includes: determining the structural similarity index according to the mean, standard deviation and covariance of the plot screen, as well as the mean, standard deviation and covariance of the structure of the original plot screen, and maximizing the structural similarity index to determine the optimized plot screen; or, determining the peak signal-to-noise ratio according to the maximum preset pixel value of the plot screen and the mean variance between the plot screen and the structure of the original plot screen, and maximizing the peak signal-to-noise ratio to determine the optimized plot screen.

[0072] Specifically, the Structural Similarity Index Measure (SSIM): SSIM is used to evaluate the structural similarity between the generated plot screen and the target screen (the structure of the original plot screen), and it takes into account the brightness, contrast and structural information of the image. A high SSIM value indicates that the generated plot screen x is similar to the structure of the original plot screen y and is visually coherent. The SSIM formula is:

[0073]

[0074] where μ x and μ y are the means of x and y respectively, σ x and σ y are the standard deviations of x and y respectively, σ xy is the covariance of x and y, and c1 and c2 are both small constants for stabilizing the denominator.

[0075] In practical applications, the calculated value of the structural similarity index between the plot screen and the structure of the original plot screen is compared with the SSIM preset value, and based on the deep learning model, the difference between the SSIM preset value and the calculated value is minimized to obtain the final plot screen.

[0076] Specifically, the Peak Signal-to-Noise Ratio (PSNR) is mainly used to measure the difference between the generated plot screen x and the structure of the original plot screen y, especially the influence of noise. The higher the PSNR value, the closer the quality of the generated screen is to the original screen. The PSNR formula is:

[0077]

[0078] where is the maximum preset pixel value of x, and MSE is the mean square error between x and y. Specifically, is the maximum pixel value supported by the image format (e.g. 255 for an 8-bit image). MSE: The mean square error between the generated image x and the original image y, calculated by pixel-by-pixel differences.

[0079] In practical applications, the difference between the peak signal-to-noise ratio corresponding to the plot picture and the preset peak signal-to-noise ratio is compared, and based on the deep learning model, the difference between the peak signal-to-noise ratio and the preset peak signal-to-noise ratio is minimized to obtain the final plot picture.

[0080] Reference Figure 2 , a plot generation device for a small theater smart machine, comprising:

[0081] A far-infrared induction flashlight 1, a small theater smart phone 2, a projector 3 and a power module 4; the far-infrared induction flashlight 1 is provided with a card slot; the card slot is used to insert a character NFC card, and the character NFC card is used to store character information; the character information includes character dialogues, scenes corresponding to the character dialogues and character information; the far-infrared induction flashlight 1 is used to read the character NFC card to trigger the corresponding plot, and to irradiate the characters on the screen of the small theater smart phone 2 with far-infrared rays to trigger the plot text corresponding to the characters; the small theater smart phone 2 includes a sensor and a speaker, and the sensor is used to sense the user's operation instructions; the speaker is used to play the sound effects in the plot; the small theater smart phone 2 is used to execute the plot generation method of the small theater smart phone; the projector 3 is used to project the plot picture corresponding to the plot text generated by the small theater smart phone 2 onto the screen; the power module 4 is used to power the small theater smart phone 2, the far-infrared induction flashlight 1 and the projector 3.

[0082] In some embodiments, a plot generation device for a small theater smart machine also includes a communication module; the communication module is arranged in the small theater smart machine; the communication module is used to wirelessly transmit data with an external terminal; the external terminal is used for users to send real-time operation instructions.

[0083] Reference Figure 3 and Figure 4 , the small theater smart machine, that is, the main device (Sugar Cube Theater Box): the core processing unit, responsible for plot generation and interactive control.

[0084] Far-infrared sensing flashlight: reads character card information and illuminates the main device to trigger interaction.

[0085] Character Card: Insert the flashlight, select the character and trigger the corresponding plot.

[0086] Sensors and actuators: perceive user operations (touch, voice, etc.) and output plot content (projection, sound, etc.).

[0087] Projector: Projects the generated plot images onto the screen.

[0088] Speaker: Plays the sound effects in the plot.

[0089] Power module: Powers the entire system.

[0090] Communication module: Communicates with external devices (such as APPs) to transmit data.

[0091] Among them, the mini-theater smart device also includes a power ball, a speaker, a battery box, a side handle, a housing, a main control board of the mini-theater smart device, a display screen, and optical glass. The power ball is used to adjust the volume of the smart device. The speaker provides excellent sound quality. The battery box contains a rechargeable lithium battery. The side handle facilitates children to pick it up. The housing has a simple sugar cube shape. After receiving the character information, the main control board of the smart device makes conditional judgments through an AI model (i.e., a deep learning model), determines the user's behavior intention, and triggers the corresponding plot. The display screen provides 16-bit images. The optical glass filters out blue light, which is beneficial to protecting children's eyesight.

[0092] Among them, the NFC card sensing area is located directly above the mini-theater smart device and is used to obtain the scenario information corresponding to the card.

[0093] Among them, the far-infrared induction flashlight includes a control button, a card slot, a flashlight main control module, a character card, and an NFC chip. The control button is the switch and projection of the flashlight. The character card has an NFC chip built in. When the character card is inserted into the card slot, the flashlight main control module will read the information in the character card and project it onto the screen of the smart device by far-infrared induction technology.

[0094] Each component is connected through internal circuits and wireless communication methods and works together: The power module supplies power, the main device starts, and the communication module connects to the APP; the user inserts the character card into the flashlight, the flashlight reads the information and transmits it to the main device through far-infrared rays, triggering the generation of the plot; the main device processes the information, generates the plot content, and outputs it through the projector and the speaker; the sensor senses the user's operation and feeds it back to the main device to adjust the plot in real time. The power module supplies power, and the communication module connects to the external APP to transmit data, realizing an interactive story game.

[0095] Embodiment 2: The educational story of this application can take "Along the River During the Qingming Festival" as the historical blueprint and "playing games + finding clues + learning knowledge" as the core design main line, aiming to create a product with rich interactive experience and educational value of Song Dynasty aesthetics for children. At the level of appearance design, a simple square sugar cube shape is adopted as the mini-theater smart device to realize the human-computer interaction scenario.

[0096] Step 1: Open the "Sugar Cube" theater box. Specifically, the execution subject: the user (child). Technical details: When the user opens the "Sugar Cube" theater box, the device automatically starts AR, AI, CV, and ASR technologies, uses a far-infrared sensing flashlight to illuminate the theater box, and displays different scenes from the Qingming Shanghe Tu on the screen.

[0097] Step 2: Insert and interact with character cards. Specifically, the executing subject: the user (child). Technical details: Hardware platform construction: Use the Arduino embedded system to build the corresponding hardware platform. Sensor and actuator connection: Install a variety of sensors on the block (small theater smart phone) and flashlight, such as infrared sensors, cameras, microphones, etc., to perceive user behavior and environmental information. Program writing: Use programming language to write programs to realize the collection and processing of sensor data and the control of actuators. Condition judgment and plot triggering: After receiving the character information, the theater box uses the AI ​​model to make conditional judgments, determine the user's behavioral intentions, and trigger the corresponding plot.

[0098] Step 3: AI plot generation. Specifically, the execution subject: deep learning model, such as AI model. Technical details: Dataset preparation: collect a large amount of plot content, including relevant stories, plots, character dialogues, etc. Preprocess the data, including noise removal, standardization, word segmentation, part-of-speech tagging, semantic parsing and other steps, and finally convert the text data into word vectors or serialized forms. Model training: Select recurrent neural network (RNN), especially long short-term memory network (LSTM) or gated recurrent unit (GRU) to process time series data and long-distance dependencies. Train the model through forward propagation and backpropagation algorithms so that it can generate reasonable plot content based on the input data. Plot generation: During the game, the AI ​​model will generate new plot content (i.e., plot text) based on the current game scenario and the user's historical choices. Specifically, the model will generate corresponding dialogues and events based on the input context information, and display them to the user through multimedia interactive projection technology. Optimization and adjustment: Evaluate and optimize the generated plot through machine learning algorithms. For example, indicators such as structural similarity index and peak signal-to-noise ratio are used to evaluate the quality of the plot screen, and adjustments are made based on user feedback to improve the coherence and interest of the plot.

[0099] Step 4: Use with the APP of the external terminal. Specifically, the theater box can be used with the APP of the external terminal to adjust the level setting and storyline according to the children's understanding ability, adding to the fun of children's exploration. Specifically, the APP can provide personalized level design and storyline content according to the children's age, interests and learning progress, making the game more suitable for children's needs.

[0100] This application integrates four new media perception technologies, namely augmented reality, artificial intelligence, computer vision, and speech recognition, and combines multimedia interactive projection technology. By shining a far-infrared induction flashlight on the "sugar cube", the linked picture is triggered to promote the development of the story, achieving a multi-modal human-computer interaction effect and creating an immersive aesthetic education experience environment of the Song Dynasty for children. Through multi-modal human-computer interaction and artificial intelligence plot generation technology, the story plot is enriched, making each game experience unique. At the same time, with the help of the exploration-style game and voice interaction mode, the child's sense of participation is enhanced; the game plot customization is supported, allowing the child to create a unique exclusive story during the play process to meet the personalized creation needs.

[0101] In an exemplary embodiment, a computer device is further provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the above method is implemented.

[0102] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the above method is implemented.

[0103] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the above method is implemented.

[0104] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.

[0105] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRdM), magnetoresistive random access memory (MRdM), ferroelectric random access memory (FRdM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0106] The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logics, data processing logics based on quantum computing, etc., without limitation.

[0107] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0108] Specific examples are used in this article to elaborate on the principles and implementation manners of the present application. The description of the above embodiments is only used to help understand the method and its core idea of the present application; at the same time, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present application.

Claims

1. A plot generation method for a small theater intelligent machine, characterized in that: The plot generation method of the small theater intelligent machine includes: Performing word segmentation processing on the current plot text to generate multiple words; the plot text includes educational stories or game plots, plot categories, and character dialogues corresponding to the plot categories; The plurality of words are transformed using a pre-trained language model of Transformer to generate a word vector corresponding to each word; The word vector corresponding to each word is used as the feature vector corresponding to the plot text at the current moment; The feature vector corresponding to the plot text at the current moment and the situation information at the current moment are input into the trained deep learning model to determine the plot text at the next moment; the situation information at the current moment includes the historical choices of the user at the current moment, the emotional state of the character at the current moment, and the real-time operation instructions of the user at the current moment; the real-time operation instructions include voice instructions and touch instructions.

2. The plot generation method of the small theater intelligent machine according to claim 1 is characterized in that: The feature vector corresponding to the plot text at the current moment and the situation information at the current moment are input into the trained deep learning model. Before determining the plot text at the next moment, the following steps are also included: Build deep learning models; The training process of the deep learning model specifically includes: Generate the historical plot text at the starting moment based on the historical user's historical choices, the emotional state of the historical role, and the real-time operation instructions of the historical user obtained at the starting moment as historical scenario information; Using the historical plot text at the starting moment as the plot text at the current moment; Inputting the feature vector and historical scenario information corresponding to the plot text at the current moment into the deep learning model to output the plot text at the next moment; With the goal of minimizing the difference between the plot category of the historical plot text at the next moment and the real plot category corresponding to the real-time operation instruction of the historical user, the deep learning model is trained to obtain a trained deep learning model.

3. The plot generation method of the small theater intelligent machine according to claim 2 is characterized in that: The deep learning model is trained with the goal of minimizing the difference between the plot category of the historical plot text at the next moment and the real plot category corresponding to the real-time operation instruction of the historical user. After obtaining the trained deep learning model, the method further includes: Obtaining a plot screen generated by the plot text at the next moment; Optimizing the plot picture according to the structural similarity index or the peak signal-to-noise ratio to determine an optimized plot picture; The optimized plot picture is used as the final plot picture.

4. The plot generation method of the small theater intelligent machine according to claim 3 is characterized in that: Optimizing the plot picture according to the structural similarity index or the peak signal-to-noise ratio to determine the optimized plot picture specifically includes: Determine the structural similarity index according to the mean, standard deviation and covariance of the plot picture and the mean, standard deviation and covariance of the original plot picture structure, and maximize the structural similarity index to determine the optimized plot picture; or, The peak signal-to-noise ratio is determined according to the maximum preset pixel value of the plot picture and the mean variance between the plot picture and the original plot picture structure, and the peak signal-to-noise ratio is maximized to determine the optimized plot picture.

5. The plot generation method of the small theater intelligent machine according to claim 1 is characterized in that: Before the current plot text is segmented to generate multiple words, it also includes: De-noise and normalize the plot text at the current moment.

6. A plot generation device for a small theater smart machine, characterized in that: include: Small theater smart phone, far infrared sensor flashlight, projector and power module; The far-infrared induction flashlight is provided with a card slot; the card slot is used to insert a character NFC card, and the character NFC card is used to store character information; the character information includes character dialogues, scenes corresponding to the character dialogues, and character information; The far-infrared induction flashlight is used to read the character NFC card to trigger the corresponding plot, and illuminate the character on the screen of the small theater smart phone through far-infrared rays to trigger the plot text corresponding to the character; The small theater smart machine includes a sensor and a speaker, wherein the sensor is used to sense the user's operation instructions; the speaker is used to play the sound effects in the plot; The small theater smart machine is used to execute the plot generation method of the small theater smart machine described in any one of claims 1 to 5; The projector is used to project the plot picture corresponding to the next moment plot text generated by the small theater smart machine onto the screen; The power module is used to supply power to the small theater smart phone, the far infrared sensor flashlight and the projector.

7. The plot generation device for a small theater smart machine according to claim 6, characterized in that: The plot generation device of the small theater smart machine also includes a communication module; The communication module is arranged in the small theater smart machine; The communication module is used for wirelessly transmitting data with an external terminal; the external terminal is used for users to send real-time operation instructions.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the plot generation method of the small theater intelligent machine described in any one of claims 1-5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the plot generation method of the small theater intelligent machine described in any one of claims 1-5 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the plot generation method of the small theater intelligent machine described in any one of claims 1-5 is implemented.

Images