Emotional fusion-based cultural creative design auxiliary method and emotion fusion-based cultural creative design auxiliary system

Abstract

The invention relates to the technical field of creative design assistance, and particularly discloses a creative design assistance method and system based on emotion fusion. The method mainly comprises the following steps: acquiring emotion data input by a user; performing sentiment analysis to extract sentiment features; matching corresponding elements from a design element database according to the emotion features; fusing the emotion features with the design elements through an emotion fusion engine to generate an emotional design draft; outputting a draft and receiving user feedback; and optimizing engine parameters according to the feedback. The corresponding system comprises a data acquisition module, an emotion analysis module, a design element database, an element matching module, an emotion fusion engine, a user interaction module and an optimization learning module. According to the method, the capability of intelligently converting the user emotion into the design element is realized, continuous optimization can be realized through interactive feedback, and the intelligent level and the personalized output quality of the cultural and creative design are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of creative design assistance, more specifically, the present application relates to a creative design assistance method and system based on emotional fusion. BACKGROUND

[0002] In recent years, the field of creative design has shown a growing trend towards digitization, personalization, and emotionalization. With the popularization of computer-aided design technology, the design process has rapidly evolved from traditional handcrafting to digital creation. The introduction of big data and artificial intelligence technology has significantly improved the efficiency of managing and combining design elements, and designers have unprecedented access to resources. The current trend focuses on how to translate the emotional factors in user or cultural connotations into specific design language to achieve a deeper emotional resonance between the work and the people, and to meet the growing demand for personalized custom design.

[0003] However, existing creative design assistance technologies have obvious shortcomings. Most systems only support mechanical matching based on keywords or tags, lacking the ability to deeply understand and integrate complex emotional intentions in user input. The combination of design elements is often rigid and template-based, making it difficult to express subtle emotional intensity, resulting in design drafts that lack personality and appeal. In addition, existing technologies have poor closed-loop performance and cannot effectively optimize themselves based on user feedback on preliminary results, making it difficult for the system to adapt to changing user preferences and limiting its long-term value.

[0004] Therefore, in view of the above problems, a creative design assistance method and system based on emotional fusion are proposed, aiming to solve the key problems in the above-mentioned existing technologies, namely: how to deeply understand and quantify user emotional input; how to achieve intelligent, subtle, and dynamic fusion between emotional features and design elements; and how to continuously optimize the design generation process through interactive feedback. SUMMARY

[0005] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a creative design assistance method and system based on emotional fusion to solve the problems raised in the above background.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical scheme: a creative design assistance method based on emotional fusion, comprising the following steps:

[0007] S1: obtaining emotional data input by a user, the emotional data including at least one of text, image, and voice;

[0008] S2: performing emotional analysis on the emotional data obtained in step S1 to extract emotional features, the emotional features including at least emotional type and emotional intensity value;

[0009] S3: matching associated design elements from a pre-constructed design element database based on the emotional features extracted in step S2, the design element database storing a plurality of design elements and an emotional label associated with each design element;

[0010] S4: generating an emotional design draft by fusing the emotional features obtained in step S2 and the design elements matched in step S3 through an emotional fusion engine, the emotional fusion engine dynamically adjusting the combination weight of the design elements according to the emotional intensity value in the range of 0% to 100%;

[0011] S5: outputting the emotional design draft generated in step S4 and receiving user interaction feedback based on the design draft;

[0012] S6: optimizing and updating the fusion parameters of the emotional fusion engine according to the user feedback received in step S5.

[0013] Preferably, in step S1, the emotional data input by the user specifically includes: receiving the emotional text description directly input by the user through the user interface, or obtaining the image or voice data provided by the user through the integrated external device; and pre-processing the obtained raw data, the pre-processing including at least one of data cleaning and format standardization.

[0014] Preferably, in step S2, the emotional analysis of the emotional data is realized by using a pre-trained emotional analysis model, the emotional analysis model being constructed based on a machine learning algorithm and being used to identify emotional polarity from text data, or to identify visual emotional elements from image data, or to identify emotional state corresponding to intonation from voice data.

[0015] Preferably, in step S3, the process of matching design elements from the design element database uses a similarity calculation algorithm to match and calculate the emotional features and the emotional labels, and selects a design element set with a similarity higher than a preset threshold, wherein the weight assigned to the i-th matched design element in the fusion is: The sum of all weights is 1, I represents the emotional intensity value analyzed from the user input, A i represents the association degree of the emotional label of the i-th design element with the current dominant emotional type, k is an adjustable gain coefficient for controlling the sensitivity of the emotional intensity I to the weight distribution, and n represents the total number of all matched design elements.

[0016] Preferably, in step S4, the emotional fusion engine performs fusion processing, and also performs conflict detection, when the emotional labels associated with the plurality of matched design elements conflict, the design element with high consistency with the dominant emotional type is preferentially retained.

[0017] A sentiment fusion-based text and creative design assistance system implementing the above method, comprising:

[0018] a data acquisition module for acquiring user input sentiment data;

[0019] a sentiment analysis module connected to the data acquisition module for sentiment analysis of the sentiment data and extraction of sentiment features;

[0020] a design element database for storing design elements and their associated sentiment labels;

[0021] an element matching module connected to the sentiment analysis module and the design element database respectively, for matching design elements based on sentiment features;

[0022] a sentiment fusion engine connected to the element matching module for fusing sentiment features with matched design elements to generate a sentiment-based design draft;

[0023] a user interaction module connected to the sentiment fusion engine for outputting the design draft and receiving user interaction feedback;

[0024] an optimization learning module connected to the user interaction module and the sentiment fusion engine respectively, for optimizing the parameters of the sentiment fusion engine according to user feedback.

[0025] Preferably, the sentiment analysis module includes at least one of a natural language processing unit for processing text data, a computer vision unit for processing image data, and a speech emotion recognition unit for processing voice data.

[0026] Preferably, the sentiment fusion engine is constructed based on a rule engine or a neural network model, for performing dynamic allocation of design element combination weights according to sentiment intensity values.

[0027] Preferably, the user interaction module provides a visual interface for presenting the design draft in the form of two-dimensional graphics, three-dimensional models or dynamic demonstrations, and integrates editing tools for users to make real-time adjustments to the design draft.

[0028] Preferably, the optimization learning module uses a reinforcement learning algorithm to adjust the weight parameters of the sentiment fusion engine through user ratings, modification behavior or usage frequency data of the design draft.

[0029] Technical effects and advantages of the present application:

[0030] Compared to existing technologies, this invention achieves intelligent design processes by constructing a method and system that incorporates sentiment analysis, element matching, and fusion generation. Specifically, the system first performs multimodal sentiment analysis on user-provided text, image, or voice data to accurately extract sentiment types and intensities; then, it intelligently matches design elements from a database based on sentiment features; finally, an sentiment fusion engine dynamically adjusts the weights of element combinations according to sentiment intensity to generate a preliminary design draft. This approach transforms abstract sentiment into concrete design language, effectively solving the problems of traditional methods relying on manual interpretation of sentiment and rigid element matching, significantly improving the efficiency of design conception and the emotional relevance of creative output.

[0031] Compared to existing technologies, this invention introduces a parameter optimization mechanism based on user feedback, enabling the system to learn and continuously improve itself. After outputting a design draft, the system collects user actions (such as modifications and ratings) as feedback data and dynamically adjusts the matching and fusion parameters of the emotion fusion engine using an optimization learning module (such as a reinforcement learning algorithm). This mechanism allows the system to continuously adapt to changes in users' personal styles and preferences, solving the problems of existing systems being static, rigid, and unable to evolve personalizedly. This ensures the long-term effectiveness and relevance of the assistance, continuously improving the user experience.

[0032] Compared to existing technologies, this invention enhances the coordination and richness of design solutions by integrating conflict detection and multi-source data fusion functions into the emotion fusion engine. When multiple matching design elements have conflicting emotion tags, the system can intelligently arbitrate based on the dominant emotion; simultaneously, it can integrate multi-source information such as user historical data and social media sentiment to generate more comprehensive emotion features. This effectively avoids the element piling up and logical confusion in the generated solutions, improves the internal consistency and cultural context adaptability of the design draft, and makes the output not only accurately express emotions but also more complete and in-depth in design. Attached Figure Description

[0033] Fig. 1 This is a flowchart of the method of the present invention.

[0034] Fig. 2 This is a system overall framework diagram of the present invention.

[0035] Fig. 3 This is a schematic diagram illustrating the emotional integration of the present invention. Detailed Implementation

[0036] 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 of 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.

[0037] Embodiment 1

[0038] As shown in the attached Figs. 1-3 A cultural and creative design assistance method and system based on emotion fusion are as follows: the complete implementation process is as follows:

[0039] The core of the present invention is to provide a closed-loop and intelligent cultural and creative design assistance process, which converts the user's emotional input into a specific design draft and can continuously self-optimize through interactive feedback. The entire implementation process can be clearly divided into six main stages: emotional data acquisition and preprocessing, emotional feature extraction, intelligent matching of design elements, emotion fusion and draft generation, draft interaction and feedback, and system optimization and learning.

[0040] The first stage: Emotional data acquisition and preprocessing

[0041] The starting point of the process is to obtain the user's emotional input. The system provides multiple data input channels through the user interface to adapt to different user preferences and scenarios. The user can choose to directly input a text to describe the emotion that the expected design is to convey (for example, "I hope this bookmark design can convey a sense of tranquility and Zen"); or upload a picture as an emotional reference (for example, a Chinese ink painting or a peaceful lake); or directly describe it through the microphone (for example, "I want a pattern full of vitality and festive atmosphere").

[0042] After receiving this raw data, the system does not immediately perform in-depth analysis, but first performs key preprocessing steps to ensure the accuracy and efficiency of subsequent analysis. For text data, preprocessing includes removing meaningless stop words (such as "de", "le"), punctuation marks, and performing word segmentation to convert the sentence into a series of meaningful lexical units. For image data, preprocessing may include size normalization, noise reduction, and color space conversion (for example, from RGB to HSV color space that is more in line with human visual perception) to highlight the color and texture features of the image. For voice data, preprocessing includes noise reduction, silent segment excision, and voice endpoint detection to ensure that only valid voice segments are analyzed. The goal of this stage is to convert the chaotic raw data into clean, regular, and standardized data for the next stage of analysis. The preprocessed data is temporarily stored and ready to be sent to the emotion analysis module.

[0043] Phase Two: Sentiment Feature Extraction

[0044] In this stage, the system will conduct in-depth analysis of the preprocessed data, aiming to transform unstructured emotional information into quantifiable and structured emotional features. It can process multimodal data (text, image, speech) and output emotional feature vectors in a unified format. The emotional features mainly include two core dimensions: emotional type and emotional intensity.

[0045] Text Sentiment Analysis: The system uses a pre-trained natural language processing model (e.g., a sentiment analysis model based on BERT architecture) to process text. The model identifies the emotional polarity (e.g., positive, negative) and specific emotional categories (e.g., "tranquil" and "Zen" belong to the calm category; "energetic" and "festive" belong to the excitement category). Simultaneously, the model outputs a confidence score, which, after normalization, serves as the sentiment intensity value (I). For example, for the phrase "very tranquil," the model might identify the sentiment type as "calm" and assign a relatively high intensity value, such as 0.9 (intensity range 0-1).

[0046] Image sentiment analysis: The system uses computer vision techniques to analyze images. First, it extracts low-level visual features, such as color histograms (to determine whether the dominant color tone is cool or warm), texture features (to determine whether it is smooth or rough), and compositional features (to determine whether it is symmetrical or asymmetrical). These features are then fed into an image sentiment recognition model, trained on a large amount of image data with sentiment labels, capable of mapping visual features to a sentiment space. Finally, the model outputs the main sentiment type expressed by the image and its intensity. For example, a picture of a lake with a predominantly blue tone and a balanced composition might be identified as "calm," with an intensity of 0.85.

[0047] Speech emotion analysis: The system identifies emotions by analyzing the acoustic features of the speech signal. These features include fundamental frequency (related to pitch, higher when excited), energy (related to volume), and speech rate. The speech emotion recognition model analyzes these features to determine the speaker's emotional state (e.g., happy, sad, calm) and assigns an intensity value.

[0048] Regardless of the input source, the final output of this stage is a standardized sentiment feature vector. This vector contains encoded information about the emotion type and the emotion intensity value I.

[0049] For example, This structure It is the core data that drives all subsequent processes.

[0050] Phase 3: Intelligent Matching of Design Elements

[0051] At this stage, the system will use the sentiment feature vector obtained in the previous step. The design elements are compared with a pre-built database of design elements to select those that best reflect the emotional characteristic. This database is a structured knowledge base storing a vast amount of basic design units, such as color, shape (geometric shapes, organic forms), texture (wood, fabric, metallic textures), pattern (totems, geometric patterns), and font styles. Crucially, each element in the database is not isolated but rather associated with one or more emotional tags. Labels can be created manually using expert knowledge or automatically generated by machine learning models after analyzing a large number of design works.

[0052] The core of the matching process is calculating the sentiment feature vector. The sentiment tag vector of each design element in the database The similarity between them. This invention uses a cosine similarity algorithm to achieve accurate semantic-level matching.

[0053]

[0054] Here, S represents the similarity score, which ranges from 0 to 1. The closer the value is to 1, the more similar the two are in terms of sentiment and semantics.

[0055] This represents the sentiment feature vector extracted from user input.

[0056] This represents the sentiment tag vector of a specific element in the design element database.

[0057] Representative vector with vector The dot product.

[0058] Representative vector The modulus (length).

[0059] Representative vector The modulus (length).

[0060] The system will calculate the relationship between each element in the database and the current element. The system then sets a similarity threshold (e.g., S > 0.7) and selects all elements that exceed this threshold to form a "matching element candidate set". This process ensures that the selected elements are highly relevant to the user's intent in terms of sentiment semantics, rather than simply matching keywords.

[0061] Phase Four: Emotional Integration and Draft Generation

[0062] The task at this stage is to take the matched design elements and the emotional intensity I as input, and generate a complete and harmonious emotional design goal through the emotional fusion engine.

[0063] The core of the emotion fusion engine is a dynamic weight allocation algorithm. It determines the "weight" of each matching element in the final design based on the emotion intensity I. This invention employs a variant algorithm based on the Softmax function to achieve this.

[0064]

[0065] Among them, W i This represents the weight assigned to the i-th matching design element during fusion. The sum of all weights is 1.

[0066] I represents the sentiment intensity value obtained from user input.

[0067] A i The correlation between the emotional label representing the i-th design element and the currently dominant emotional type is a preset value (for example, the correlation A_i between a "sky blue" color block and the "calm" emotion may be 0.95).

[0068] k is an adjustable gain coefficient used to control the sensitivity of the emotion intensity I to the weight distribution.

[0069] n represents the total number of all design elements matched in this instance.

[0070] e is the natural constant.

[0071] j is the loop variable.

[0072] Example of calculation process: Assume 3 elements (color blocks) are matched, the sentiment intensity I = 0.8, k = 1. The A1 of element 1 (dark blue) is 0.9, the A2 of element 2 (light blue) is 0.7, and the A3 of element 3 (light green) is 0.6.

[0073] First, calculate the molecule:

[0074] e^(0.8*0.9)≈e^0.72≈2.054, e^(0.8*0.7)≈e^0.56≈1.751, e^(0.8*0.6)≈e^0.48≈1.616.

[0075] The denominator is the sum of the three: 2.054 + 1.751 + 1.616 = 5.421.

[0076] The weights are as follows:

[0077] W1=2.054 / 5.421≈0.379, W2=1.751 / 5.421≈0.323, W3=1.616 / 5.421≈0.298.

[0078] It is evident that dark blue, which is most strongly associated with dominant emotions, received the highest weight.

[0079] While assigning weights, the fusion engine also incorporates a conflict detection mechanism. It checks the candidate set for elements with severe emotional semantic conflicts (e.g., the simultaneous presence of cool colors representing "calm" and warm colors representing "excitement"). If a conflict is detected, the engine prioritizes retaining elements that are highly consistent with the dominant emotional type, while suppressing or eliminating conflicting elements to ensure the internal consistency of the design draft.

[0080] Finally, based on the calculated weights, the engine combines the various design elements according to design rules (such as color matching rules and composition principles) to generate a preliminary, visual, and emotionally engaging design draft. This can be a two-dimensional graphic design, a three-dimensional model rendering, or a simple dynamic effect demonstration.

[0081] Phase 5: Draft Interaction and Feedback

[0082] The system presents the generated emotional design draft to the user through a graphical user interface. Users can directly view and interact with this preliminary design. The system provides a series of editing tools that allow users to fine-tune the draft, such as adjusting the brightness of a color, scaling a graphic, replacing an unsatisfactory element, or directly rating the overall draft (e.g., 1-5 stars).

[0083] All these user actions, whether explicit ratings or implicit modifications, are automatically captured by the system and transformed into structured feedback data. For example, a user changing the system-generated dark blue tone to light blue is a strong signal that the user is reserved about high-intensity "calm" expression and prefers medium intensity. This feedback data is a valuable resource for the system to optimize itself.

[0084] Phase 6: System Optimization and Learning

[0085] The key to realizing the long-term value of this invention lies in this final closed-loop stage. The optimization learning module collects all feedback data generated by the user in the previous stage.

[0086] Its core optimization goal is to adjust the parameters in the emotion fusion engine (such as the gain coefficient k in the weight calculation formula or the emotional relevance A of the elements). iThis allows the system to generate design drafts for users that better align with their preferences in the next iteration. This process typically employs reinforcement learning algorithms. The system treats positive user feedback (such as high ratings or unmodified adoption) as "rewards" and negative feedback (such as low ratings or significant modifications) as "penalties." By continuously experimenting and adjusting its strategy based on feedback (i.e., fusing parameters), the algorithm eventually learns a personalized generation strategy best suited for a specific user.

[0087] For example, if the system repeatedly detects that when it generates a "calm" theme design with a high intensity I (dark colors), the user consistently lightens the colors, the optimization algorithm will gradually reduce the relevance A of dark elements to that user under the "calm" emotional state. i Alternatively, the coefficient k can be adjusted so that when a draft is generated in the future, the initial weight W of the dark elements is... i

[0088] The system naturally lowers its color intensity, favoring medium-brightness colors. This allows for personalized adaptation and continuous improvement of the system.

[0089] The above six stages constitute the complete and closed-loop implementation process of this invention. From the input of emotional data to precise feature extraction and element matching, then to the core emotional fusion generation, and finally to the system's self-iteration through user interaction feedback, the entire process has a clear data flow and a well-defined algorithm calculation process, which together ensures the intelligent, personalized, and continuous optimization capabilities of the cultural and creative design assistance process.

[0090] Finally, the following points should be noted: First, in the description of this application, it should be noted that, unless otherwise specified and limited, the terms "installation", "connection", and "linkage" should be interpreted broadly, and can be mechanical or electrical connections, or internal connections between two components, or direct connections. "Up", "down", "left", "right", etc. are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may change.

[0091] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0092] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An emotional fusion-based cultural and creative design assistance method, characterized in that, The method comprises the following steps: S1: obtaining user input emotion data, the emotion data comprising at least one of text, image and voice; S2: performing emotion analysis on the emotion data obtained in step S1 to extract emotion features, the emotion features comprising at least emotion type and emotion intensity value; S3: based on the emotion features extracted in step S2, matching associated design elements from a pre-constructed design element database, the design element database storing a plurality of design elements and emotion labels associated with each design element; S4: performing fusion processing on the emotion features obtained in step S2 and the design elements matched in step S3 by an emotion fusion engine to generate an emotional design draft, the emotion fusion engine dynamically adjusting the combination weight of the design elements according to the emotion intensity value within the range of 0% to 100%; S5: outputting the emotional design draft generated in step S4 and receiving user interaction feedback based on the design draft; S6: optimizing and updating the fusion parameters of the emotion fusion engine according to the user feedback received in step S5.

2. The emotion fusion-based design assistance method for cultural and creative products of claim 1, wherein, In step S1, the obtaining of user input emotion data specifically comprises: receiving emotion text description directly input by the user through a user interface, or obtaining image or voice data provided by the user through an integrated external device; and performing preprocessing on the obtained raw data, the preprocessing comprising at least one of data cleaning and format standardization. 3.The emotion fusion-based design assistance method for cultural and creative products of claim 1, wherein, In step S2, the emotion analysis on the emotion data is realized by using a pre-trained emotion analysis model, the emotion analysis model being constructed based on a machine learning algorithm and being used to identify emotion polarity from text data, or to identify visual emotion elements from image data, or to identify emotion state corresponding to intonation from voice data.

4. The emotion fusion-based design assistance method for cultural and creative products of claim 1, wherein, In step S3, the process of matching design elements from the design element database adopts a similarity calculation algorithm, matches and calculates the emotional characteristics with the emotional labels, and selects a design element set with a similarity higher than a preset threshold, wherein the i th matching design element is assigned a weight in fusion: The sum of all weights is 1, I represents the emotional intensity value analyzed from the user input, A i represents the association degree of the emotional label of the i th design element with the current dominant emotional type, k is an adjustable gain coefficient for controlling the sensitivity of the emotional intensity I to the weight distribution, and n represents the total number of all design elements matched this time.

5. The emotion fusion-based design assistance method for cultural and creative products of claim 1, wherein, In step S4, when the emotion fusion engine performs fusion processing, it also performs conflict detection, and when the emotion labels associated with the matched multiple design elements conflict, the design element with high consistency with the dominant emotion type is preferentially retained.

6. An emotional fusion-based text creative design assistance system for implementing the method according to any one of claims 1 to 5. The method comprises: a data acquisition module for acquiring user input emotion data; an emotion analysis module connected with the data acquisition module, for performing emotion analysis on the emotion data and extracting emotion features; a design element database for storing design elements and emotion labels associated therewith; an element matching module connected with the emotion analysis module and the design element database, respectively, for matching design elements based on emotion features; an emotion fusion engine connected with the element matching module, for fusing emotion features with matched design elements to generate an emotional design draft; a user interaction module connected with the emotion fusion engine, for outputting the design draft and receiving user interaction feedback; an optimization learning module connected with the user interaction module and the emotion fusion engine, respectively, for optimizing the parameters of the emotion fusion engine according to user feedback.

7. The emotion fusion-based design assistance system for cultural and creative products of claim 6, wherein The emotion analysis module comprises at least one of a natural language processing unit for processing text data, a computer vision unit for processing image data, and a voice emotion recognition unit for processing voice data. The emotion analysis module comprises at least one of a natural language processing unit for processing text data, a computer vision unit for processing image data, and a voice emotion recognition unit for processing voice data. 8.The emotion fusion-based creative design auxiliary system according to claim 6, wherein, The emotional fusion engine is constructed based on a rule engine or a neural network model, and is used to perform dynamic allocation of design element combination weights according to emotional intensity values. 9.The emotion fusion-based creative design auxiliary system according to claim 6, wherein, The user interaction module provides a visual interface for presenting a design draft in the form of a two-dimensional graph, a three-dimensional model or a dynamic demonstration, and integrates editing tools for real-time adjustment of the design draft by the user. 10.The emotion fusion-based creative design auxiliary system according to claim 6, wherein, The optimization learning module adopts a reinforcement learning algorithm to adjust weight parameters of the emotional fusion engine based on user scores, modification behaviors or usage frequency data of the design draft.

Images