Visual Communication Design and Display System Based on Model Analysis

Abstract

The present invention discloses a visual communication design and display system based on model analysis, which relates to the technical field of visual communication design, and includes a design requirement analysis unit, a design output unit, a design display unit, a user feedback analysis unit, and a design optimization unit; the present invention identifies each natural language command through a large language model, determines the design requirements expressed by each natural language command, determines the corresponding design system, page image, and page template based on the design requirements, generates a trial-run user interface design draft and displays it to the user side, establishes a user portrait based on the user's basic information, obtains a user feature coefficient, and at the same time constructs a user behavior table based on behavior data, analyzes the user behavior table through a deep learning model, and then evaluates the trial-run simulation interface to obtain an evaluation result, and performs user interface responsive design optimization for the page module corresponding to the trial-run simulation interface to adapt to different user needs.

Description

Technical Field

[0001] The present invention relates to the technical field of visual communication design, and particularly to a visual communication design and display system based on model analysis. Background Art

[0002] Visual communication design is a form of design that conveys information, expresses ideas and emotions through visual elements (such as graphics, text, colors, layouts, etc.). It aims to enhance the readability, attractiveness and influence of information through visual means. The display system refers to the overall design scheme and technical system for displaying information, products, artworks or brand images. It is not only the spatial layout of physical display, but also a combination of a whole set of elements such as comprehensive design, interactive technology, visual communication, etc.

[0003] Visual communication design and display system are closely related fields, and the two work together to affect the way we obtain information, perceive the brand image and experience art and culture. With the continuous development of technology and design concepts, visual communication design and display system will increasingly rely on digital technology, interactive means and creative elements to provide a richer and more interactive experience. Among them, user interface design is an important application field of visual communication design, including the interface design of mobile applications, websites, and software, ensuring a smooth user experience and conveying correct information. User interface design is a very comprehensive task that needs to consider multiple aspects such as user needs, functional requirements, platform characteristics, visual design, interaction methods, accessibility, etc. Excellent user interface design not only needs to meet technical requirements, but also pays attention to user experience. However, user experience and user needs are multi-faceted, and it is difficult to meet the different needs of different users. Therefore, the existing user interface design is difficult to comprehensively meet the actual feedback of user experience.

[0004] In view of the above technical deficiencies, a solution is proposed. Summary of the Invention

[0005] The purpose of the present invention is to: evaluate the trial operation simulation interface based on user characteristic coefficients and behavior data to obtain an evaluation result, and perform user interface responsive design optimization on the page module corresponding to the trial operation simulation interface to adapt to different user needs.

[0006] To achieve the above purpose, the present invention adopts the following technical solution: a visual communication design and display system based on model analysis, including a design requirement analysis unit, a design output unit, a design display unit, a user feedback analysis unit and a design optimization unit;

[0007] The design requirement analysis unit includes a model recognition module and a requirement analysis module. The model recognition module is used to obtain natural language commands, identify each natural language command through a large language model, determine the design requirements expressed by each natural language command, and send them to the requirement analysis module. The large language model is trained based on the structured knowledge in the field of user interface design;

[0008] The requirement analysis module is used to obtain design requirements, determine the corresponding design system, page image, and page template based on the design requirements, construct the pseudo-code corresponding to the page template using the resources of the design system, and send the pseudo-code to the design output unit;

[0009] The design output unit is used to convert the pseudo-code into the corresponding vector design draft, determine the corresponding style and filling content based on the page image, and use the style and the filling content to fill the vector design draft to obtain the trial-run user interface design draft and then output it to the design display unit;

[0010] The design display unit is used to obtain the trial-run user interface design draft and the target user interface, generate a trial-run simulation interface based on the trial-run user interface design draft, and display the trial-run simulation interface to the user side. At the same time, generate an actual operation interface based on the target user interface and display the actual operation interface to the user side;

[0011] The user feedback analysis unit includes a user analysis module and a feedback analysis module. The user analysis module is used to obtain the basic information of the user, establish a user portrait according to the basic information, classify the user portrait according to the clustering model to obtain the user feature recognition result, and calculate the user feature coefficient based on the user feature recognition result and send it to the feedback analysis unit;

[0012] The feedback analysis module is used to obtain the behavior data of the user on the trial-run simulation interface, construct a user behavior table based on the behavior data, analyze the user behavior table through a deep learning model to obtain the user behavior analysis data, evaluate the trial-run simulation interface according to the user feature coefficient and the user behavior analysis data to obtain the evaluation result, and send it to the interface optimization unit;

[0013] The design optimization unit is used to obtain and process the evaluation result, and perform user interface responsive design optimization on the page module corresponding to the trial-run simulation interface to adapt to different user needs to obtain the target user interface and then send it to the design display unit again.

[0014] Furthermore, the specific process of obtaining the large language model is as follows:

[0015] S101. The natural language commands include user requirements and goals, functional requirements, platform and device characteristics, visual design principles, interaction methods, accessibility, brand and visual identity, and technical feasibility;

[0016] S102. Build a knowledge graph based on data from different natural language command sources. Use the depth-first search algorithm to obtain the language entities of each entity in each knowledge graph based on the search path corresponding to each entity. After cleaning and tokenizing the language entities, obtain the training data;

[0017] S103. Use the clustering algorithm to obtain the relevance of the training data for each knowledge graph and obtain the target task data corresponding to the training data;

[0018] S104. Build the base of a large language training model based on the Transformer architecture. Input the training data into the base of the large language training model for pre-training to obtain a pre-trained large language model. On the basis of the pre-trained model, use the target task data for fine-tuning to obtain an optimized large language model.

[0019] Further, the specific process of obtaining the trial-run user interface design draft is as follows:

[0020] S201. Retrieve relevant styles and filling contents from the network database based on the page image, and integrate the styles and filling contents into an alternative data set;

[0021] S202. Calculate the correlation coefficients between each style and filling content in the alternative data set and the design requirements according to the clustering algorithm, and represent them with specific numerical values. Mark the correlation coefficients at the subscripts of the styles and filling contents;

[0022] S203. Rearrange the styles and filling contents in the alternative data set from largest to smallest correlation coefficient, and fill the top several groups of styles and filling contents into the vector design draft one by one to obtain a quasi-user interface design draft;

[0023] S204. Input the quasi-user interface design draft into the target model to obtain the verification result output by the target model, and select the trial-run user interface design draft according to the verification result;

[0024] The target model is a neural network model obtained through iterative training with a number of normal user interface design drafts and a number of abnormal user interface design drafts. The abnormal user interface design draft is a user interface component that does not conform to the preset design rules.

[0025] Further, the specific process of generating the trial-run simulation interface is as follows:

[0026] S301. Obtain the interface configuration data corresponding to the user interface design draft, and generate an initial user interface based on the interface configuration data. The interface configuration data is used to represent the style, filled content, and page template of the user interface design draft, and the initial user interface is displayed on the user side;

[0027] S302. Configure the initial user interface to generate a trial operation simulation interface, and the trial operation simulation interface matches the user interface design draft.

[0028] Further, the specific process of calculating the user feature coefficient is as follows:

[0029] S401. Obtain the basic information of the user. The basic information includes the user's personal information, historical behavior data, and historical social data. The user's personal information includes the user's gender, age, and occupation. The historical behavior data includes historical browsing records, historical click records, historical search data, and historical purchase data. The historical social data includes the user's interaction behaviors on the social platform, including the number of likes, comment content, and sharing data;

[0030] S402. Add tags to the user's personal information, historical behavior data, and historical social data to obtain a user tag set. Determine a feature vector set based on the user tag set, and construct a user portrait according to the user tag set and the feature vector set;

[0031] S403. Construct a feature matrix of the user based on the user's behavior characteristics and consumption characteristics, and identify the feature types of the user portrait through a clustering model, specifically hierarchical clustering;

[0032] S404. Extract the feature keywords of the feature type, identify the portrait feature factors in the feature type, and then obtain the user feature recognition result. Calculate the user feature coefficient R(G, F) according to the following formula: , where e1 and e2 are preset proportionality coefficients, n represents the number of samples corresponding to the portrait feature factor, Gi is the i-th sample value of the corresponding parameters of other factors in the portrait feature factor, is the sample mean of the corresponding parameters of other factors in the portrait feature factor, Fi is the i-th sample value of the corresponding parameters of other factors in the portrait feature factor, is the sample mean of the corresponding parameters of other factors in the portrait feature factor. The user feature coefficient is used to represent the degree of correlation between the user and the corresponding feature type. The larger the user feature coefficient, the higher the degree of correlation between the user and the corresponding feature type. Conversely, the smaller the user feature coefficient, the lower the degree of correlation between the user and the corresponding feature type.

[0033] Further, the specific process of evaluating the trial operation simulation interface is as follows:

[0034] S501. Obtain the behavior data of the user on the trial operation simulation interface. The behavior data includes the user ID and the corresponding user characteristic coefficient, the routed page, the stay time, the number of link jumps, and the number of function uses;

[0035] S502. Construct a timeline based on the user's operation time, mark the behavior data according to the corresponding time nodes on the timeline to obtain a user behavior table, and analyze the user behavior table, including aggregation comparison and sorting, to obtain user behavior analysis data. The user behavior analysis data includes the user ID, the number of clicks on high-frequency pages, the number of clicks on high-frequency links, and the number of clicks on high-frequency functions;

[0036] S503. Obtain the user characteristic coefficient, and calculate the correlation coefficient Ui between the portrait feature factor in the user characteristic coefficient and the user behavior analysis data one by one according to the following formula: , where d is a preset weight coefficient, N is the total number of user operations, and fi is the number of clicks on high-frequency pages or the number of clicks on high-frequency links or the number of clicks on high-frequency functions;

[0037] S504. Obtain the preset correlation judgment threshold. If the correlation coefficient Ui is greater than or equal to the correlation judgment threshold, the evaluation result of the trial operation simulation interface meets the user requirements;

[0038] If the correlation coefficient Ui is less than the correlation judgment threshold, the evaluation result of the trial operation simulation interface does not meet the user requirements.

[0039] Further, the specific operations for responsive design optimization of the trial operation simulation interface are as follows:

[0040] Adjust the interface layout of the trial operation simulation interface;

[0041] Add new controls to the trial operation simulation interface;

[0042] Delete existing controls from the trial operation simulation interface;

[0043] Adjust the control attributes of the controls in the trial operation simulation interface;

[0044] Adjust the control styles of the controls in the trial operation simulation interface;

[0045] Adjust the link positions to be jumped by the controls in the trial operation simulation interface.

[0046] In summary, due to the adoption of the above technical solutions, the beneficial effects of the present invention are:

[0047] The visual communication design and display system based on model analysis identifies each natural language command through a large language model, determines the design requirements expressed by each natural language command, and determines the corresponding design system, page image, and page template based on the design requirements. After generating a trial-run user interface design draft, it is displayed to the user side. A user portrait is established based on the user's basic information, and a user feature coefficient is obtained. At the same time, a user behavior table is constructed based on behavior data, and the user behavior table is analyzed through a deep learning model. Then, the trial-run simulation interface is evaluated to obtain an evaluation result, and the page module corresponding to the trial-run simulation interface is optimized for user interface responsive design to adapt to different user needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 FIG. shows the overall system structure diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to 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. Embodiment 1

[0050] As Figure 1 shown, the visual communication design and display system based on model analysis includes a design requirement analysis unit, a design output unit, a design display unit, a user feedback analysis unit, and a design optimization unit;

[0051] The design requirement analysis unit includes a model recognition module and a requirement analysis module. The model recognition module is used to obtain natural language commands, identify each natural language command through a large language model, determine the design requirements expressed by each natural language command, and send them to the requirement analysis module. The large language model is trained based on the structured knowledge in the field of user interface design;

[0052] The specific process of obtaining the large language model is as follows:

[0053] S101. The natural language commands include user needs and goals, functional requirements, platform and device characteristics, visual design principles, interaction methods, accessibility, brand and visual identity, and technical feasibility;

[0054] S102. Construct a knowledge graph based on the data from different natural language command sources, use the depth-first search algorithm to obtain the language entities of each entity based on the search path corresponding to each entity in each knowledge graph, and obtain the training data after cleaning and word segmentation processing of the language entities;

[0055] S103. Use a clustering algorithm to obtain the relevance of the training data for each knowledge graph and obtain the target task data corresponding to the training data;

[0056] S104. Build the base of a large language training model based on the Transformer architecture, input the training data into the base of the large language training model for pre-training to obtain a pre-trained large language model, and on the basis of the pre-trained model, use the target task data for fine-tuning to obtain an optimized large language model.

[0057] The requirements analysis module is used to obtain design requirements, determine the corresponding design system, page image, and page template based on the design requirements, construct the pseudo-code corresponding to the page template using the resources of the design system, and send the pseudo-code to the design output unit;

[0058] The design output unit is used to convert the pseudo-code into a corresponding vector design draft, determine the corresponding style and filling content based on the page image, and use the style and the filling content to fill the vector design draft to obtain a trial-run user interface design draft and then output it to the design display unit;

[0059] The specific process of obtaining the trial-run user interface design draft is as follows:

[0060] S201. Retrieve relevant styles and filling content from the network database based on the page image, and integrate the styles and filling content into an alternative data set;

[0061] S202. Calculate the correlation coefficients between each style and filling content in the alternative data set and the design requirements according to the clustering algorithm, represent them with specific numerical values, and mark the correlation coefficients at the subscripts of the styles and filling content;

[0062] S203. Rearrange the styles and filling content in the alternative data set from largest to smallest correlation coefficient, and fill the top several groups of styles and filling content into the vector design draft one by one to obtain a quasi-user interface design draft;

[0063] S204. Input the quasi-user interface design draft into the target model to obtain the verification result output by the target model, and select the trial-run user interface design draft according to the verification result;

[0064] The target model is a neural network model obtained by iterative training with a number of normal user interface design drafts and a number of abnormal user interface design drafts, and the abnormal user interface design draft is a user interface component that does not conform to the preset design rules.

[0065] The design display unit is used to obtain the user interface design draft and the target user interface for trial operation, generate a trial operation simulation interface based on the user interface design draft for trial operation, and display the trial operation simulation interface to the user side. At the same time, generate an actual operation interface based on the target user interface and display the actual operation interface to the user side;

[0066] The specific process of generating the trial operation simulation interface is as follows:

[0067] S301. Obtain the interface configuration data corresponding to the user interface design draft, and generate an initial user interface based on the interface configuration data. The interface configuration data is used to represent the style, filled content and page template of the user interface design draft, and the initial user interface is displayed on the user side;

[0068] S302. Configure the initial user interface to generate a trial operation simulation interface, and the trial operation simulation interface matches the user interface design draft.

[0069] The user feedback analysis unit includes a user analysis module and a feedback analysis module. The user analysis module is used to obtain the basic information of the user, establish a user portrait according to the basic information, classify the user portrait according to the clustering model to obtain the user feature recognition result, and calculate the user feature coefficient based on the user feature recognition result and send it to the feedback analysis unit;

[0070] The specific process of calculating the user feature coefficient is as follows:

[0071] S401. Obtain the basic information of the user. The basic information includes the user's personal information, historical behavior data and historical social data. The user's personal information includes the user's gender, age and occupation. The historical behavior data includes historical browsing records, historical click records, historical search data and historical purchase data. The historical social data includes the user's interaction behaviors on the social platform, including the number of likes, comment content and sharing data;

[0072] S402. Add labels to the user's personal information, historical behavior data and historical social data to obtain a user label set, determine a feature vector set based on the user label set, and construct a user portrait according to the user label set and the feature vector set;

[0073] S403. Construct a feature matrix of the user based on the user's behavior characteristics and consumption characteristics, and identify the feature type of the user portrait through a clustering model, specifically hierarchical clustering;

[0074] S404. Extract the feature keywords of the feature type, identify the portrait feature factors in the feature type, and then obtain the user feature recognition result, and calculate the user feature coefficient R(G, F) according to the following formula: , where e1 and e2 are preset proportionality coefficients, n represents the number of samples corresponding to the portrait feature factor, Gi is the i-th sample value of the parameters corresponding to other factors in the portrait feature factor, is the sample mean of the parameters corresponding to other factors in the portrait feature factor, Fi is the i-th sample value of the parameters corresponding to other factors in the portrait feature factor, is the sample mean of the parameters corresponding to other factors in the portrait feature factor. The user feature coefficient is used to represent the degree of correlation between the user and the corresponding feature type. The larger the user feature coefficient, the higher the degree of correlation between the user and the corresponding feature type. Conversely, the smaller the user feature coefficient, the lower the degree of correlation between the user and the corresponding feature type.

[0075] The feedback analysis module is used to obtain the behavior data of the user on the trial operation simulation interface, construct a user behavior table based on the behavior data, analyze the user behavior table through a deep learning model to obtain user behavior analysis data, evaluate the trial operation simulation interface according to the user feature coefficient and the user behavior analysis data to obtain an evaluation result, and send it to the interface optimization unit;

[0076] The specific process of evaluating the trial operation simulation interface is as follows:

[0077] S501. Obtain the behavior data of the user on the trial operation simulation interface. The behavior data includes the user ID and the corresponding user feature coefficient, the routed page, the stay time, the number of link jumps, and the number of function uses;

[0078] S502. Construct a timeline based on the user's operation time, and mark the behavior data according to the corresponding time nodes on the timeline to obtain a user behavior table. Analyze the user behavior table, including aggregation comparison and sorting, to obtain user behavior analysis data. The user behavior analysis data includes the user ID, the number of clicks on the high-frequency page, the number of clicks on the high-frequency link, and the number of clicks on the high-frequency function;

[0079] In the specific aggregation process, taking the stay time as the judgment factor, record the stay time through the attribute of the routed page, then accumulate the stay time of the same user on the same routed page, and then sort the stay time from large to small to obtain the top data as the high-frequency page;

[0080] S503. Obtain the user feature coefficient, and calculate the correlation coefficient Ui between the portrait feature factor in the user feature coefficient and the user behavior analysis data one by one according to the following formula: , where d is a preset weight coefficient, N is the total number of user operations, and fi is the number of clicks on the high-frequency page or the number of clicks on the high-frequency link or the number of clicks on the high-frequency function;

[0081] S504. Obtain a preset correlation judgment threshold. If the correlation coefficient Ui is greater than or equal to the correlation judgment threshold, the evaluation result of the trial operation simulation interface meets the user requirements;

[0082] If the correlation coefficient Ui is less than the correlation judgment threshold, the evaluation result of the trial operation simulation interface does not meet the user requirements.

[0083] The design optimization unit is used to obtain and process the evaluation result, and based on the evaluation result, perform user interface responsive design optimization on the page module corresponding to the trial operation simulation interface to adapt to different user requirements, and send the target user interface to the design display unit again.

[0084] The specific operations for performing responsive design optimization on the trial operation simulation interface are as follows:

[0085] Adjust the interface layout of the trial operation simulation interface;

[0086] Add new controls to the trial operation simulation interface;

[0087] Delete existing controls from the trial operation simulation interface;

[0088] Adjust the control attributes of the controls in the trial operation simulation interface;

[0089] Adjust the control styles of the controls in the trial operation simulation interface;

[0090] Adjust the link positions that the controls in the trial operation simulation interface need to jump to.

[0091] The present invention identifies each natural language command through a large language model, determines the design requirements expressed by each natural language command, determines the corresponding design system, page image, and page template based on the design requirements, generates a trial operation user interface design draft and displays it to the user side, establishes a user portrait based on the user's basic information, obtains a user feature coefficient, and constructs a user behavior table based on the behavior data. Analyze the user behavior table through a deep learning model, and then evaluate the trial operation simulation interface to obtain an evaluation result. Based on the evaluation result, perform user interface responsive design optimization on the page module corresponding to the trial operation simulation interface to adapt to different user requirements.

[0092] The setting of the size of the threshold is for the convenience of comparison. Regarding the size of the threshold, it depends on the amount of sample data and the base quantity set by those skilled in the art for each group of sample data; as long as it does not affect the proportional relationship between the parameters and the quantified values.

[0093] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters in the formulas are set by technicians in this field according to the actual situation;

[0094] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field of the present invention within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered within the protection scope of the present invention.

Claims

1. Visual communication design and display system based on model analysis, characterized by: It includes design requirement analysis unit, design output unit, design display unit, user feedback analysis unit and design optimization unit; The design requirement analysis unit includes a model recognition module and a requirement analysis module, wherein the model recognition module is used to obtain natural language commands, and recognize each natural language command through a large language model, determine the design requirements expressed by each natural language command, and send them to the requirement analysis module, wherein the large language model is obtained by training based on structured knowledge in the field of user interface design; The requirement analysis module is used to obtain design requirements, and determine the corresponding design system, page image and page template based on the design requirements, use the resources of the design system to construct the pseudo code corresponding to the page template, and send the pseudo code to the design output unit; The design output unit is used to convert the pseudo code into a corresponding vector design draft, and determine the corresponding style and filling content based on the page image, fill the vector design draft with the style and the filling content, and output the obtained trial user interface design draft to the design display unit; The design display unit is used to obtain the user interface design draft of the trial run and the target user interface, and generate a trial run simulation interface based on the user interface design draft of the trial run, and display the trial run simulation interface to the user end, and generate an actual operation interface based on the target user interface, and display the actual operation interface to the user end; The user feedback analysis unit includes a user analysis module and a feedback analysis module. The user analysis module is used to obtain basic information of the user, establish a user portrait according to the basic information, and classify the user portrait according to the clustering model to obtain a user feature recognition result. The user feature coefficient is calculated based on the user feature recognition result and sent to the feedback analysis unit; The specific process of calculating the user characteristic coefficient is as follows: S401. Obtaining basic information of the user, wherein the basic information includes user personal information, historical behavior data, and historical social data. The user personal information includes the user's gender, age, and occupation. The historical behavior data includes historical browsing records, historical click records, historical search data, and historical purchase data. The historical social data includes the user's interactive behavior on the social platform, including the number of likes, comment content, and sharing data. S402, adding tags to the user's personal information, historical behavior data, and historical social data to obtain a user tag set, determining a feature vector set based on the user tag set, and constructing a user portrait based on the user tag set and the feature vector set; S403, constructing a user feature matrix based on the user's behavioral characteristics and consumption characteristics, and identifying the feature type of the user portrait through a clustering model, specifically hierarchical clustering; S404: extract the feature keywords of the feature type, identify the portrait feature factors in the feature type, and then obtain the user feature recognition result, and calculate the user feature coefficient R (G, F) according to the following formula: , where e1 and e2 are preset proportional coefficients, n represents the number of samples corresponding to the image feature factor, Gi is the i-th sample value of the parameter corresponding to other factors in the image feature factor, is the sample mean of the parameters corresponding to other factors in the image feature factor, Fi is the i-th sample value of the parameters corresponding to other factors in the image feature factor, It is the sample mean of the corresponding parameters of other factors in the portrait feature factor. The user feature coefficient is used to indicate the relevance between the user and the corresponding feature type. The larger the user feature coefficient is, the higher the relevance between the user and the corresponding feature type is. On the contrary, the smaller the user feature coefficient is, the lower the relevance between the user and the corresponding feature type is. The feedback analysis module is used to obtain the user's behavior data on the trial operation simulation interface, construct a user behavior table based on the behavior data, analyze the user behavior table through a deep learning model to obtain user behavior analysis data, evaluate the trial operation simulation interface according to the user characteristic coefficient and the user behavior analysis data to obtain an evaluation result, and send it to the interface optimization unit; The design optimization unit is used to obtain and process the evaluation results, and based on the evaluation results, perform user interface responsive design optimization on the page modules corresponding to the trial operation simulation interface to adapt to different user needs, and obtain the target user interface and send it to the design display unit again.

2. The visual communication design and display system based on model analysis according to claim 1, characterized in that: The specific process of obtaining a large language model is as follows: S101, the natural language command includes user needs and goals, functional requirements, platform and device characteristics, visual design principles, interaction methods, accessibility, brand and visual identity, and technical feasibility; S102, constructing a knowledge graph based on data from different natural language command sources, using a depth-first search algorithm to obtain a language entity for each entity based on a search path corresponding to each entity in each knowledge graph, and performing cleaning and word segmentation on the language entity to obtain training data; S103, using a clustering algorithm to obtain the relevance of training data for each knowledge graph, and obtaining target task data corresponding to the training data; S104. Build a large language training model base based on the Transformer architecture, input the training data into the large language training model base for pre-training to obtain a pre-trained large language model, and fine-tune the pre-trained model using the target task data to obtain an optimized large language model.

3. The visual communication design and display system based on model analysis according to claim 1, characterized in that: The specific process of getting the user interface design draft for trial operation is as follows: S201, retrieving associated styles and filling contents from a network database based on the page image, and integrating the styles and filling contents into a candidate data set; S202, calculating the correlation coefficient between each style and filling content in the candidate data set and the design requirement according to the clustering algorithm, and expressing it through a specific numerical value, and marking the correlation coefficient at the subscript of the style and filling content; S203, rearrange the styles and filling contents in the candidate data sets according to the correlation coefficients from large to small, fill the several groups of styles and filling contents arranged in the front row into the vector design draft one by one, and obtain the pseudo user interface design draft; S204, inputting the proposed user interface design draft into the target model, obtaining the verification result output by the target model, and selecting the user interface design draft for trial operation based on the verification result; The target model is a neural network model obtained by iteratively training a number of normal user interface design drafts and a number of abnormal user interface design drafts, wherein the abnormal user interface design drafts are user interface components that do not conform to preset design rules.

4. The visual communication design and display system based on model analysis according to claim 1, characterized in that: The specific process of generating the trial run simulation interface is as follows: S301, obtaining interface configuration data corresponding to the user interface design draft, generating an initial user interface based on the interface configuration data, wherein the interface configuration data is used to represent the style, filling content and page template of the user interface design draft, and the initial user interface is displayed in the user terminal; S302: Configure the initial user interface to generate a trial operation simulation interface, where the trial operation simulation interface matches the user interface design draft.

5. The visual communication design and display system based on model analysis according to claim 1, characterized in that: The specific process of evaluating the trial operation simulation interface is as follows: S501, obtaining user behavior data on the trial operation simulation interface, wherein the behavior data includes a user ID and a corresponding user characteristic coefficient, a routed page, a stay time, a link jump number, and a function usage number; S502: construct a timeline based on the user's operation time, and mark the behavior data according to the corresponding time nodes on the timeline to obtain a user behavior table, analyze the user behavior table, including aggregation comparison and sorting, to obtain user behavior analysis data, the user behavior analysis data including user ID, number of clicks on high-frequency pages, number of clicks on high-frequency links, and number of clicks on high-frequency functions; S503: Obtain user feature coefficients, and calculate the correlation coefficients Ui between the portrait feature factors in the user feature coefficients and the user behavior analysis data one by one according to the following formula: , where d is the preset weight coefficient, N is the total number of user operations, and fi is the number of clicks on high-frequency pages or high-frequency links or high-frequency functions; S504, obtaining a preset association judgment threshold, if the association coefficient Ui is greater than or equal to the association judgment threshold, then the evaluation result of the trial operation simulation interface meets the user's needs; If the correlation coefficient Ui is less than the correlation judgment threshold, the evaluation result of the trial operation simulation interface is that it does not meet the user needs.

6. The visual communication design and display system based on model analysis according to claim 1, characterized in that: The specific operations for responsive design optimization of the trial run simulation interface are as follows: Adjusting the interface layout of the trial operation simulation interface; Adding new controls in the trial run simulation interface; Deleting existing controls from the trial run simulation interface; Adjusting the control properties of the controls in the trial operation simulation interface; Adjusting the control style of the control in the trial operation simulation interface; The link position of the control that needs to jump in the trial operation simulation interface is adjusted.

Images