Book design optimization method and system based on artificial intelligence

Abstract

The invention discloses a book design optimization method and system based on artificial intelligence. The method comprises the steps of data collection and processing, concept book intelligent design, entity book design enhancement, electronic function interaction optimization and book design optimization. The invention relates to the technical field of intelligent design of books, in particular to a book design optimization method and system based on artificial intelligence, and adopts a detailed intelligent and automatic book design optimization method from intelligent design of a concept book to two levels of a physical book and an electronic book. Furthermore, bidirectional optimization of the entity book and the electronic book is carried out according to the content of the concept book, so that the reference performance of the intelligent book design is improved; carrying out intelligent design of a concept book by adopting an integrated transformer model combined with information clustering optimization; a semantic feature optimization time sequence is adopted to generate an adversarial hybrid integrated network, and entity book design enhancement is carried out; and carrying out electronic function interaction optimization by adopting an augmented reality method combined with a convolutional graph neural network.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent book design, and specifically refers to a method and system for optimizing book design based on artificial intelligence. Background Art

[0002] A method and system for optimizing book design based on artificial intelligence is a system that uses advanced artificial intelligence technologies to intelligently optimize various design aspects of books. Through technologies such as deep learning, machine learning, and natural language processing, the system can analyze and generate optimal solutions for book cover design, content structure, page layout, and interactive functions. Its main functions are to enhance the visual effects of books, improve the reading experience, and provide personalized design and function optimization according to the needs and preferences of different readers, ultimately achieving efficient, intelligent, and precise book design.

[0003] However, in existing book design optimization methods, there are technical problems as follows: traditional book design methods often mainly focus on the content of book design. At the same time, current intelligent book design means mainly tend to the separate design of physical books and e-books. However, in practical applications, there are parts of the design and practical use of physical books and e-books that are interconnected and can be integrated. Existing methods lack the combined design of e-books and physical books from both the form and content perspectives, resulting in poor reference and practicality of intelligent book design methods; in existing intelligent design methods for concept books, there are technical problems as follows: existing methods similar to concept book design basically only focus on the extraction of abstract information from the content of books and perform technical tasks similar to abstract extraction or keyword extraction. However, in fact, the structure of concept books not only involves the general text content of books, but also should design the form design of multiple aspects such as the table of contents structure, chapter structure, and page layout of books. The existing book abstraction process lacks these elements; in existing physical book design methods, there are technical problems as follows: physical books not only involve complex semantic understanding content, but also involve entity book indicators and specifications related to pure data. Therefore, the design optimization of physical books needs to be considered and optimized from the perspective of multiple types of data, and existing technologies lack the optimization related to entity book specifications and indicators; in existing e-book design optimization methods, there are technical problems as follows: existing e-books have poor adaptability to users and support for users' reading habits when using e-books, which in turn leads to difficulty in improving the reading experience of e-books. This part involves both the content of the book itself and the form presentation of e-books. Summary of the Invention

[0004] In view of the above situation, to overcome the defects of the prior art, the technical solution adopted by the present invention is as follows: The method for optimizing book design based on artificial intelligence provided by the present invention includes the following steps:

[0005] Step S1: Data collection and processing;

[0006] Step S2: Intelligent design of concept books;

[0007] Step S3: Enhancement of physical book design;

[0008] Step S4: Optimization of electronic function interaction;

[0009] Step S5: Optimization of book design.

[0010] Further, in step S1, the data collection and processing is used to collect the original data required for book design optimization and perform basic preprocessing. Specifically, the original dataset for book design optimization is obtained through the collection of book content and e-book platform data, and the dataset for book design optimization is obtained by performing data preprocessing on the original dataset for book design optimization;

[0011] The original dataset for book design optimization specifically includes book content data, reader behavior data, book preference research data, and e-book interaction feedback data;

[0012] The dataset for book design optimization specifically includes book text optimization data, user requirement optimization data, book preference optimization data, and user feedback optimization data.

[0013] Further, in step S2, the intelligent design of concept books is used to design the conceptual structure of the book. Specifically, based on the original dataset for book design optimization, an integrated transformer model combined with information clustering optimization is adopted to perform intelligent design of concept books, and the conceptual book ontology data is obtained. The specific steps are as follows:

[0014] Step S21: Clustering information optimization, specifically, clustering information optimization is performed through semantic similarity calculation, weighted word frequency measurement, and dynamic clustering weight adjustment to obtain the intelligent information cluster of concept books, including the following steps:

[0015] Step S211: Clustering initialization, specifically, the book text optimization data and user requirement optimization data in the original dataset for book design optimization are used as the clustering original data input, and an initial weight is assigned to each vocabulary through word frequency calculation;

[0016] Step S212: Semantic similarity calculation, specifically, semantic similarity calculation is performed based on a standard semantic representation model to obtain semantic similarity data. The calculation formula is:

[0017]

[0018] Wherein, Sem Sim (·) is a semantic similarity calculation function, and W i is the first vocabulary index, and W j is the second vocabulary index. The vocabulary index is used to represent each word in the clustering original data input. is the semantic embedding vector corresponding to the first vocabulary, is the semantic embedding vector corresponding to the second vocabulary;

[0019] Step S213: Weighted word frequency metric. Specifically, by introducing a weighting coefficient to improve the word frequency metric coefficient, and through word frequency metric, the word frequency mutual information is calculated. The calculation formula is:

[0020]

[0021] Wherein, NP Weight (·) is a weighted calculation function of word frequency mutual information, and W i is the first vocabulary index, and W j is the second vocabulary index, w(·) is the weighting coefficient, and the initial value is taken as the initial weight of each vocabulary. Sem Sim (·) is a semantic similarity calculation function, and p(·) is a joint representation function of word occurrence probability;

[0022] Step S214: Dynamic clustering weight adjustment. Specifically, by dynamically adjusting the weighting coefficient in the weighted word frequency metric, introducing a dynamic weight adjustment mechanism, and performing clustering weighted dynamic update to obtain the dynamic clustering cluster weight. The calculation formula is:

[0023] w n = a·w 0 +(1 - a)·metric(C);

[0024] Wherein, w n is the dynamic clustering cluster weight, a is the learning rate parameter, w 0 is the weighting coefficient before dynamic adjustment, metric(·) is the clustering quality index, specifically referring to the average similarity within the cluster, and C is the clustering cluster representation index;

[0025] Step S215: Iterative information clustering. Specifically, through the clustering initialization, the semantic similarity calculation, the weighted word frequency metric, and the dynamic clustering weight adjustment, and setting iterative conditions, iterative information clustering is performed to obtain the concept book intelligent information cluster. The concept book intelligent information cluster is used as the data input of the integrated transformer model;

[0026] Step S22: Build an integrated transformer model, specifically by building a pre-trained variant semantic representation model, and through integrating and fusing the pre-trained variant semantic representation model, extracting text feature information, and performing feature integration to obtain the integrated transformer output text features;

[0027] The pre-trained variant semantic representation model specifically includes a sentence semantic representation model, a medical semantic representation model, and a cross-domain semantic representation model;

[0028] Step S23: Concept book generation, specifically generating a book concept structure, a chapter structure, and a content summary based on the integrated transformer output text features, and through fusing the book concept structure, the chapter structure, and the content summary, constructing and outputting a concept book ontology structure;

[0029] Step S24: Training of the concept book intelligent design model, specifically training the concept book intelligent design model through the clustering information optimization, the construction of the integrated transformer model, and the concept book generation to obtain the concept book intelligent design model Model DG ;

[0030] Step S25: Concept book intelligent design, specifically using the concept book intelligent design model Model DG based on the book design optimization dataset to perform concept book intelligent design and obtain concept book ontology data;

[0031] The concept book ontology data specifically includes a summary of the core content of the book, the book chapter structure, the type of target audience of the book, the book keywords, and the book content structure.

[0032] Furthermore, in step S3, the entity book design enhancement is used to intelligently design the specifications of the entity book. Specifically, based on the book design optimization dataset and the concept book ontology data, a semantic feature optimization time series generation adversarial hybrid integration network is adopted to perform entity book design enhancement to obtain entity book specification layout optimization data, which specifically includes the following steps:

[0033] Step S31: Build a multi-layer perceptron semantic recurrent subnet, specifically by building two multi-layer perceptrons, which are respectively used to receive the book preference optimization data in the book design optimization dataset and the concept book ontology data, and through introducing an alternating update rule mechanism, performing odd and even time step alternating feature extraction, and on the basis of the standard long short-term memory neural network, adding an update gate for feature integration to build a multi-layer perceptron semantic recurrent subnet to obtain the original features of entity book design enhancement;

[0034] Step S32: Construct a generative adversarial subnet. Specifically, by constructing a generator and a discriminator model, a standard generative adversarial network is built as the generative adversarial subnet, and a design scheme for the entity book is generated by enhancing the original features according to the entity book, obtaining a reference output for the entity book design;

[0035] Step S33: Joint loss optimization. Specifically, by introducing a contrast loss, a word-level training loss, and a target label loss as the joint loss function of the entity book design enhancement model, and based on the joint loss function, the model training is optimized;

[0036] Step S34: Training of the entity book design enhancement model. Specifically, through the construction of the multi-layer perceptron semantic recurrent subnet, the construction of the generative adversarial subnet, and the joint loss optimization, the entity book design enhancement model is trained to obtain the entity book design enhancement model Model RG ;

[0037] Step S35: Enhancement of the entity book design. Specifically, based on the book design optimization dataset and the conceptual book ontology data, the entity book design enhancement model Model RG is used to enhance the entity book design, obtaining optimized data for the entity book specification layout;

[0038] The optimized data for the entity book specification layout specifically includes reference for cover design, reference for book size and thickness, reference for book typesetting design, reference for book paper selection, and reference for book printing method.

[0039] Furthermore, in step S4, the electronic function interaction optimization is used to adjust the presentation mode of the e-book by combining augmented reality technology. Specifically, based on the book design optimization dataset and the conceptual book ontology data, an augmented reality method combined with a convolutional graph neural network is adopted to optimize the electronic function interaction, obtaining optimized data for the e-book layout adjustment, which specifically includes the following steps:

[0040] Step S41: Training of the convolutional graph neural network. Specifically, by constructing a convolutional graph neural network including an input layer, a convolutional reference feature layer, and a graph neural network layer, the structured information of the e-book is processed, and by combining the feature outputs of the convolutional reference feature layer and the graph neural network layer, the overall optimization of the e-book page layout is carried out to obtain a reference output for the e-book page adjustment;

[0041] Step S42: Integration of augmented reality functions. Specifically, by introducing augmented reality technology and combining the reference output for the e-book page adjustment, user operation support, dynamic text display, and chapter switching of the e-book are carried out;

[0042] Step S43: Implement personalized settings. Specifically, adjust the reference output according to the e-book page, combine the augmented reality function, conduct the user's e-book reading operation preferences, and perform dynamic adjustments to the page layout, font, and text to obtain reference data for the personalized e-book interaction solution;

[0043] Step S44: Optimize the e-book layout adjustment. Specifically, through the training of the convolutional graph neural network, the integration of the augmented reality function, and the implementation of the personalized settings, perform the e-book layout adjustment optimization to obtain the e-book layout adjustment optimization data;

[0044] The e-book layout adjustment optimization data specifically includes reference data for the interactive function design, reference data for the e-book page layout adjustment, reference data for the e-book page font adjustment, and reference data for the user's personalized settings adaptive adjustment.

[0045] Furthermore, in step S5, the book design optimization is used to comprehensively optimize the book design by combining the concept book design, physical design, and electronic function design. Specifically, based on the concept book ontology data, design the overall content and form of the book, and by combining the physical book specification layout optimization data, generate an intelligent physical book design scheme, and by combining the e-book layout adjustment optimization data, generate an intelligent e-book design scheme to obtain the intelligent book comprehensive design optimization reference scheme data;

[0046] The intelligent book comprehensive design optimization reference scheme data specifically includes an overview of the overall design optimization of the data, reference data for personalized book design recommendations, and reference data for book market analysis.

[0047] The book design optimization system based on artificial intelligence provided by the present invention includes a data collection and processing module, a concept book intelligent design module, a physical book design enhancement module, an electronic function interaction optimization module, and a book design optimization module;

[0048] The data collection and processing module is used for data collection and processing. Through data collection and processing, a book design optimization data set is obtained, and the book design optimization data set is sent to the concept book intelligent design module, the physical book design enhancement module, and the electronic function interaction optimization module;

[0049] The concept book intelligent design module is used for concept book intelligent design. Through concept book intelligent design, concept book ontology data is obtained, and the concept book ontology data is sent to the physical book design enhancement module, the electronic function interaction optimization module, and the book design optimization module;

[0050] The entity book design enhancement module is used for enhancing the entity book design. Through the enhancement of the entity book design, the optimized data of the entity book specification layout is obtained, and the optimized data of the entity book specification layout is sent to the electronic function interaction optimization module and the book design optimization module;

[0051] The electronic function interaction optimization module is used for optimizing the electronic function interaction. Through the optimization of the electronic function interaction, the optimized data of the e-book layout adjustment is obtained, and the optimized data of the e-book layout adjustment is sent to the book design optimization module;

[0052] The book design optimization module is used for optimizing the book design. Through the optimization of the book design, the optimized reference scheme data of the intelligent book comprehensive design is obtained.

[0053] The beneficial effects achieved by the present invention using the above solution are as follows:

[0054] (1) In view of the existing book design optimization methods, the traditional book design methods often mainly focus on the content of book design. At the same time, the current intelligent book design means mainly tend to the compartmentalized design of physical books and e-books. However, in fact, at the practical level, the design and application of physical books and e-books have some common and integrated parts. The existing methods lack the combined design of e-books and physical books from both the form and content perspectives, resulting in poor reference and practicality of intelligent book design methods. This solution creatively adopts a detailed intelligent and automated book design optimization method from the intelligent design of concept books to two levels of physical books and e-books. Through the intelligent generation of concept books and further two-way optimization of physical books and e-books based on the content of concept books, the referenceability of intelligent book design is improved. At the same time, the design of the data structure of the concept book itself also has better versatility and extensibility, providing a practical exploration experience for the intelligent thinking of book design;

[0055] (2) In the existing intelligent design methods for concept books, there are methods that are similar to concept book design, which mainly focus on extracting abstract information from the content of the book and perform technical tasks similar to abstract extraction or keyword extraction. However, in fact, the structure of a concept book not only involves the general text content of the book, but also should design the formal aspects such as the table of contents structure, chapter structure, and page layout of the book. The existing book abstraction process lacks these elements. This solution creatively uses an integrated transformer model combined with information clustering optimization for intelligent concept book design. By improving the way of information clustering, while optimizing the feature representation between chapter structures, it can also effectively extract the formal features and content features of the book. Then, by integrating three variant semantic representation models, it effectively provides the extraction of book content information for multi-domain and multi-type texts, thus improving the data quality of concept books and providing strong data guarantee and support for subsequent intelligent steps;

[0056] (3) In the existing physical book design methods, there is a problem that physical books not only involve complex semantic understanding content, but also involve data such as physical book indicators and specifications related to pure data. Therefore, the design optimization of physical books needs to be considered and optimized from the perspective of multi-type data integration, while the existing technology lacks the optimization related to physical book specifications and indicators. This solution creatively uses a semantic feature optimized time series generative adversarial hybrid integrated network for enhancing physical book design. Through semantic feature fusion optimization, it extracts the content and formal features of physical books, thus realizing the optimization and design of physical books in multiple aspects such as content, typesetting, layout, and book size, and improving the depth and intelligence level of book design;

[0057] (4) In the existing e-book design optimization methods, there is a problem that existing e-books have poor adaptability to users and lack support for users' reading habits when using e-books, which leads to difficulty in improving the reading experience of e-books. This part involves both the content of the book itself and the formal presentation of e-books. This solution creatively uses an augmented reality method combined with a convolutional graph neural network for optimizing electronic function interaction. By combining the structured book design data of the concept book ontology data part and the personal data of readers for parallel analysis and prediction, and presenting it relying on augmented reality technology, it improves the design accuracy of e-books in the design stage and also realizes the parallel design optimization of physical books and e-books. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a schematic flowchart of the book design optimization method based on artificial intelligence provided by the present invention;

[0059] Figure 2Schematic diagram of the book design optimization system based on artificial intelligence provided by the present invention;

[0060] Figure 3 Schematic flow diagram of the intelligent design of the concept book in step S2;

[0061] Figure 4 Schematic flow diagram of the enhancement of the physical book design in step S3;

[0062] Figure 5 Schematic flow diagram of the optimization of the electronic function interaction in step S4;

[0063] Figure 6 Schematic flow diagram of the optimization of the clustering information in step S21;

[0064] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention. Detailed implementation manners

[0065] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0066] In the description of the present invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc. indicating the orientation or positional relationship are based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation to the present invention.

[0067] Embodiment 1, referring to Figure 1 , the book design optimization method based on artificial intelligence provided by the present invention includes the following steps:

[0068] Step S1: Data collection and processing;

[0069] Step S2: Intelligent design of the concept book;

[0070] Step S3: Enhancement of the physical book design;

[0071] Step S4: Optimization of the electronic function interaction;

[0072] Step S5: Optimization of the book design.

[0073] By performing the above operations, in the existing book design optimization methods, the traditional book design methods often mainly focus on the content of book design. At the same time, the current intelligent book design means mainly tend to the compartmentalized design of physical books and e-books. However, in practical terms, there are some parts that are interconnected and can be integrated in the design and practical use of physical books and e-books. The existing methods lack the combination design of e-books and physical books from the two perspectives of form and content, resulting in the technical problem that the referenceability and practicality of intelligent book design methods are relatively poor. This solution creatively adopts a detailed intelligent and automated book design optimization method from conceptual book intelligent design to the two levels of physical books and e-books. Through the intelligent generation of conceptual books and further two-way optimization of physical books and e-books based on the content of conceptual books, the referenceability of intelligent book design is improved. At the same time, the design of the data structure of the conceptual book ontology also has better versatility and extensibility, providing a practical exploration experience for the intelligent thinking of book design.

[0074] Example 2, refer to Figure 1 and Figure 2 In step S1, the data collection and processing is used to collect the original data required for book design optimization and perform basic preprocessing. Specifically, through the collection of book content and e-book platform data, the original data set for book design optimization is obtained, and through the data preprocessing of the original data set for book design optimization, the data set for book design optimization is obtained;

[0075] The original data set for book design optimization specifically includes book content data, reader behavior data, book preference research data, and e-book interaction feedback data;

[0076] The book content data specifically includes the full text data of the book body in the book library, book chapter division data, book keyword data, and book content summary data;

[0077] The reader behavior data specifically includes e-book platform user behavior data, reading duration data, most frequently browsed chapter data, search record data, user marking data, and user page turning speed reference data;

[0078] The book preference research data specifically includes user preference data for book types, cover style data, physical book size requirements data, and e-book specification data;

[0079] The e-book interaction feedback data specifically includes user evaluations and feedback data on e-book interaction functions;

[0080] The data preprocessing specifically includes data cleaning, data labeling, data standardization and initial feature extraction, and obtains a book design optimization data set by performing data preprocessing operations on the original data set of book design optimization;

[0081] The book design optimization data set specifically includes book text optimization data, user demand optimization data, book preference optimization data and user feedback optimization data.

[0082] Example 3, see Figure 1 , Figure 2 , Figure 3 and Figure 6 This embodiment is based on the above embodiment. In step S2, the concept book intelligent design is used to design the concept structure of the book. Specifically, the original data set is optimized according to the book design, and the integrated transformer model combined with information clustering optimization is used to perform concept book intelligent design to obtain concept book ontology data. Specifically, the following steps are included:

[0083] Step S21: clustering information optimization, specifically, optimizing clustering information through semantic similarity calculation, weighted word frequency measurement and dynamic clustering weight adjustment to obtain concept book intelligent information clusters, including the following steps:

[0084] Step S211: clustering initialization, specifically, inputting the book text optimization data and user demand optimization data in the book design optimization original data set as clustering original data, and assigning an initial weight to each word through word frequency calculation;

[0085] Step S212: semantic similarity calculation, specifically, semantic similarity calculation is performed based on a standard semantic representation model to obtain semantic similarity data, and the calculation formula is:

[0086]

[0087] In the formula, Sem Sim (·) is the semantic similarity calculation function, W i is the first vocabulary index, W j is a second vocabulary index, the vocabulary index is used to represent each word in the clustering original data input, is the semantic embedding vector corresponding to the first word, is the semantic embedding vector corresponding to the second word;

[0088] Step S213: weighted word frequency measurement, specifically, improving the word frequency measurement coefficient by introducing a weighted coefficient, and calculating the word frequency mutual information through the word frequency measurement, and the calculation formula is:

[0089]

[0090] wherein, NP Weight (·) is a weighted calculation function of word frequency mutual information, W i is the first vocabulary index, W j is the second vocabulary index, w(·) is the weighting coefficient, and the initial value is taken as the initial weight of each vocabulary, Sem Sim (·) is a semantic similarity calculation function, and p(·) is a joint representation function of word occurrence probability;

[0091] Step S214: Dynamic clustering weight adjustment, specifically by dynamically adjusting the weighting coefficient in the weighted word frequency metric, introducing a dynamic weight adjustment mechanism, and performing dynamic update of clustering weights to obtain the dynamic clustering cluster weight. The calculation formula is:

[0092] w n = a·w 0 +(1 - a)·metric(C);

[0093] wherein, w n is the dynamic clustering cluster weight, a is the learning rate parameter, w 0 is the weighting coefficient before dynamic adjustment, metric(·) is the clustering quality index, specifically referring to the average similarity within the cluster, and C is the clustering cluster representation index;

[0094] Step S215: Iterative information clustering, specifically by performing the clustering initialization, the semantic similarity calculation, the weighted word frequency metric, and the dynamic clustering weight adjustment, and setting iterative conditions to perform iterative information clustering to obtain the concept book intelligent information cluster, which is used as the data input of the integrated transformer model;

[0095] Step S22: Construct an integrated transformer model, specifically by constructing a pre-trained variant semantic representation model, and integrating and fusing the pre-trained variant semantic representation model to extract text feature information and perform feature integration to obtain the integrated transformer output text feature;

[0096] The pre-trained variant semantic representation model specifically includes a sentence semantic representation model, a medical semantic representation model, and a cross-domain semantic representation model;

[0097] The sentence semantic representation model specifically refers to the pre-trained S-BERT model;

[0098] The medical semantic representation model specifically refers to the pre-trained PubMed-BERT model;

[0099] The cross-domain semantic representation model specifically refers to the pre-trained Blue-BERT model;

[0100] The calculation formula for the feature integration is as follows:

[0101] C S = Concat(f S , f PM , f B );

[0102] In the formula, C S is the output text feature of the integrated transformer, Concat(·) is the feature splicing function, f S is the feature output of the sentence semantic representation model, f PM is the feature output of the medical semantic representation model, f B is the feature output of the cross-domain semantic representation model;

[0103] Step S23: Concept book generation, specifically, based on the output text feature of the integrated transformer, generate the book concept structure, chapter structure, and content summary, and construct the output of the concept book ontology structure by fusing the book concept structure, chapter structure, and content summary;

[0104] Step S24: Training of the concept book intelligent design model, specifically, through the clustering information optimization, the construction of the integrated transformer model, and the concept book generation, perform the training of the concept book intelligent design model to obtain the concept book intelligent design model Model DG ;

[0105] Step S25: Intelligent design of the concept book, specifically, based on the book design optimization dataset, use the concept book intelligent design model Model DG , perform the intelligent design of the concept book to obtain the concept book ontology data;

[0106] The concept book ontology data specifically includes the summary of the core content of the book, the chapter structure of the book, the type of the target audience of the book, the keywords of the book, and the content structure of the book.

[0107] By performing the above operations, in the existing intelligent design method of concept books, there are existing methods similar to the design of concept books, which basically only focus on the extraction of abstract information of book content and perform technical tasks similar to abstract extraction or keyword extraction. However, in fact, the structure of a concept book not only involves the general text content of the book, but also the formal design of multiple aspects such as the table of contents structure, chapter structure, and page layout of the book. The existing book abstraction process lacks the elements of this part. This solution creatively uses an integrated transformer model combined with information clustering optimization to perform intelligent design of concept books. By improving the way of information clustering, while optimizing the feature representation between chapter structures, it can also effectively extract the formal features and content features of books. Then, through the integration of three variant semantic representation models, it effectively provides the extraction of book content information for multi-domain and multi-type texts, thus improving the data quality of concept books and providing strong data guarantee and support for subsequent intelligent steps.

[0108] Example 4, refer to Figure 1 、 Figure 2 and Figure 4 This embodiment is based on the above embodiment. In step S3, the entity book design enhancement is used to perform intelligent design on the specifications of the entity book. Specifically, based on the book design optimization dataset and the concept book ontology data, a semantic feature optimization time series generation adversarial hybrid integrated network is used to perform entity book design enhancement to obtain entity book specification layout optimization data, which specifically includes the following steps:

[0109] Step S31: Construct a multi-layer perceptron semantic recurrent subnet. Specifically, by constructing two multi-layer perceptrons, which are respectively used to receive the book preference optimization data in the book design optimization dataset and the concept book ontology data, and by introducing an alternating update rule mechanism, perform odd and even time step alternating feature extraction, and on the basis of the standard long short-term memory neural network, add an update gate for feature integration to construct a multi-layer perceptron semantic recurrent subnet to obtain the original features of entity book design enhancement. The calculation formula is:

[0110]

[0111] In the formula, h 0 is the initial semantic long short-term memory subnet hidden state, MLP 1 (·) is the first multi-layer perceptron representation function, which is used to receive the book preference optimization data in the book design optimization dataset, z 1 is the book preference optimization data, c 0 is the initial semantic long short-term memory subnet cell state, MLP 2 (·) is the second multi-layer perceptron representation function, which is used to receive the concept book ontology data, z2 is the ontology data of the concept book;

[0112] v t is the hidden state of the concept book, which is used to represent the comprehensive features after semantic fusion of the book preference optimization data and the data features of the ontology data of the concept book in the book design optimization dataset. sig(·) is the S-shaped activation function, Q is the first alternating update rule parameter, h t-1 is the hidden state of the entity book design at 1 time unit before the current moment, v t-2 is the hidden state of the concept book at 2 time units before the current moment, R is the second alternating update rule, v t-1 is the hidden state of the concept book at 1 time unit before the current moment, h t-2 is the hidden state of the entity book design at 2 time units before the current moment, ⊙ is the dot product operation;

[0113] f t is the output of the forget gate, W f is the weight of the forget gate, b f is the bias term of the forget gate, u t is the output of the update gate, W u is the weight of the update gate, b u is the bias term of the update gate;

[0114] h t is the enhanced original feature of the entity book design, which is used to represent the final output of the multi-layer perceptron semantic recurrent subnet, o t is the output of the output gate of the multi-layer perceptron semantic recurrent subnet, tanh(·) is the hyperbolic tangent function, c t-1 is the candidate cell state at 1 time unit before the current moment, i t is the output of the input gate of the multi-layer perceptron semantic recurrent subnet;

[0115] Step S32: Construct a generative adversarial subnet, specifically by constructing a generator and a discriminator model to perform the construction of a standard generative adversarial network as the generative adversarial subnet, and generating an entity book design scheme based on the enhanced original feature of the entity book design to obtain a reference output of the entity book design;

[0116] Step S33: Joint loss optimization, specifically by introducing a contrast loss, a word-level training loss, and a target label loss as the joint loss function of the entity book design enhancement model, and optimizing the model training according to the joint loss function;

[0117] The calculation formula of the joint loss function is:

[0118] L total = a real ·L real + afake ·L fake +a word ·L word +a w ·L w ;

[0119] In the formula, L total is the combined loss function, a real is the real data loss weight in the contrastive loss function, L real is the real data loss function in the contrastive loss function, a fake is the generated data loss weight in the contrastive loss function, L fake is the generated data loss function in the contrastive loss function, a word is the loss weight of the word-level loss function, L word is the word-level loss function, and the word-level loss function specifically adopts the binary cross-entropy loss function, a w is the loss weight of the target label loss function, L w is the target label loss function, and the target label loss function specifically adopts the classification loss function;

[0120] Step S34: Training the entity book design enhancement model, specifically, training the entity book design enhancement model through the construction of the multi-layer perceptron semantic recurrent subnet, the construction of the generative adversarial subnet, and the combined loss optimization to obtain the entity book design enhancement model Model RG ;

[0121] Step S35: Entity book design enhancement, specifically, based on the book design optimization dataset and the concept book ontology data, using the entity book design enhancement model Model RG , perform entity book design enhancement to obtain entity book specification layout optimization data;

[0122] The entity book specification layout optimization data specifically includes cover design reference, book size and thickness reference, book layout design reference, book paper selection reference, and book printing method reference.

[0123] By performing the above operations, in the existing physical book design methods, physical books not only involve complex semantic understanding content, but also involve physical book indicators and specifications such as pure data-related ones. Therefore, the design optimization of physical books needs to be considered and optimized from the perspective of multiple types of data integration. However, the prior art lacks the technical problem of optimizing the specifications and indicators related to physical books. This solution creatively uses a semantic feature optimization time series generative adversarial hybrid integration network to enhance the physical book design. Through semantic feature fusion optimization, the content and form features of physical books are extracted, thus realizing the optimization and design of physical books in multiple aspects such as content, typesetting, layout, and book size, and improving the depth and intelligence of book design.

[0124] Example Five, refer to Figure 1 、 Figure 2 and Figure 5 In this example, based on the above example, in step S4, the electronic function interaction optimization is used to adjust the presentation mode of e-books by combining augmented reality technology. Specifically, according to the book design optimization data set and the conceptual book ontology data, an augmented reality method combined with a convolutional graph neural network is used to perform electronic function interaction optimization to obtain e-book layout adjustment optimization data, which specifically includes the following steps:

[0125] Step S41: Train a convolutional graph neural network. Specifically, by constructing a convolutional graph neural network including an input layer, a convolutional reference feature layer, and a graph neural network layer, perform structured information processing of e-books, and through combining the feature outputs of the convolutional reference feature layer and the graph neural network layer, perform overall optimization of the e-book page layout to obtain an e-book page adjustment reference output;

[0126] The convolutional reference feature layer is used to receive the user feedback optimization data in the book design optimization data set, perform page visual feature optimization, extract font features, color features, and typesetting features, and obtain a page visual feature output;

[0127] The graph neural network layer is used to receive the book content features in the conceptual book ontology data, and by representing the book content structure as a graph, representing chapter content as nodes, and representing chapter relationships as edges, construct a book interaction graph structure, and construct a standard graph neural network as the graph neural network layer to obtain an e-book content layout feature output;

[0128] The overall optimization of the e-book page layout is specifically to perform overall optimization of the e-book page layout by combining the page visual feature output and the e-book content layout feature output to obtain an e-book page adjustment reference output;

[0129] Step S42: Augmented reality function integration, specifically by introducing augmented reality technology, combining the reference output of the e-book page adjustment, providing user operation support and dynamic text display and chapter switching of the e-book;

[0130] Step S43: Implementation of personalized settings, specifically based on the reference output of the e-book page adjustment, combining the augmented reality function, obtaining the user's e-book reading operation preferences, and performing dynamic adjustments to the page layout, font, and text to obtain reference data for a personalized e-book interaction solution;

[0131] Step S44: Optimization of e-book layout adjustment, specifically through the training of the convolutional graph neural network, the augmented reality function integration, and the implementation of personalized settings, performing optimization of the e-book layout adjustment to obtain data for e-book layout adjustment optimization;

[0132] The data for e-book layout adjustment optimization specifically includes reference data for interactive function design, reference data for e-book page layout adjustment, reference data for e-book page font adjustment, and reference data for adaptive adjustment of user personalized settings.

[0133] By performing the above operations, in the existing e-book design optimization methods, there are technical problems that the existing e-books have poor adaptability to users and poor support for users' reading habits when using e-books, resulting in difficulty in improving the reading experience of e-books. This part involves both the content of the book itself and the form presentation of e-books. This solution creatively uses an augmented reality method combined with a convolutional graph neural network to optimize electronic function interaction. By combining the structured book design data of the concept book ontology data part and the personal data of readers for parallel analysis and prediction, and presenting it relying on augmented reality technology, it improves the design accuracy of e-books in the design stage and also realizes the parallel design optimization of physical books and e-books.

[0134] Example 6, refer to Figure 1 、 Figure 2 and Figure 6 In this example, based on the above example, in step S5, the book design optimization is used to comprehensively optimize the book design by combining concept book design, physical design, and electronic function design. Specifically, by relying on the concept book ontology data, the overall content and form of the book are designed, and by combining the optimized data of the physical book specification layout, a smart physical book design scheme is generated, and by combining the data for e-book layout adjustment optimization, a smart e-book design scheme is generated to obtain reference data for a comprehensive design optimization of the smart book.

[0135] The intelligent book comprehensive design optimization reference scheme data specifically includes the overall design optimization summary of the data, personalized recommendation reference data for book design, and reference data for book market analysis.

[0136] Example 7. Refer to Figure 1 and Figure 2 , based on the above embodiments, the book design optimization system based on artificial intelligence provided by the present invention includes a data collection and processing module, a concept book intelligent design module, an entity book design enhancement module, an electronic function interaction optimization module, and a book design optimization module;

[0137] The data collection and processing module is used for data collection and processing. Through data collection and processing, a book design optimization data set is obtained, and the book design optimization data set is sent to the concept book intelligent design module, the entity book design enhancement module, and the electronic function interaction optimization module;

[0138] The concept book intelligent design module is used for concept book intelligent design. Through concept book intelligent design, concept book ontology data is obtained, and the concept book ontology data is sent to the entity book design enhancement module, the electronic function interaction optimization module, and the book design optimization module;

[0139] The entity book design enhancement module is used for entity book design enhancement. Through entity book design enhancement, entity book specification layout optimization data is obtained, and the entity book specification layout optimization data is sent to the electronic function interaction optimization module and the book design optimization module;

[0140] The electronic function interaction optimization module is used for electronic function interaction optimization. Through electronic function interaction optimization, electronic book layout adjustment optimization data is obtained, and the electronic book layout adjustment optimization data is sent to the book design optimization module;

[0141] The book design optimization module is used for book design optimization. Through book design optimization, intelligent book comprehensive design optimization reference scheme data is obtained.

[0142] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprises", "comprising" or any other variation thereof is intended to cover a non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device.

[0143] Although embodiments of the present invention have been shown and described, those of ordinary skill in the art will understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention.

[0144] The above description of the present invention and its embodiments is not restrictive. What is shown in the drawings is only one of the embodiments of the present invention, and the actual structure is not limited thereto. In general, if those of ordinary skill in the art are inspired by it and, without departing from the purpose of the present invention, creatively design a structural manner and embodiments similar to the technical solution, they shall fall within the protection scope of the present invention.

Claims

1. A book design optimization method based on artificial intelligence, characterized by: The method comprises the following steps: Step S1: data collection and processing to obtain a book design optimization data set; Step S2: Intelligent design of concept book, using the integrated transformer model combined with information clustering optimization to perform intelligent design of concept book and obtain the concept book ontology data, specifically including the following steps: Step S21: Clustering information optimization; Step S22: Constructing the integrated transformer model; Step S23: Concept book generation; Step S24: Concept book intelligent design model training; Step S25: Concept book intelligent design; Step S3: Physical book design enhancement, using semantic feature optimization time-series generative adversarial hybrid integrated network to perform physical book design enhancement, and obtain physical book specification layout optimization data, specifically including the following steps: Step S31: constructing a multi-layer perceptual semantic loop subnet; Step S32: constructing a generative adversarial subnet; Step S33: joint loss optimization; Step S34: physical book design enhancement model training; Step S35: physical book design enhancement; Step S4: electronic function interaction optimization, using an augmented reality method combined with a convolutional graph neural network to optimize the electronic function interaction and obtain the electronic book layout adjustment optimization data, specifically including the following steps: step S41: training the convolutional graph neural network; step S42: augmented reality function integration; step S43: personalized setting implementation; step S44: electronic book layout adjustment optimization; Step S5: Book design optimization, obtaining intelligent book comprehensive design optimization reference solution data.

2. The book design optimization method based on artificial intelligence according to claim 1, characterized in that: In step S1, the data collection process is used to collect the original data required for book design optimization and perform basic preprocessing, specifically, to obtain a book design optimization original data set through book content and e-book platform data collection, and to obtain a book design optimization data set by performing data preprocessing on the book design optimization original data set; The book design optimization data set specifically includes book text optimization data, user demand optimization data, book preference optimization data and user feedback optimization data.

3. The book design optimization method based on artificial intelligence according to claim 2 is characterized in that: In step S2, the concept book intelligent design is used to design the concept structure of the book, specifically, based on the book design optimization of the original data set, the integrated transformer model combined with information clustering optimization is used to perform the concept book intelligent design to obtain the concept book ontology data, specifically including the following steps: Step S21: clustering information optimization, specifically, optimizing clustering information through semantic similarity calculation, weighted word frequency measurement and dynamic clustering weight adjustment to obtain concept book intelligent information clusters; Step S22: constructing an integrated transformer model, specifically constructing a pre-trained variant semantic representation model, and extracting text feature information by integrating and fusing the pre-trained variant semantic representation model, and performing feature integration to obtain integrated transformer output text features; The pre-trained variant semantic representation model specifically includes a sentence semantic representation model, a medical semantic representation model and a cross-domain semantic representation model; Step S23: generating a concept book, specifically generating a book concept structure, a chapter structure and a content summary according to the output text features of the integrated transformer, and constructing a concept book body structure output by fusing the book concept structure, chapter structure and content summary; Step S24: training the concept book intelligent design model, specifically, training the concept book intelligent design model through the clustering information optimization, the construction of the integrated transformer model and the concept book generation, to obtain the concept book intelligent design model Model DG ; Step S25: Concept book intelligent design, specifically, using the concept book intelligent design model Model according to the book design optimization data set DG , perform intelligent design of concept books and obtain concept book ontology data; The conceptual book ontology data specifically includes a summary of the core content of the book, a book chapter structure, a book target audience type, book keywords and a book content structure.

4. The book design optimization method based on artificial intelligence according to claim 3 is characterized in that: In step S21, the concept book intelligent design specifically includes the following steps: Step S211: clustering initialization, specifically, inputting the book text optimization data and user demand optimization data in the book design optimization original data set as clustering original data, and assigning an initial weight to each word through word frequency calculation; Step S212: semantic similarity calculation, specifically, semantic similarity calculation is performed based on a standard semantic representation model to obtain semantic similarity data, and the calculation formula is: In the formula, Sem Sim (·) is the semantic similarity calculation function, W i is the first vocabulary index, W j is a second vocabulary index, the vocabulary index is used to represent each word in the clustering original data input, is the semantic embedding vector corresponding to the first word, is the semantic embedding vector corresponding to the second word; Step S213: weighted word frequency measurement, specifically, improving the word frequency measurement coefficient by introducing a weighted coefficient, and calculating the word frequency mutual information through the word frequency measurement, and the calculation formula is: In the formula, NP Weight (·) is the weighted calculation function of word frequency mutual information, W i is the first vocabulary index, W j is the second vocabulary index, w(·) is the weighting coefficient, and the initial value is the initial weight of each vocabulary, Sem Sim (·) is the semantic similarity calculation function, p(·) is the word occurrence probability joint representation function; Step S214: Dynamic clustering weight adjustment, specifically, dynamically adjusting the weight coefficient in the weighted word frequency measurement, introducing a dynamic weight adjustment mechanism, dynamically updating the clustering weight, and obtaining the dynamic clustering weight. The calculation formula is: w n =a·w0+(1-a)·metric(C); In the formula, w n is the dynamic cluster weight, a is the learning rate parameter, w0 is the weight coefficient before dynamic adjustment, metric(·) is the cluster quality indicator, specifically the average similarity within the cluster, and C is the cluster representation index; Step S215: iterative information clustering, specifically, performing iterative information clustering through the cluster initialization, the semantic similarity calculation, the weighted word frequency measurement and the dynamic clustering weight adjustment, and setting iteration conditions to obtain concept book intelligent information clusters, and the concept book intelligent information clusters are used as data input for the integrated transformer model.

5. The book design optimization method based on artificial intelligence according to claim 4 is characterized in that: In step S3, the physical book design enhancement is used to intelligently design the specifications of the physical book. Specifically, based on the book design optimization data set and the conceptual book ontology data, a semantic feature optimized temporal generative adversarial hybrid integrated network is used to enhance the physical book design to obtain physical book specification layout optimization data, which specifically includes the following steps: Step S31: constructing a multi-layer perceptual semantic recurrent subnet, specifically by constructing two multi-layer perceptrons, respectively used to receive the book preference optimization data and the conceptual book ontology data in the book design optimization data set, and by introducing an alternating update rule mechanism, performing odd-even time step alternating feature extraction, and adding an update gate on the basis of a standard long short-term memory neural network for feature integration, constructing a multi-layer perceptual semantic recurrent subnet, and obtaining the original features of the physical book design enhancement; Step S32: constructing a generative adversarial subnet, specifically, constructing a standard generative adversarial network by constructing a generator and a discriminator model as the generative adversarial subnet, and generating a physical book design scheme by enhancing the original features according to the physical book design, and obtaining a physical book design reference output; Step S33: joint loss optimization, specifically, introducing contrast loss, word-level training loss and target label loss as a joint loss function of the physical book design enhancement model, and optimizing model training according to the joint loss function; Step S34: Physical book design enhancement model training, specifically, through the construction of the multi-layer perceptual semantic recurrent subnet, the construction of the generative adversarial subnet and the joint loss optimization, the physical book design enhancement model training is performed to obtain the physical book design enhancement model Model RG ; Step S35: Physical book design enhancement, specifically, using the physical book design enhancement model Model according to the book design optimization data set and the conceptual book ontology data RG , enhance the physical book design and obtain the physical book specification layout optimization data; The physical book specification layout optimization data specifically includes a cover design reference, a book size thickness reference, a book typesetting design reference, a book paper selection reference, and a book printing method reference.

6. The book design optimization method based on artificial intelligence according to claim 5, characterized in that: In step S4, the electronic function interaction optimization is used to adjust the presentation mode of the electronic book in combination with the augmented reality technology, specifically, based on the book design optimization data set and the conceptual book ontology data, an augmented reality method combined with a convolutional graph neural network is used to perform electronic function interaction optimization to obtain the electronic book layout adjustment optimization data, specifically including the following steps: Step S41: training a convolutional graph neural network, specifically, constructing a convolutional graph neural network including an input layer, a convolutional reference feature layer and a graph neural network layer to process the structured information of the electronic book, and optimizing the overall layout of the electronic book page by combining the feature outputs of the convolutional reference feature layer and the graph neural network layer to obtain a reference output for adjusting the electronic book page; Step S42: integrating augmented reality functions, specifically, by introducing augmented reality technology, combining the electronic book page adjustment reference output, performing user operation support and electronic book dynamic text display and chapter switching; Step S43: implementing personalized settings, specifically adjusting the reference output according to the e-book page, combining the augmented reality function, performing the user's e-book reading operation preferences, and dynamically adjusting the page layout, font and text to obtain personalized e-book interaction solution reference data; Step S44: electronic book layout adjustment and optimization, specifically, by performing electronic book layout adjustment and optimization through the training convolutional graph neural network, the augmented reality function integration and the personalized setting implementation, and obtaining electronic book layout adjustment and optimization data; The electronic book layout adjustment optimization data specifically includes interactive function design reference data, electronic book page layout adjustment reference data, electronic book page font adjustment reference data and user personalized setting adaptive adjustment reference data.

7. The artificial intelligence-based book design optimization method according to claim 6, characterized in that: In step S5, the book design optimization is used to combine the concept book design, the physical design and the electronic function design to perform comprehensive book design optimization, specifically, by designing the overall content and form of the book according to the concept book ontology data, and by combining the physical book specification layout optimization data, to generate an intelligent physical book design solution, and by combining the electronic book layout adjustment optimization data, to generate an intelligent electronic book design solution, to obtain intelligent book comprehensive design optimization reference solution data; The intelligent book comprehensive design optimization reference solution data specifically includes a data overall design optimization summary, book design personalized recommendation reference data and book market analysis reference data.

8. An artificial intelligence-based book design optimization system, used to implement the artificial intelligence-based book design optimization method as described in any one of claims 1 to 7, characterized in that: It includes data collection and processing module, concept book intelligent design module, physical book design enhancement module, electronic function interaction optimization module and book design optimization module.

9. The book design optimization system based on artificial intelligence according to claim 8, characterized in that: The data collection and processing module is used for data collection and processing, and obtains a book design optimization data set through data collection and processing, and sends the book design optimization data set to the concept book intelligent design module, the physical book design enhancement module and the electronic function interaction optimization module; The concept book intelligent design module is used for concept book intelligent design, obtains concept book ontology data through concept book intelligent design, and sends the concept book ontology data to the physical book design enhancement module, the electronic function interaction optimization module and the book design optimization module; The physical book design enhancement module is used for physical book design enhancement, obtains physical book specification layout optimization data through physical book design enhancement, and sends the physical book specification layout optimization data to the electronic function interaction optimization module and the book design optimization module; The electronic function interaction optimization module is used for electronic function interaction optimization, obtains electronic book layout adjustment optimization data through electronic function interaction optimization, and sends the electronic book layout adjustment optimization data to the book design optimization module; The book design optimization module is used for book design optimization, and obtains intelligent book comprehensive design optimization reference solution data through book design optimization.

Images