Automobile design optimization method based on perceptual requirements

Abstract

The invention discloses an automobile design optimization method based on perceptual requirements. The method comprises the steps of obtaining target comment data about an automobile; obtaining a comment keyword and a user group type according to the target comment data; performing matching relationship verification on the comment keyword to obtain a target keyword pair; obtaining attention features according to the comment keywords; obtaining a user perceptual demand portrait according to the target key vocabulary pair, the attention feature and the user group type; and optimizing the automobile design according to the user perceptual demand portrait. According to the scheme, the matching relation between the design elements and the perceptual words in the user comments can be systematically mined, the perceptual cognition structure model of the user is constructed in combination with the user background and attention characteristics, the perceptual demand difference of different user groups under different car purchase purposes is visually presented, and the user experience is improved. Therefore, a scientific and data-driven support means is provided for automobile appearance design.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to an optimization method for automobile design based on emotional needs. Background Technology

[0002] With the continued development of intelligent manufacturing and personalized consumption trends, users' emotional needs for automotive products are becoming increasingly diverse. Automotive exterior design has shifted from a traditional function-oriented approach to a user-oriented approach centered on emotional experience. Especially in the new energy vehicle sector, where product homogenization is severe, exterior design has become a crucial means of shaping brand awareness and emotional value. Traditional automotive design methods rely on designers' experience and subjective judgment, lacking systematic modeling and quantitative analysis of target users' emotional perceptions, making it difficult to meet diverse market demands. Summary of the Invention

[0003] In view of the above, the present invention aims to provide an optimization method for automobile design based on emotional needs, in order to solve the aforementioned technical problems.

[0004] The technical solution adopted in this invention is as follows:

[0005] This invention provides an optimization method for automobile design based on emotional needs, including:

[0006] Obtain target review data about cars;

[0007] Based on the target comment data, the comment keywords and user group types are obtained;

[0008] The matching relationship of the comment keywords is verified to obtain the target keyword pairs;

[0009] Based on the keywords in the comments, attention characteristics are obtained;

[0010] Based on the target keyword pairs, the attention characteristics, and the user group types, a user emotional needs profile is obtained;

[0011] Based on the user's emotional needs profile, optimize the car design.

[0012] Optionally, obtain targeted review data about the car, including:

[0013] Scrape user review data about cars from online platforms;

[0014] After preprocessing the user review data, entity annotation is performed to obtain target review data about cars. The target review data includes user ID, car model name, purchase price, purchase time, purchase IP, purchase purpose, review time, and review text content.

[0015] Optionally, based on the target comment data, the user group type can be obtained, including:

[0016] Based on the purchase price, the IP address of the purchaser, and the purpose of the purchase, the user group type is determined.

[0017] Optionally, based on the target comment data, comment keywords are obtained, including:

[0018] The comment text is input into a pre-trained text recognition model for entity recognition to obtain comment keywords; the comment keywords include design elements and emotional vocabulary.

[0019] Optionally, the matching relationship of the comment keywords is verified to obtain target keyword pairs, including:

[0020] Randomly pair the design elements and emotional words in the keywords to obtain multiple pairing combinations;

[0021] Multiple pairings are filtered to obtain target keyword pairs that meet the preset pairing relationship. Each target keyword pair includes a target design element and a target emotional word.

[0022] Optionally, attention features can be obtained based on the comment keywords, including:

[0023] Based on the frequency of the target design elements appearing in the target comment data, attention features are determined.

[0024] Optionally, based on the target keyword pair, the attention characteristics, and the user group type, a user emotional needs profile is obtained, including:

[0025] Based on the target keyword pairs, determine the perceptual dimension score of the features of interest;

[0026] The emotional dimension scoring of the focus feature includes emotional dimension word scores of multiple categories, and the emotional dimension words are determined by the frequency of occurrence of emotional words in the target comment data;

[0027] Based on the emotional dimension scores of the aforementioned attention features and the user group type, a user emotional needs profile is obtained.

[0028] Optionally, based on the emotional dimension scores of the aforementioned attention features and the user group type, a user emotional needs profile is obtained, including:

[0029] Calculate the average score of each user group type on each emotional dimension of the features they care about to obtain a profile of user emotional needs.

[0030] Optionally, based on the target keywords, determine the emotional dimension score of the features of interest, including:

[0031] When the target emotional vocabulary belongs to the emotional dimension words, the score of the corresponding emotional dimension words of the focus feature is increased by one point, while the scores of other emotional dimension words remain unchanged.

[0032] When the target emotional words do not belong to the emotional dimension words, the correlation between the target emotional words and each emotional dimension word is calculated based on the cosine distance, and the emotional dimension word score is determined based on the correlation.

[0033] Based on the set of scores for multiple perceptual dimension words, determine the perceptual dimension score of the feature of interest.

[0034] Optionally, the text recognition model is trained, including:

[0035] Word embedding is used to convert words in text comments into fixed-dimensional text vectors.

[0036] The text vectors are input into the text recognition model, which captures the contextual semantic information of each text vector in the text sequence and concatenates them to obtain a comprehensive representation vector of each text vector in the context. This representation vector is then input into the CRF layer, which outputs words with pre-defined annotations.

[0037] The above-described solution of the present invention has at least the following beneficial effects:

[0038] The above-described solution of the present invention obtains target review data about automobiles; based on the target review data, it obtains review keywords and user group types; it verifies the matching relationship of the review keywords to obtain target keyword pairs; based on the review keywords, it obtains attention characteristics; based on the target keyword pairs, the attention characteristics, and the user group types, it obtains a user emotional needs profile; and based on the user emotional needs profile, it optimizes automobile design. This system can systematically mine the pairing relationship between design elements and emotional words in user reviews, combine user background and attention characteristics, construct a structural model of user emotional cognition, and visually present the differences in emotional needs of different user groups under different car purchase purposes in the form of radar charts, thereby providing a scientific and data-driven support method for automobile exterior design. Attached Figure Description

[0039] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described below with reference to the accompanying drawings, wherein:

[0040] Figure 1 A flowchart illustrating an optimization method for automobile design based on emotional needs, provided in an embodiment of the present invention.

[0041] Figure 2 A radar diagram illustrating the optimization method for automotive design based on emotional needs provided in this embodiment of the invention. Detailed Implementation

[0042] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0043] This invention proposes an embodiment of an optimization method for automobile design based on emotional needs, specifically, as follows: Figure 1 As shown, it includes:

[0044] Step 11: Obtain target review data about the car;

[0045] In this embodiment, a web crawler is used to scrape data from mainstream automotive vertical websites. The target data scraped includes user ID, vehicle model name, purchase price, purchase time, purchase IP address, purchase purpose, review time, and review text content.

[0046] Next, the crawled data is preprocessed: regular expressions are used to remove HTML tags, advertising text, and meaningless characters; images, tags, and special symbols in the text are filtered; duplicate detection and deduplication are performed on the data, and comments with ambiguous content, empty content, or only non-sentimental words are removed; the Jieba library is used to segment the text and generate corresponding columns; similar words are unified; and standardized comment data is obtained.

[0047] Furthermore, Docano was used to annotate the comment text content, labeling design elements, sentiment words, and pairing relationships. An example comment is: "DMS is too sensitive, triggering frequently during normal driving." The annotation results are: Design element: DMS; Sensation words: sensitive, frequent; Correct pairing relationships: DMS + sensitive, DMS + frequent.

[0048] Finally, export the annotated comments in JSONL file format.

[0049] Step 12: Based on the target comment data, obtain the comment keywords and user group types;

[0050] In this embodiment, this section uses three latent variables—purchase price, purchase IP address, and purchase purpose—to classify user groups.

[0051] By identifying user group types, it's easier to pinpoint market segments and user preferences, thus gaining more comprehensive information on user needs. City-level car-buying IPs help capture information such as living environment, culture, and habits; purchase price represents the consumer's economic status and the importance they place on the car; and purchase purpose helps analyze whether different purposes lead to different emotional needs.

[0052] Furthermore, keywords are extracted from the comment text. A neural network model is used here for automatic keyword recognition in the comment text.

[0053] Before performing entity recognition, the model needs to be trained to obtain a model suitable for text content entity recognition. Specifically, the design elements and sentiment words in JSONL format are transformed into a structured data structure optimized for machine learning models.

[0054] The conversion style is as follows: {"text": "DMS is too sensitive, frequently triggered during normal driving;", "label": {"Design element": {"DMS": [[0, 3]]}, "Sentiment words": {"Sensitive": [[4, 6]], "Frequent": [[11, 13]]}}}. The data is divided into training, validation, and test sets for model training and evaluation.

[0055] In this embodiment, a BiLSTM-CRF model is used to train preprocessed data and extract keywords from comments. The model consists of a Bidirectional Long Short-Term Memory (BiLSTM) network and a Conditional Random Field (CRF), and the training process is as follows:

[0056] Word embedding is used to convert words in text comments into fixed-dimensional text vectors.

[0057] BiLSTM encoding: The BiLSTM model captures the contextual semantic information of each position in the text sequence from both the forward and backward directions, and concatenates the final hidden state outputs from the two directions to obtain the comprehensive representation vector of each input unit in the context.

[0058] CRF Decoding: The vector sequence output by the BiLSTM model is input into the CRF layer. The CRF layer calculates the score of all possible label sequences given the input sequence by modeling the transition relationship between labels, and selects the globally optimal label sequence through Viterbi algorithm dynamic programming.

[0059] Joint training: Using training data with word-level labels as supervision signals, and employing the negative log-likelihood loss of CRF as the loss function, the parameters of the BiLSTM model and the CRF layer are jointly trained end-to-end to maximize the probability of the correct label sequence.

[0060] After training, the BiLSTM-CRF model can predict the named entity label category for each word in the input text comment and perform BIO annotation. B indicates that a word is the start of a named entity, i.e., a design element; I indicates that a word belongs to the interior of a named entity, i.e., a sentiment word; and O indicates that a word does not belong to any named entity. Ultimately, it achieves automatic identification and extraction of design elements or sentiment words in the text.

[0061] The comment text is input into a pre-trained text recognition model for entity recognition to obtain comment keywords; these keywords include design elements and emotional terms.

[0062] Step 13: Verify the matching relationship of the comment keywords to obtain target keyword pairs;

[0063] In this embodiment, after the BiLSTM-CRF model identifies the comment keywords, it is necessary to verify whether the pairing relationship between the design elements and the emotional words is correct. Since it is only necessary to determine whether the two match, this relationship extraction task can be modeled as a binary classification problem. For this purpose, a multilayer perceptron (MLP) model is adopted, which enhances the feature representation ability through hidden layers, uses nonlinear activation functions to handle complex associations, and can flexibly adjust the output layer structure according to task requirements.

[0064] Before training the MLP model, “design elements-emotional vocabulary” need to be extracted from the original labeled data. Positive and negative samples are constructed through full combination to generate all possible pairings. The correctness of the pairing relationship is determined by manual or rule-based labeling (correct pairing is labeled as 1, and incorrect pairing is labeled as 0).

[0065] The pairings are shown in the table below:

[0066] Table 1:

[0067] Text content Design Elements Emotional vocabulary Pairing relationship The DMS is very sensitive, and the wheels look great. DMS, wheel hub Sensitive and beautiful <DMS, Sensitive> Correct (1) The DMS is very sensitive, and the wheels look great. DMS, wheel hub Sensitive and beautiful <Wheels look good> Correct (1) The DMS is very sensitive, and the wheels look great. DMS, wheel hub Sensitive and beautiful <DMS, Looks Good> Error (0) The DMS is very sensitive, and the wheels look great. DMS, wheel hub Sensitive and beautiful <Hub, Sensitivity> Error (0)

[0068] After training, the keyword pairs of "design element" and "emotional vocabulary" are extracted from the original text, and context-enhanced word vectors are obtained through BiLSTM. Finally, these two vectors are concatenated as input to the MLP to predict whether the pair constitutes a semantically correct relation (1 or 0). Each target keyword pair includes a target design element and a target emotional vocabulary.

[0069] This embodiment can automatically identify "design element - emotional vocabulary" pairs in comment text and extract users' emotional evaluations of specific vehicle parts (such as headlights, grilles, wheels, etc.). It achieves accurate acquisition of users' emotional and aesthetic reactions to the car's appearance from unstructured text, significantly improving the accuracy and efficiency of information extraction.

[0070] Step 14: Obtain attention features based on the comment keywords;

[0071] In this embodiment, attention features are determined based on the frequency of the target design elements appearing in the target comment data. Here, the target design elements can be discretized into 5 categories according to the style characteristics of the design parts: wheels, grille, body, headlights, and taillights.

[0072] Specifically, the importance of design element a (e.g., wheel hub) in comment c is evaluated using attention frequency-inverse corpus frequency (AF-ICF), calculated as follows:

[0073] ;

[0074] in, The frequency of design element 'a' in a certain comment 'c'; N is the total number of comments; The total frequency of mention of design element 'a' in all comments.

[0075] To handle references to the same design element in different expressions (such as "front fascia" and "headlights"), two rules were established to summarize word frequencies and calculate attention values:

[0076] Rule 1: If a term explicitly refers to a design element (such as "grille"), only that element is counted;

[0077] Rule 2: If the term is vague but refers to a certain area (such as "front face"), then multiple related components (such as "grille" and "headlights") are counted together.

[0078] Step 15: Based on the target keyword pair, the attention features, and the user group type, obtain a user emotional needs profile;

[0079] In this embodiment, the emotional dimension score of the focus feature is determined based on the target keyword pair; wherein the emotional dimension score of the focus feature includes multiple categories of emotional dimension word scores, and the emotional dimension words are determined by the frequency of occurrence of emotional words in the target comment data.

[0080] Based on the frequency of emotional terms and their presence in reviews, word frequency analysis can identify the top 6 high-frequency emotional dimension words (high-end, smooth, stylish, technological, elegant, and sporty) as emotional dimensions for each design element of the vehicle. However, not all emotional dimension words appear in every user review, so user emotional scores must be calculated using specified rules. Vectorizing words using word embeddings before entity recognition allows for the calculation of word relevance scores.

[0081] Define the following rules to calculate each user's emotional score:

[0082] Rule 1: If a user review contains a clear emotional dimension word (e.g., "This car looks high-tech"), the "high-tech" score will increase by 1, while the scores for other emotional dimensions will remain unchanged.

[0083] Rule 2: If no emotive dimension words appear in the user review (e.g., "The car looks sublime," where "sublime" is not a selected emotive dimension word), then calculate the correlation between "sublime" and each emotive dimension word based on cosine distance, and sum the correlations as scores:

[0084] ;

[0085] in, For words in the comments, "N" represents the emotional dimension, and "n" represents the total number of comments.

[0086] In word vectors, cosine similarity is usually located in [0,1]. The closer it is to 1, the more similar the two words are.

[0087] Next, based on the emotional dimension scores of the features of interest and the user group type, a user emotional needs profile is obtained. Specifically, a structural equation model (SEM) is used to construct the user's emotional cognitive process in perceiving automotive exterior design, thus obtaining the user emotional needs profile. The path is: User features → Features of interest (users will pay more attention to certain components), Design features → Features of interest (certain design elements are more eye-catching), Features of interest → Emotional evaluation (attention influences perceived results). The path influence relationship of each emotional dimension is determined through model estimation. The emotional score assigned to each review represents the user's preference strength for each emotional dimension. Using features of interest as input variables, the cognitive level of users on six emotional dimensions (high-end feel, smooth feel, fashionable feel, technological feel, elegant feel, and sporty feel) is predicted, and the emotional dimension scores are output as output variables.

[0088] After obtaining the emotional dimension scores for each user, users were further grouped according to their car purchase purpose and vehicle type (new energy vehicle or gasoline vehicle). Each user was categorized into a specific subgroup, such as "new energy vehicle users for dating purposes" or "gasoline vehicle users for business purposes," thus creating conditions for comparing the emotional needs characteristics of different groups. For example, new energy vehicle users have a significant influence on the technological and fashionable feel of the vehicle in a dating (car purchase purpose) scenario.

[0089] After grouping, the scores for each user group across six emotional dimensions are aggregated. By calculating the average score for each user group on each emotional dimension, a profile of the group's emotional needs is created.

[0090] This embodiment uses structural equation modeling to model the causal relationships between user characteristics, design characteristics, attention characteristics, and emotional dimensions. It can quantitatively reveal the influence paths of different design factors on users' multi-dimensional emotional images such as "high-end," "technological," and "fashionable," providing a scientific basis for emotional design decisions. By combining car purchase purpose, purchase price, and regional information to create user profiles and groups, it analyzes the differences in preferences of different groups in the emotional dimension, effectively supporting scenario-based customized design strategies and improving user satisfaction and product competitiveness.

[0091] Step 16: Optimize the car design based on the user's emotional needs profile.

[0092] In this embodiment, the results of user emotional needs profiling constitute the data for drawing radar charts, which can reflect the overall preference levels of different groups in terms of emotional needs. For example... Figure 2 As shown, the perceptual mean data for each group are plotted as a radar chart to graphically illustrate the differences in perceptual evaluations between new energy vehicle and gasoline vehicle users under different purchase purposes. This intuitively demonstrates the differentiated perceptions of different user groups regarding "fashion," "premium," and "sporty" aspects, providing empirical evidence for product design and market positioning.

[0093] This embodiment uses radar charts to visually represent the user group's perceptual profile, facilitating designers' quick understanding of the target users' aesthetic expectations and enhancing the interpretability and practicality of the perceptual needs analysis results in the design process. This strengthens the intelligence and data-driven capabilities of product design.

[0094] This invention presents an optimization method for automotive design based on emotional needs. It systematically mines the pairing relationships between design elements and emotional terms in user reviews, combines user background and attention characteristics to construct a structural model of user emotional cognition, and visually presents the differences in emotional needs among different user groups under different car-buying purposes using radar charts and other formats. This provides a scientific and data-driven support method for automotive exterior design. By transforming user language behavior into structured modeling input, this invention realizes the transformation of emotional design from "experience-driven" to "data-driven," providing a replicable and scalable systematic method for optimizing intelligent vehicle exterior design and improving user experience.

[0095] An embodiment of the present invention also provides a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method described in the above embodiments. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0096] In this embodiment of the invention, a computer-readable storage medium is also provided, storing instructions that, when executed on a computer, cause the computer to perform the method described in the above embodiments. All implementations of the methods described in the above embodiments are applicable to this embodiment and can achieve the same technical effect.

[0097] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0098] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0099] In the embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0100] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0101] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0102] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0103] Furthermore, it should be noted that in the apparatus and method of the present invention, it is obvious that the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered equivalent solutions of the present invention. Moreover, the steps performing the above series of processes can naturally be executed in the order described, but are not necessarily required to be executed in chronological order; some steps can be executed in parallel or independently of each other. Those skilled in the art will understand that all or any step or component of the method and apparatus of the present invention can be implemented in any computing device (including processors, storage media, etc.) or network of computing devices, in hardware, firmware, software, or a combination thereof. This is something that those skilled in the art can achieve by using their basic programming skills after reading the description of the present invention.

[0104] Therefore, the object of the present invention can also be achieved by running a program or a set of programs on any computing device. The computing device can be a known general-purpose device. Therefore, the object of the present invention can also be achieved simply by providing a program product containing program code implementing the method or apparatus. That is, such a program product also constitutes the present invention, and the storage medium storing such a program product also constitutes the present invention. Obviously, the storage medium can be any known storage medium or any storage medium developed in the future. It should also be noted that in the apparatus and method of the present invention, it is obvious that the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered equivalent to the present invention. Furthermore, the steps performing the above series of processes can naturally be performed in the order described, but are not necessarily required to be performed in chronological order. Some steps can be performed in parallel or independently of each other.

[0105] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for optimizing automobile design based on emotional needs, characterized by, include: Obtain target review data about cars; Based on the target comment data, the comment keywords and user group types are obtained; The matching relationship of the comment keywords is verified to obtain the target keyword pairs; Based on the keywords in the comments, attention characteristics are obtained; Based on the target keyword pairs, the attention characteristics, and the user group types, a user emotional needs profile is obtained; Based on the user's emotional needs profile, optimize the car design.

2. The method for optimizing a car design based on emotional needs according to claim 1, wherein, Obtain targeted review data about cars, including: Scrape user review data about cars from online platforms; After preprocessing the user review data, entity annotation is performed to obtain target review data about cars. The target review data includes user ID, car model name, purchase price, purchase time, purchase IP, purchase purpose, review time, and review text content.

3. The method for optimizing a car design based on emotional needs according to claim 2, wherein Based on the target review data, the user group types are obtained, including: Based on the purchase price, the IP address of the purchaser, and the purpose of the purchase, the user group type is determined.

4. The method for optimizing a car design based on emotional needs according to claim 2, wherein Based on the target comment data, comment keywords are obtained, including: The comment text is input into a pre-trained text recognition model for entity recognition to obtain comment keywords; the comment keywords include design elements and emotional vocabulary.

5. The method for optimizing a car design based on emotional needs according to claim 4, wherein The matching relationship of the comment keywords is verified to obtain target keyword pairs, including: Randomly pair the design elements and emotional words in the keywords to obtain multiple pairing combinations; Multiple pairings are filtered to obtain target keyword pairs that conform to the preset pairing relationship. Each target keyword pair includes a target design element and a target emotional word.

6. The method for optimizing a car design based on emotional needs according to claim 5, wherein Based on the keywords in the comments, attention characteristics are obtained, including: Based on the frequency of the target design elements appearing in the target comment data, attention features are determined.

7. The method of claim 3, wherein the method further comprises: Based on the target keyword pairs, the attention characteristics, and the user group types, a user emotional needs profile is obtained, including: Based on the target keyword pairs, determine the perceptual dimension score of the features of interest; The emotional dimension scoring of the focus feature includes emotional dimension word scores of multiple categories, and the emotional dimension words are determined by the frequency of occurrence of emotional words in the target comment data; Based on the emotional dimension scores of the aforementioned attention features and the user group type, a user emotional needs profile is obtained.

8. The method for optimizing a car design based on emotional needs according to claim 7, wherein Based on the emotional dimension scores of the aforementioned attention features and the user group type, a user emotional needs profile is obtained, including: Calculate the average score of each user group type on each emotional dimension of the features they care about to obtain a profile of user emotional needs.

9. The method of claim 7, wherein the method further comprises: Based on the target keywords, determine the emotional dimension score of the features of interest, including: When the target emotional vocabulary belongs to the emotional dimension words, the score of the corresponding emotional dimension words of the focus feature is increased by one point, while the scores of other emotional dimension words remain unchanged. When the target emotional words do not belong to the emotional dimension words, the correlation between the target emotional words and each emotional dimension word is calculated based on the cosine distance, and the emotional dimension word score is determined based on the correlation. Based on the set of scores for multiple perceptual dimension words, determine the perceptual dimension score of the feature of interest.

10. The optimization method for automobile design based on emotional needs according to claim 4, characterized in that, Training the text recognition model includes: Word embedding is used to convert words in text comments into fixed-dimensional text vectors. The text vectors are input into the text recognition model, which captures the contextual semantic information of each text vector in the text sequence and concatenates them to obtain a comprehensive representation vector of each text vector in the context. This representation vector is then input into the CRF layer, which outputs words with pre-defined annotations.

Images