Design of panoramic virtual museum interactive interface based on entropy weight TOPSIS and PSO-SVR | npj Heritage Science

npj Heritage Science volume 13, Article number: 201 (2025) Cite this article

381 Accesses

Metrics details

This study addresses the neglect of user emotional needs in digital cultural heritage displays by proposing a quantitative methodology to optimize the Panoramic Virtual Museum Interactive Interface (PVMUI). Leveraging Kansei Engineering, we first extracted three core affective dimensions (Technological, Innovative, Fancy) through factor analysis of 45 Kansei terms. Morphological decomposition and entropy-weighted TOPSIS objectively prioritized six critical design features from 14 candidates. A nonlinear mapping model integrating Particle Swarm Optimization Support Vector Regression (PSO-SVR) was developed to correlate these features with user emotions, validated empirically via 100 participants. Results identified an optimal configuration: top-bottom layout (A1-1), 30% screen occupancy (A2-3), transparent navigation icons (A5-5), planar maps (A7-3), arrow-shaped movement icons (A9-7), and no scene interaction (A10-6). The framework advances virtual museum design by establishing emotion-driven optimization principles, bridging theoretical gaps in affective computing for PVMUI while offering actionable guidelines for enhancing user immersion and heritage dissemination.

With the passage of time, the function of museums has gradually shifted from collecting and storing artifacts to engaging in a unique form of cultural exchange with society. This transformation has made museums one of the most popular scientific institutions outside of schools, and such cultural exchanges can bring positive benefits to society, economy, education, and other aspects1. With the continuous development of the so-called “Experience Economy,” technological innovation and the widespread use of the internet have made people more eager to obtain novel museum experiences. Traditional museum visiting methods can no longer satisfy the emotional needs of users2. There is an urgent need for more innovative, high-tech, and interactive museum visiting methods to facilitate cultural exchange with museums. In this context, the digitization of museum experiences is considered a highly anticipated solution3.

Accompanied by continuous exploration of digital technologies such as VR, AR, MR, panoramic photography, mobile platforms, and interactive animations in the field of museum exhibitions, scholars have combined technologies like computer graphics4 and artificial intelligence5. Starting from two approaches—improving offline and online museums—they have integrated digital technologies into museum exhibitions, promoting the frontier development of cross-disciplinary research in museum digital exhibitions. In the direction of improving offline museums, these technologies and methods have been widely applied in forms such as immersive museum VR installations6, AR-assisted museum experiences7,8, and personalized museum experiences9. These forms are all considered helpful in enhancing public satisfaction with offline museum exhibitions. In the direction of online museums, these technologies and methods have been applied in forms such as virtual exhibition systems10,11,12, serious games13,14, and Panoramic Virtual Museums (PVMs)15. These forms can all provide the public with more convenient, interesting, and immersive online museum experiences.

As digital museum exhibition technologies continue to mature, museums are increasingly integrating diverse digital methods into their exhibitions—ranging from offline AR and VR interactions to interactive touchscreen digital displays, as well as online panoramic virtual museums and gamified virtual museums. These technologies hold the potential to profoundly shape the future of museum experiences. Virtual digital museums provide the public with more immersive and convenient museum viewing experiences, meeting the public’s diverse cultural, educational, and emotional experience needs for museum visits, and promoting sustainable social and economic development.

As museum digitization projects continue to advance, online virtual museums - which include immersive virtual reality exhibitions and interactive online platforms - are receiving increasing attention and recognition from the public and museum institutions. However, in academia, there is still no unified definition of “virtual museum,” but it is generally agreed that its main feature is to provide the public with a three-dimensional virtual museum experience through an interactive interface on the internet. The advantages of Virtual Museums can be summarized as:

Visiting without leaving home: Visitors can enjoy museum exhibitions and services without being physically present, saving time and economic costs.

Simple and easy interactive operation: A user-friendly interactive interface and intuitive operation methods allow users to get started easily, enhancing the convenience of the visiting experience.

Relatively low construction cost: Compared with physical museums, Virtual Museums have lower construction and maintenance costs, higher resource utilization efficiency, and are conducive to the widespread dissemination of museum resources.

The Panoramic Virtual Museum (PVM) is a representative project among Virtual Museums and has been widely applied in various museums around the world, such as cultural museums16, and science museums17. It uses three-dimensional panoramic technology to scan and stitch real museum exhibitions into a large number of high-definition panoramic images, transmits them to the internet through web plugins, and generates PVM scenes. The public can enter the interactive interface via web links18 to realize functions such as roaming exhibitions, point-to-point movement, voice explanations, and VR tours. Previous studies have shown that visitors regard panoramic virtual museums as a form of leisure activity, finding them both interesting and memorable. Zhang19 pointed out that virtual museums employ 3D panoramic, VR, and AR technologies to provide immersive experiences, allowing audiences to visit exhibitions without spatial constraints. In addition, as an innovative medium for knowledge dissemination, panoramic virtual museums can offer off-campus educational services for students. Aylana and Gökb’s research on Turkey’s Panorama 1453 Historical Museum indicated that its three-dimensional narrative mode facilitates the transmission of historical and cultural content while enhancing students’ interest in learning20.

Meanwhile, panoramic virtual museums can serve as specialized exhibition platforms for artifacts limited by real-world conditions, thereby generating cultural tourism benefits for local regions. In a VR game study involving the Terracotta Warriors, Zhang et al.21 noted that the combination of 360° panoramic photography and 3D modeling enables participants to gain deeper insights into the Mausoleum of the First Qin Emperor, while simultaneously promoting cultural tourism. Moreover, this form of presentation aids museums in preserving, restoring, and displaying ancient cultural heritage while allowing for public oversight. Doulamis et al.22 proposed that serious games enhance the dissemination of cultural heritage by providing interactive experiences through which audiences can better understand historical cultures.

Interaction between the public and virtual spaces remains a critical focus in VR research. Liarokapis et al.23 summarized six types of interfaces in AR, VR, and MR systems, suggesting that technological advancements in panoramic virtual museums optimize user experience and offer substantial prospects for cultural education and entertainment. However, Past research on PVMs often focused on how to create an online PVM but did not conduct related design research on the PVM Interactive Interface (PVMUI), ignoring users’ emotional needs. To bridge this gap, this study aims to develop a PVMUI design framework to enhance user interaction during exhibitions and promote the advancement and adoption of PVM technology.

In the study of users’ emotional design, KE is used as an effective translation method to convert users’ Kansei perceptions of products into design specifications that can be used by designers, thereby solving the imbalance between product output strategies and market input strategies caused by the increase in users’ Kansei needs. KE originated in the automotive industry and was first applied by Mazda in the production of its “Persona” model24, which was released in November 1988 and achieved excellent sales. In the subsequent nearly 40 years of development, KE has been applied in various design fields, such as logistics service design25, space design26, community service design27, clothing design28, and numerous industrial product design fields29,30,31. As a rigorous design framework, KE can be divided into three stages: (1) determining users’ perception of the design object, (2) determining key design elements, and (3) constructing a mapping model between user perception and key design elements. In this process, researchers can use various types of mathematical models and computer reasoning systems to make the final design specifications more in line with users’ Kansei needs.

For the first stage, most researchers choose to provide product samples to users in the form of product pictures or sketches and obtain Kansei feedback32. Some researchers complete this step using new technologies such as virtual reality33. For the second stage, methods such as Analytic Hierarchy Process (AHP)34, Kano model35, Rough Set Theory (RST)36,37, Grey Relational Analysis (GRA)38, etc., are applied. These methods can be used to determine the design elements with higher weights among the design elements, which helps researchers reduce the computational cost of the subsequent mapping model. For the third stage, since the relationship between users’ Kansei perception and design elements is often a nonlinear function, traditional linear statistical analysis (such as factor analysis) cannot perform this task. Therefore, artificial intelligence technologies such as Backpropagation Neural Network (BPNN)39 and Genetic Algorithm (GA)40 are widely used in this stage.

This study proposes a systematic PVMUI design method based on the Kansei Engineering framework, integrating the entropy weight method, TOPSIS, and PSO-SVR to enhance users’ emotional experience and sense of immersion. With the rapid digitization of cultural heritage and museums, PVMs have become an essential avenue for cultural transmission. However, existing research on PVMUIs primarily focuses on technological implementation while overlooking users’ emotional experiences, leading to lower user engagement and weaker emotional connection with online museums. To date, no studies have applied the Kansei Engineering framework to optimize PVMUI design. Therefore, the core objective of this research is to establish a scientific approach capable of quantifying users’ emotional needs and optimizing PVMUI design, thereby improving the user experience of virtual museums and filling a gap in the literature.

Following the Kansei Engineering approach, the study first identifies core user emotions and then employs the entropy-weighted TOPSIS method to determine key design features. Next, a PSO-SVR model is used to construct a nonlinear mapping between user emotions and these key features, enabling accurate predictions of optimal interface designs. Experimental results confirm the effectiveness of this method for enhancing users’ emotional experience and informing PVMUI design decisions.

By providing scientific decision support for virtual museum designers, the findings of this research can optimize PVMUI to boost immersive online viewing and interactive experiences. Additionally, this study expands virtual museums’ display forms, attracting the public to explore museums online and improving digital dissemination. It also meets the public’s emotional needs for PVMs, enriches the visual experience of virtual exhibitions, and facilitates the continued development and innovation of cultural heritage digitization, promoting broader dissemination and inheritance of cultural heritage.

The innovations of this paper can be summarized as follows:

Combining the Entropy Weight TOPSIS and PSO-SVR methods for design research. The integration of these two methods provides a new approach for the nonlinear mapping between user emotions and design elements.

Offering new perspectives and methods for the field, contributing to the advancement of theory and practice in virtual museum interactive interface design.

Using the Entropy Weight TOPSIS method to screen interactive interface design elements, determining key design elements, and reducing research costs. Through an objective weighting method, the design elements that have the most significant impact on user emotions are precisely extracted, improving the efficiency and effectiveness of the research.

Employing PSO-SVR to replace traditional linear statistical analysis, establishing a nonlinear mapping model between user perceptions and design elements. By optimizing the parameters of the SVR model using the PSO algorithm, the prediction accuracy of the model is enhanced, overcoming the limitations of traditional methods in nonlinear relationship modeling.

The structure of the paper is as follows: “Methods” provides an overview of the main methods used in this paper; “Research Framework and Process” elaborates on the research framework of the PVMUI design proposed in this paper and introduces the specific research process; “Results and Discussion” analyzes and discusses the research results; “Conclusion and Future Work” summarizes the content of this paper and proposes future research prospects.

With the continuous improvement of public living standards, consumer demands have shifted from merely functional value to emotional value. In 1986, Mazda first introduced the term “Kansei Engineering,” which was further developed in 1995 by Mitsuo Nagamachi at Hiroshima University into a concrete product development technique24. Kansei Engineering aims to explore the relationship between consumers’ intangible emotional needs and the tangible elements of product design41.

In recent years, KE has been widely applied in interactive interface research, particularly in evaluating and optimizing user experience. Table 1 clearly summarizes the research methods, techniques, and outcomes related to applying KE in interactive interface design. Despite various existing KE-based studies on interface evaluation or design, no scholars have paid attention to the field of PVM, and there is a lack of design studies on the use of complex integrated models to build mapping models between user perceptual factors and interactive interface design elements. To address this gap, this study selects a web-based PVMUI as the research object and constructs a scientific and rational KE-TOPSIS-SVR integrated model to optimize existing PVMUIs. This approach aims to strengthen digital museum exhibitions and further promote PVM in the field of cultural heritage protection.

To determine the weight of each design feature in PVMUI, we apply the Entropy Weight TOPSIS method. The Entropy Weight Method (EWM) is an objective weighting approach based on information theory, which calculates entropy values to measure the amount of information in the dataset. By avoiding subjective bias, EWM effectively reflects the weighting of user perceptions42,43. In Kansei Engineering research, EWM is often used to determine the weights of design elements. For example, Zou et al. employed EWM to assign weights to lychee quality indicators and used a linear weighted sum method to evaluate changes in overall shelf-life quality42. Wang et al. integrated EWM with an improved TOPSIS approach, constructing a decision-making system that combines subjective and objective evaluations to provide ergonomic optimization solutions44.

TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) is one of the multi-criteria decision-making (MCDM) methods, which ranks alternatives by calculating their distances to an ideal solution45,46. Its advantages include intuitive logic, computational efficiency, and suitability for multi-attribute decision-making47,48. This method has been extensively applied in industry and design. For instance, Lin et al. combined neural networks with TOPSIS to develop a Kansei Engineering decision support model for optimizing toy designs49,50. Wang et al. introduced a KE-TOPSIS-AISM method that uses TOPSIS for partial-order comparisons of product styles and incorporates AISM to establish a topological hierarchy, enhancing the optimization of product appearance design51. Furthermore, Quan et al. fused EWM, the Analytic Hierarchy Process (AHP), game theory, and Grey Relational Analysis (GRA) to propose a KE-GRA-TOPSIS approach, demonstrating its unique advantages in Kansei evaluation52.

In summary, the Entropy Weight TOPSIS method can objectively determine the weights of design elements in multi-criteria decision scenarios. It extracts key features, discards low-weight factors, reduces computational costs, and provides robust support in Kansei Engineering for identifying essential design elements.

The calculation process of the Entropy Weight TOPSIS is as follows.

The EWM is an objective weighting method. It uses the information entropy of the data itself to calculate weights and obtain objective weighting results. The calculation process is as follows:

Step 1: Construct the decision matrix. Suppose there are m elements and n expert scores, the evaluation decision matrix is as in Eq. (1):

Step 2: Calculate the normalized standardized matrix. The normalized matrix \({\boldsymbol{Q}}\) is formed by taking the ratio of each element in the standardized matrix to the sum of all elements in its column vector. The calculation process is as shown in Eqs. (2) and (3):

where \({q}_{ij}={x}_{ij}/\mathop{\sum }\nolimits_{i=1}^{m}{x}_{ij}^{2}\)

Step 3: Calculate the information entropy and weights of each risk evaluation indicator. Use Eq. (4) to obtain the information entropy \({E}_{j}\) of NO j, and use Eq. (5) to obtain the weight \({w}_{j}\) of NO j:

where \(\lambda =\frac{1}{\mathrm{ln}\,n}\), \({p}_{ij}=\frac{{x}_{ij}^{{\prime} }}{{\sum }_{i=1}^{n}{x}_{ij}^{{\prime} }}\), \(i=1,2,\cdots ,n,j=1,2,\cdots ,m\), Where \(0\le {w}_{j}\le 1\) and \(\mathop{\sum }\nolimits_{j=1}^{m}{w}_{j}=1\)。

TOPSIS is a decision-making method widely used for multi-attribute problems across various industries. The calculation process is as follows:

Step 1: Construct the weighted decision matrix \({\boldsymbol{B}}\) based on the indicator weights.

where \(({b}_{ij})\) is calculated according to Eq. (6):

Step 2: Determine the positive ideal solution \({{\boldsymbol{U}}}^{+}\) and the negative ideal solution \({{\boldsymbol{U}}}^{-}\) based on the weighted decision matrix \({\boldsymbol{B}}\), as shown in Eqs. (7) and (8):

Step 3: Based on the positive and negative ideal solutions, determine the weighted Euclidean distances between the PVM design features and the ideal solutions, as shown in Eqs. (9) and (10):

Step 4: Calculate the relative closeness coefficient of each target and rank them, as shown in Eq. (11):

By sorting the \({v}_{i}\) values in descending order, a larger \({v}_{i}\) indicates a higher importance of the evaluation indicator. Based on the decision results, the design features are ranked according to their importance. Several PVM design features with high importance are selected, effectively extracting design features that have a significant impact on users’ Kansei needs.

In Kansei Engineering research, the relationship between Kansei words and design elements is often nonlinear, making it difficult for traditional linear methods to accurately map these associations. Consequently, neural network techniques have been widely adopted for constructing such mapping models. Among them, Support Vector Machine (SVM) has gained attention for its “kernel” technology, which effectively addresses nonlinear problems53. Support Vector Regression (SVR), an extension of SVM, performs nonlinear regression analysis for optimizing function estimation54,55. Studies have shown that SVR outperforms traditional neural networks in various applications. For instance, Yang et al. demonstrated that SVR yielded better predictive performance than BPNN in experiments involving mobile phone and electric vehicle datasets56.

Particle Swarm Optimization (PSO) is a global optimization algorithm that simulates the search behavior of particle swarms to find optimal solutions in a given search space57. In this research, we employ PSO to optimize SVR parameters, thereby constructing a mapping model between the core design features of the PVMUI and user perceptual needs. This approach enhances both the accuracy and efficiency of design optimization.

SVR has also been widely used in previous design studies. For example, Yang et al. combined SVR with a multi-objective genetic algorithm (MOGA) to develop a hybrid Kansei Engineering system (HKES)58. Yuan et al. integrated LDA (Latent Dirichlet Allocation) and rough set theory (RST) to optimize medical care bed design59. Shieh et al. proposed a method merging SVR with a multi-objective evolutionary algorithm (MOEA), which they validated in a vase design context60. Although SVR has been applied in industrial and product design, it has yet to be adopted for interface design. Therefore, this study utilizes PSO-SVR to build a mapping model that links the core design features of the PVMUI to users’ perceptual needs.

The following content briefly introduces the basic principles and calculation methods of PSO-SVR.

SVR maps the input sample vectors into a high-dimensional feature space using a nonlinear function \(\phi (x)\), thereby approximating a linear function in that feature space. The linear regression function of SVR can be expressed as Eq. (12):

In Eq. (12), \(\phi ({x}_{i})\) is the nonlinear mapping function that maps the sample data into the feature space, \(\omega\) is the coefficient of the independent variable, and \(b\) is the bias term. \(\omega\) and \(b\) are obtained by minimizing the following Eq. (13):

In Eq. (13), \(\min \,f(\omega ,{\xi }_{i}^{\ast },{\xi }_{i})\) is the generalized optimal classification hyperplane function that considers the minimum misclassified samples and the maximum classification margin; \(\frac{1}{2}{\Vert \omega \Vert }^{2}\) is the regularization term; \(n\) is the total number of samples; \(C\) is the penalty factor; and \({\mathcal E}\) is the insensitive loss function (error control function). To find suitable constraint conditions, the original problem can be minimized by introducing the upper and lower slack variables. By introducing the upper and lower slack variables \({\xi }_{i}^{\ast }\) and \({\xi }_{i}\), the original problem is minimized.

Using the Lagrangian equation and dual theory, by setting the partial derivatives of \(\omega\), \(b\), \({\xi }_{i}^{\ast }\) and \({\xi }_{i}\) to zero respectively, the dual form is obtained, as shown in Eq. (15):

Where \(K({x}_{i},{x}_{j})\) represents the kernel function, and \(a,{a}^{\ast }\) denotes the Lagrange multipliers. Based on the above dual calculation equation, the regression function can be obtained as shown in Eq. (17):

The basic concept of PSO can be summarized as particles moving continuously in an n-dimensional space, where each particle continuously adjusts its position and velocity based on its own experience or that of neighboring particles. The functional equations can be expressed as Eqs. (18) and (19):

where \(i=(1,2,\cdots ,n)\)\(j=(1,2,\cdots ,D)\),\(n\) represent the number of particles; \(D\) represents the dimension of the search space; \({x}_{ij}\) represents the position of the particles; \({v}_{ij}\) represents the velocity of the particles; \(w\) represents the inertia weight; \({c}_{1}\) and \({c}_{2}\) represent the learning factors; \({R}_{1}\) and \({R}_{2}\) are random numbers between [0, 1]; \({p}_{i}\) represents the best position of particle \(i\); and \({p}_{g}\) represents the best position among all particles \(i\) in the entire group.

The purpose of this study is to develop an interactive interface design model for PVMs under the global background of the virtualization and digitization of museums and cultural heritage. The research method combines qualitative and quantitative approaches. First, PVM websites were retrieved through web searches to screen and obtain a sample set of PVM interactive interfaces. By releasing an online interactive interface survey questionnaire, 45 Kansei words targeting PVMs were obtained, and 10 Kansei words representing user perceptions were identified through the KJ method (Affinity Diagram Method). After an initial sorting of the sample set, 12 PVM interactive interface design elements were derived, which were reduced to 6 key design elements based on the Entropy Weight TOPSIS method. An SVR model was used to establish a mapping between user perceptions and key design elements, ultimately obtaining the optimal design combination for PVMs. Based on this result, a design practice was conducted, and a satisfaction evaluation of the practice results was performed to verify the feasibility of the design process. Figure 1 shows the specific research steps.

This figure presents the proposed research framework for optimizing the Panoramic Virtual Museum Interactive Interface (PVMUI). The framework consists of three phases: Phase 1: Image Collection and Factor Analysis Extraction of Kansei Words • PVMUI images are gathered from various online sources. • Kansei words are collected through online questionnaires and screened. • The KJ Method is used to filter Kansei words, followed by Factor Analysis (FA) Clustering to extract core emotional dimensions (Technological, Innovative, and Fancy). Phase 2: Entropy Weight-TOPSIS Morphological Deconstruction and Extraction. • The design features of PVMUI are deconstructed. • The Entropy Weight-TOPSIS method is applied to identify and extract critical interface design features based on user perception. Phase 3: PSO-SVR Mapping of Kansei Words and Design Features. • Particle Swarm Optimization (PSO) is used to optimize Support Vector Regression (SVR) parameters. • The SVR model is then used to predict the mapping between user emotions and design features, refining the optimal PVMUI configuration.

This study selects the content of the interactive interface that first appears after entering the web link of a PVM as the design sample. By accessing the official websites of various provinces, cities, and specific cultural relic museums, more than 90 interactive interface design samples that meet the research content were extensively collected. To avoid the influence of different exhibited artifacts within PVMs on user decisions, the viewpoints are oriented as much as possible towards non-exhibition areas (such as walls, corridors, etc.). Through expert group interviews, issues that may affect user perception, such as duplicate sample forms and blurred interactive interfaces, were resolved. Representative PVM interactive interfaces were selected from the original sample set, and an interactive interface sample set containing 60 design samples was established (see Fig. 2) for subsequent analysis.

This figure displays 60 representative Panoramic Virtual Museum Interactive Interfaces (PVMUI) collected from various online sources. These samples include diverse interface layouts, color schemes, navigation styles, and interaction elements, reflecting a broad spectrum of virtual museum design approaches. The dataset was curated by ensuring: • A variety of museum types (e.g., historical, cultural, science museums). • Standardized perspectives that minimize content bias and focus on interface design.

During the Kansei vocabulary collection phase, based on the previously constructed set of 60 design samples, a user perception evaluation questionnaire targeting the PVMUI was distributed to focus groups, collecting a total of 1120 Kansei evaluations. At the same time, a panel of six experts was established, consisting of two designers, two experts in Kansei Engineering, and two representative users with experience in virtual museum interaction. After removing evaluations that did not conform to the form of Kansei words, were repetitive, or had significant semantic errors, 45 qualified Kansei words were selected. Using the KJ method (Affinity Diagram), these 45 Kansei words were organized and summarized into ten Kansei word groups. After discussion within the expert panel, a Kansei word that matched the content of each group was assigned, ultimately resulting in ten of the most representative Kansei words (see Table 2).

Subsequently, questionnaires were administered to 20 design students, 20 designers, and 20 potential users of PVMs. They rated all 60 design samples using the ten categories of Kansei words. The questionnaire employed a five-point Likert scale, where 5 represented “highly consistent,” 3 represented “moderately consistent,” and 1 represented “not consistent.” A total of 53 valid questionnaires were collected. To ensure the objectivity of the questionnaire, the average result corresponding to each sample and the ten Kansei words were calculated, constructing a Kansei semantic evaluation matrix (see Table 3 and Supplementary Data 1). This paper uses factor analysis to reduce the dimensionality of the Kansei semantic evaluation matrix. The matrix was imported into IBM SPSS Statistics 27, and the structural reliability and validity of the factor analysis scale were verified through KMO and Bartlett’s tests. The experimental results showed that the KMO measure of sampling adequacy was 0.890, and Bartlett’s test of sphericity yielded an approximate chi-square value of 464.327, degrees of freedom of 45, and a significance less than 0.001, meeting the standards for data suitability for factor analysis (KMO value greater than 0.5, Bartlett’s test of sphericity less than 0.05). Therefore, it was determined that the data have statistical significance and are suitable for factor analysis (see Table 4). In the total variance explained obtained by principal component analysis, the cumulative contribution rate of the first three factors was 79.805%, close to 80%, proving that the ten Kansei words can be reduced to three representative Kansei words (see Table 5). The ten Kansei words were orthogonally rotated using the Kaiser normalization maximum variance method. To reduce interference, factor loadings with absolute values less than 0.5 were not displayed (see Table 6).

Finally, the component score coefficient matrix was obtained (see Table 7). Three Kansei word groups were derived from the ones targeting the PVMUI and were named based on the rotated component matrix scores. The first word group was defined as “Technological,” consisting of five words: “immersive”, “Trendy”, “mysterious”, “technological”, and “solemn”, with “technological” having the highest score of 0.843. The second word group consisted of “warm,” “Minimalistic,” and “novel,” with “warm” having the highest score of 0.878. Since “warm” as a Kansei word is too broad in meaning, after expert panel discussion, this word group was named “Innovative.” The third word group consisted of two Kansei words: “Fancy” and “esthetic,” with “Fancy” having a higher score of 0.792. Therefore, this word group was defined as “Fancy.”

We extracted design elements from the 60 design samples in the sample set. To ensure the representativeness and objectivity of the extraction results, a panel of four experts in the field of interaction design was invited to form an expert group. After group discussions and morphological analysis, the PVMUI was decomposed into 12 design elements: overall layout style, interface content occupying screen size, function bar position, function bar image style, navigation bar icon color, navigation bar text color, map form, map position, movement icon style, scene interaction style, scene selection area position, and scene selection area form. Each design element can be further subdivided, totaling 64 types (see Fig. 3).

This figure presents the morphological deconstruction of Panoramic Virtual Museum Interactive Interface (PVMUI) design features. The table categorizes different design elements and their corresponding types, which serve as the basis for feature extraction in the Entropy Weight-TOPSIS analysis. Key design features include: • Overall Layout Style. • Proportion of Interface Content on Screen. • Function Bar and Navigation Bar. • Map Format and Position. • Moving Icon Styles. • Scene Interaction and Selection Area.

Since different design elements in the PVMUI have different perceptual weights for users, design features with lower or even no influence may lead to inaccurate subsequent results. Therefore, this paper uses the Entropy Weight TOPSIS method to perform dimensionality reduction and screening on the 12 design elements to determine the design elements with higher weights among them.

First, we recruited an expert group consisting of five design students, two Kansei Engineering experts, and two designers, all of whom hold master’s or doctoral degrees in design. Using a combination of online and offline questionnaire evaluations, the 12 design elements were compared in detail with the three representative Kansei words selected earlier for evaluation and scoring. A five-point Likert scale was used, where 5 represented “very important for this word,” 3 represented “generally important for this word,” and 1 represented “very unimportant for this word.” A total of 9 questionnaires were collected. By calculating the average value of each target, a mean matrix of design elements corresponding to the representative Kansei words was constructed(Supplementary Data 2). According to Eq. (1), the mean matrix was normalized to obtain the standardized matrix. Using Eqs. (2)–(5), we calculated the entropy values and weight coefficients of the three Kansei words, obtaining the information entropy values, information utility values, and weight values of the Kansei words (see Table 8).

By integrating the weight coefficients of the Kansei words for the PVMUI into TOPSIS, there were 12 evaluation objects, each with 3 Kansei evaluation attributes. According to Eq. (6), each attribute of the morphological components was vector normalized to form a decision matrix. Using Eqs. (7)–(11), we calculated the positive ideal solution (D+) and negative ideal solution (D-) for the 12 design elements, as well as the distance between each indicator and the best and worst situations (see Table 9) and obtained the comprehensive score index. We selected design elements with a comprehensive score index greater than 0.6 as key design elements. Therefore, through the Entropy Weight TOPSIS method, we identified 6 key design elements that have a significant impact on user perception of the PVMUI: overall layout style, interface content occupying screen size, navigation bar icon color, map form, movement icon style, and scene interaction style.

After performing factor analysis for dimensionality reduction on the Kansei words in the earlier stage, the three representative Kansei words (“Technological,” “Innovative,” and “Fancy”) were introduced into 60 PVMUI samples. A 5-point Likert scale was constructed, and a total of 100 subjects were recruited to participate in the evaluation. The subjects included a small number of designers with a background in design studies and users with experience using virtual museum interactive interfaces. Combined with the key interactive interface design features obtained from the Entropy Weight TOPSIS ranking, a sample-key design feature-representative Kansei word matrix was constructed, as shown in Table 10 and Supplementary Data 3.

At this stage, we used the PSO-SVR model to construct a mapping model between Kansei words and design features. By using the Matlab® software package, PSO-SVR was used to learn the matrix data in Table 10. Randomly selected 48 samples were used as the model’s training set, and the remaining 12 samples were used as the test set. The key design features were used as input of feature values, and the evaluation values from 100 participants were used as outputs. According to Eqs. (18) and (19), the PSO settings were determined. The PSO size was set to 20, the learning factors were both set to 2, and the number of iterations was set to 20 to optimize the SVR accordingly. The optimized SVR was used to establish the mapping relationship between Kansei words and design elements, as per Eqs. (12)–(17). Both the test set and the training set showed strong convergence, and the fitting graphs of the Kansei words are shown in Fig. 4. Good parameter results were obtained (training set: \({R}^{2}=\) 0.243, \(MSE=\)0.035; test set: \({R}^{2}=\)−0.140, \(MSE=\)0.056) (see Table 11, Supplementary Data 5 and Supplementary Data 6), indicating that the model used in this study has good predictive ability and high fitting degree, and can be used to establish the mapping relationship between user emotions and PVMUI design elements.

This figure presents the comparison between predicted and actual values of the PSO-SVR model for the three Kansei dimensions: Technological, Innovative, and Fancy. The evaluation is conducted on both the training set (top row) and the test set (bottom row). Training Set (Top Row): • Each graph compares the predicted (solid lines) and real (dotted lines) values across 45 samples. • The model shows varying accuracy across the three Kansei dimensions, with Fancy having the closest alignment between predicted and actual values. Test Set (Bottom Row): • Predictions for the Technological and Innovative dimensions show a small deviation from actual values, while the Fancy dimension maintains a relatively stable fit. • These results demonstrate the model’s capability in predicting user emotional responses to PVMUI design features, validating its effectiveness for design optimization.

In order to obtain the combination of PVMUI features with the highest user recognition, this study used the KJ method and factor analysis to collect and reduce the dimensionality of users’ perception needs Kansei words, obtaining the three reduced Kansei words: “Technological,” “Innovative,” and “Fancy.” Through the Entropy Weight TOPSIS method, 12 design features were extracted to 5 key design features. Each design feature has 7, 5, 5, 4, 9, 6 categories respectively, and their combinations can yield 7 × 5 × 5 × 4 × 9 × 6 = 37,800 design combination schemes (Supplementary Data 4). The PSO-SVR algorithm was used to calculate the corresponding values of the three Kansei words for each design combination scheme. The three corresponding values were summed to obtain the corresponding value of each design combination scheme as the user perception evaluation standard. Through calculation of the results, the maximum corresponding value of the design combination scheme was 10.0457, corresponding to the design feature categories of 1, 3, 5, 3, 7, 6. This is the optimal PVMUI design scheme.

To demonstrate the applicability of the method proposed in this paper, the author used the optimal PVMUI design scheme obtained from the experiment as a reference. Specifically, the design features selected were a1-1 (overall layout style with upper and lower distribution), a2-3 (interface content occupying 30% of the screen), a5-5 (transparent navigation bar icon color), a7-3 (plane-type map style), a9-7 (arrow-shaped movement icon style), and a10-6 (no scene interaction style added). Using 2D and 3D design software such as Adobe Photoshop 2020, Adobe Illustrator 2020, SketchUp Pro 2022, and Enscape 3.2, the rendering images and interactive interface of the PVM were obtained, as shown in Fig. 5, and placed into usage scenarios as shown in Fig. 6.

This figure presents the interactive interface design for the Bai Tie-Dye Digital Museum, developed based on the optimized Panoramic Virtual Museum Interactive Interface (PVMUI) configuration identified in this study. The design integrates the optimal interface elements derived from Entropy Weight-TOPSIS and PSO-SVR analysis to enhance user engagement and emotional resonance. Top Image (Exterior View): Implements the top-bottom layout style (A1-1), ensuring a clear hierarchical interface structure. The interface content occupies 30% of the screen space (A2-3), providing a balanced visual experience. Transparent navigation bar icons (A5-5) and a planar-style map (A7-3) are employed to reduce cognitive load while maintaining effective wayfinding. Bottom Image (Interior View): The movement interaction follows the arrow-shaped movement icon style (A9-7), facilitating intuitive navigation. The scene maintains a non-interactive environmental setting (A10-6), reducing unnecessary distractions and improving focus on exhibition content.

This figure showcases the optimized Panoramic Virtual Museum Interactive Interface (PVMUI) in different usage environments, illustrating its visual adaptability across various settings. The interface maintains consistent usability and esthetic coherence, ensuring an engaging experience in diverse contexts.

This paper adopts the KE-Entropy Weight TOPSIS-SVR research method to determine the optimal PVMUI design combination scheme that meets users’ perceptual needs. Using Chinese PVMs as the sample set, ten Kansei words were collected and screened through questionnaires and the KJ method. Factor analysis was used for dimensionality reduction to obtain three key Kansei words: “Technological,” “Innovative,” and “Fancy.” By decomposing the morphology of 60 PVM samples, 12 design features and a total of 64 corresponding types were obtained. The Entropy Weight TOPSIS method was used to screen the design features, resulting in six high-weight design features: overall layout style, interface content occupying screen size, navigation bar icon color, map form, movement icon style, and scene interaction style. It was found that users pay more attention to interface layout, interactive icons, and map forms when experiencing PVMs. The six design features corresponded to the 60 design samples. Finally, the three Kansei words were introduced into the SVM to predict the design combinations that would obtain higher Kansei weights for users. The results showed that selecting types 1, 3, 5, 3, 7, and 6 in the six design features could obtain the optimal design combination scheme. The comprehensive Kansei score of this scheme was 10.0457, higher than the other 37,799 schemes.

Based on the key design features identified from the optimal design combination, this study refined the Panoramic Virtual Museum Interactive Interface (PVMUI) by emphasizing user interaction behavior and interface simplicity, thereby more effectively enhancing the user experience. To evaluate acceptance of the optimized interface, 100 participants were recruited, and a 7-point Likert scale was used to rate the interface shown in Fig. 5. The results indicate that the interface’s average rating is 5.19, which is significantly higher than the theoretical midpoint of 4.00 on the 7-point scale, suggesting a high level of user recognition for the optimized PVMUI.

These findings validate the effectiveness of the proposed approach in quantifying online museum users’ emotional needs and demonstrate that the optimized PVMUI aligns more closely with user preferences. Looking ahead, museum authorities or interface designers can refer to the optimal design solution proposed in this study to improve online museum interactive interfaces, thereby increasing user satisfaction and retention. This, in turn, will help advance the digital display and dissemination of museum cultural heritage.

In the field of Kansei Engineering, establishing a mapping model between user emotional preferences and design features is one of the core research areas. Traditional linear statistical methods, such as Quantification Theory Type I (QT-I)61 and Principal Component Analysis (PCA)62, exhibit significant limitations when dealing with highly subjective and uncertain emotional data. These methods cannot directly measure and quantify user emotional experiences and struggle to accurately capture the nonlinear characteristics of user emotional needs, thereby limiting their application in complex design tasks.

With the rapid development of artificial intelligence technologies, nonlinear modeling techniques such as Neural Networks (NN)63, Genetic Algorithms (GA)64, and Support Vector Regression (SVR) have been gradually introduced into KE to establish nonlinear mapping relationships between emotions and design features. Among them, Deep Convolutional Neural Networks (DCNN)65 perform excellently in image processing and pattern recognition but have high requirements for input image sizes and certain limitations in handling non-image data. Genetic Algorithms have been widely and successfully applied in engineering optimization problems, but their search efficiency significantly decreases when facing high-dimensional spaces and complex fitness functions, and they can only solve a limited number of optimal solutions.

In contrast, SVR performs exceptionally well in handling small samples, high-dimensional data, and nonlinear problems. Compared to traditional parametric statistical methods and neural network methods, SVR has stronger generalization ability and data adaptability. By finding the optimal hyperplane in a high-dimensional feature space, it can directly learn complex nonlinear relationships from known data and make accurate predictions. Therefore, when constructing mapping models between emotions and design features, SVR is more suitable for various types of data. Relevant studies have shown that SVR outperforms other methods in Kansei mapping performance, as detailed in Table 12.

Despite the demonstrated effectiveness of the proposed Kansei Engineering–based PVMUI design framework in terms of user ratings, several limitations remain in practical applications. These should be addressed in future work:

Real-World Constraints: The implementation of the PSO-SVR model requires technical support from trained professionals, which may be challenging for local museums with limited resources. In addition, existing digital infrastructure in museums may not fully support the proposed optimized interface design, thus preventing it from achieving its intended effectiveness when presented to users. Future studies could explore lighter-weight algorithmic designs and collaborate with museums to validate these approaches across different scenarios.

User Group Constraints: Although digital museum platforms are becoming more widespread, certain user segments—particularly the elderly and other populations affected by the digital divide—may be unfamiliar or uncomfortable with digital tools. Even though the optimized PVMUI prioritizes ease of use and user-friendly interactions, these groups might still encounter usage barriers, negatively affecting their engagement and recognition of virtual museum experiences. In the future, usability testing should be conducted to collect objective data (e.g., task completion time, error rates, navigation paths) and subjective feedback (e.g., satisfaction, immersion, perceived workload) to fully assess how well the interface design meets the needs of different user groups.

Design Details Constraints: The current PVMUI design primarily focuses on the visual and functional layout of the interactive interface. However, a more holistic enhancement of user interactions within PVM requires improvements in functionality, service quality, and feedback mechanisms to address diverse user needs.

Methodological Constraints: This study employed a mixed qualitative–quantitative approach, which may inevitably be influenced by subjective factors. Future research could incorporate high-precision affective computing tools, such as electroencephalography (EEG), galvanic skin response (GSR), and eye-tracking devices, to capture and analyze users’ emotional feedback more accurately.

To further refine the PVMUI design framework, this study addresses both user experience enhancement and the broader dissemination of cultural heritage in virtual environments. By integrating the Kansei Engineering (KE) framework, factor analysis was applied to extract representative Kansei words, and the Entropy Weight TOPSIS method was employed to identify key design features that significantly impact user experience. Subsequently, a PSO-SVR model was constructed to establish a mapping between user Kansei words and key design features. The validation study, involving 100 participants, confirmed that the optimized design combination aligns with users’ emotional preferences.

Building on these findings, this study makes several key contributions:

Under the research framework of KE, a design process based on KE-Entropy Weight TOPSIS-PSO-SVR is proposed. This method integrates Kansei Engineering, Entropy Weight TOPSIS and PSO-SVR, providing a new methodology for design research.

Currently, there is no design research on the PVMUI based on the KE framework. This paper fills the research gap in this field and offers a new perspective for the interactive interface design of virtual museums.

The Entropy Weight TOPSIS method was used to more objectively identify key design features that strongly affect user emotions. By objective weighting, the accuracy and reliability of key design feature extraction were improved.

SVR was adopted to replace traditional linear analysis methods, establishing a mapping model between Kansei words and key design features. By determining the optimal hyperplane, the optimal design combination scheme of the PVMUI was obtained, satisfying user experience needs and promoting the display and dissemination of museum cultural heritage.

All datasets generated and/or analyzed during the current study are included in the manuscript and its supplementary files. The full raw dataset necessary for interpreting, replicating, and building upon the findings reported in this article is available in the supplementary materials. These datasets can be accessed directly in the supplementary files section of the manuscript. For any further inquiries or access to additional data, please contact the corresponding author.

Vallance, E. & Roberts, L. C. From knowledge to narrative: Educators and the changing museum. Hist. Educ. Q. 38, 344–346 (1998).

Article Google Scholar

Radder, L. & Han, X. An examination of the museum experience based on Pine and Gilmore’s experience economy realms. J. Appl. Bus. Res. 31, 455 (2015).

Article Google Scholar

Proctor, N. Digital: Museum as platform, curator as champion, in the age of social media. Curator 53, 35–46 (2010).

Article Google Scholar

Miller, G. et al. The virtual museum: Interactive 3D navigation of a multimedia database. J. Vis. Comput. Animat. 3, 183–197 (1992).

Article Google Scholar

Dal Falco, F. & Vassos, S. Museum experience design: A modern storytelling methodology. Des. J. 20, S3975–S3983 (2017).

Google Scholar

Carrozzino, M. & Bergamasco, M. Beyond virtual museums: Experiencing immersive virtual reality in real museums. J. Cult. Herit. 11, 452–458 (2010).

Article Google Scholar

He, Z., Wu, L. & Li, X. R. When art meets tech: The role of augmented reality in enhancing museum experiences and purchase intentions. Tour. Manag. 68, 127–139 (2018).

Article Google Scholar

Sylaiou, S. et al. Exploring the relationship between presence and enjoyment in a virtual museum. Int. J. Hum. -Comput. Stud. 68, 243–253 (2010).

Article Google Scholar

Ardissono, L., Kuflik, T. & Petrelli, D. Personalization in cultural heritage: The road travelled and the one ahead. 22, 73–99 (2012).

Styliani, S. et al. Virtual museums, a survey and some issues for consideration. J. Cult. Herit. 10, 520–528 (2009).

Article Google Scholar

Kersten, T., Tschirschwitz, F. & Deggim, S. Development of a virtual museum including a 4D presentation of building history in virtual reality. In TC II & CIPA 3D Virtual Reconstruction and Visualization of Complex Architectures, 1–3 March 2017, Nafplio, Greece 361–367 (Copernicus, 2017).

Bruno, F. et al. From 3D reconstruction to virtual reality: A complete methodology for digital archaeological exhibition. J. Cult. Herit. 11, 42–49 (2010).

Article Google Scholar

Mortara, M. et al. Learning cultural heritage by serious games. J. Cult. Herit. 15, 318–325 (2014).

Article Google Scholar

Anderson, E. F. et al. Developing serious games for cultural heritage: A state-of-the-art review. Virtual Real. 14, 255–275 (2010).

Article Google Scholar

Tsita, C. et al. An approach to facilitate visitors’ engagement with contemporary art in a virtual museum. In International Conference on Transdisciplinary Multispectral Modeling and Cooperation for the Preservation of Cultural Heritage 207–217 (Springer, 2021).

Cao, K. Development and design case function comparison of panoramic roaming system of virtual museum based on Pano2VR. Mob. Inf. Syst. 2022, 7363221 (2022).

Google Scholar

Ahn, B. Y. et al. Construction of cyber science museum. In Proceedings. 2003 International Conference on Cyberworlds 440–445 (IEEE, 2003).

Zhang, J. Innovative application of virtual display technique in virtual museum. IOP Conf. Ser.: Mater. Sci. Eng. 231, 012043 (2017).

Article Google Scholar

Aylan, F. K. & Gök, H. S. Panoramic museum visits as cultural recreation activities: Panorama 1453 History Museum example. J. Tour. Leis. Hosp. 3, 103–117 (2021).

Google Scholar

Guo, Y. et al. Design of virtual scenes for revolutionary education based on VR technology. In Proceedings of the World Conference on Intelligent and 3-D Technologies (WCI3DT 2022) 649–658 (Springer, 2023).

Zhang, L. et al. VR games and the dissemination of cultural heritage. In Distributed, Ambient and Pervasive Interactions: Understanding Humans 439–451 (Springer, 2018).

Doulamis, A. et al. Serious games for cultural applications. Intell. Comput. Graph. 2012, 97–115 (2012).

Google Scholar

Bekele, M. K. et al. A survey of augmented, virtual, and mixed reality for cultural heritage. J. Comput. Cult. Herit. 11, 1–36 (2018).

Article Google Scholar

Nagamachi, M. Kansei engineering: A new ergonomic consumer-oriented technology for product development. Int. J. Ind. Ergon. 15, 3–11 (1995).

Article Google Scholar

Hsiao, Y. H., Chen, M. C. & Liao, W. C. Logistics service design for cross-border E-commerce using Kansei engineering with text-mining-based online content analysis. Telemat. Inform. 34, 284–302 (2017).

Article Google Scholar

Li, M. & Dai, Y. Optimization strategies for the modular resource construction of art gallery’s exhibition halls based on Kansei engineering. IEEE Access 12, 26800–26818 (2024).

Google Scholar

Llinares, C. & Page, A. F. Differential semantics as a Kansei engineering tool for analyzing the emotional impressions which determine the choice of neighbourhood: The case of Valencia, Spain. Landsc. Urban Plan. 87, 247–257 (2008).

Article Google Scholar

Huang, J. M. et al. Costume innovative design based on Kansei engineering. Appl. Mech. Mater. 651, 2024–2027 (2014).

Article Google Scholar

Jindo, T. & Hirasago, K. Application studies to car interior of Kansei engineering. Int. J. Ind. Ergon. 19, 105–114 (1997).

Article Google Scholar

Lai, H. H. et al. User-oriented design for the optimal combination on product design. Int. J. Prod. Econ. 100, 253–267 (2006).

Article Google Scholar

Hsu, S. H., Chuang, M. C. & Chang, C. C. A semantic differential study of designers’ and users’ product form perception. Int. J. Ind. Ergon. 25, 375–391 (2000).

Article Google Scholar

Jiao, J. R., Zhang, Y. & Helander, M. A Kansei mining system for affective design. Expert Syst. Appl. 30, 658–673 (2006).

Article Google Scholar

Liu, X. & Yang, S. Study on product form design via Kansei engineering and virtual reality. J. Eng. Des. 33, 412–440 (2022).

Article Google Scholar

Sugihara, K., Ishii, H. & Tanaka, H. Interval priorities in AHP by interval regression analysis. Eur. J. Oper. Res. 158, 745–754 (2004).

Article Google Scholar

Chen, C. C. & Chuang, M. C. Integrating the Kano model into a robust design approach to enhance customer satisfaction with product design. Int. J. Prod. Econ. 114, 667–681 (2008).

Article Google Scholar

Nagamachi, M., Okazaki, Y. & Ishikawa, M. Kansei engineering and application of the rough sets model. Proc. Inst. Mech. Eng. I J. Syst. Control Eng. 220, 763–768 (2006).

Google Scholar

Wang, T. & Zhou, M. Integrating rough set theory with customer satisfaction to construct a novel approach for mining product design rules. J. Intell. Fuzzy Syst. 41, 331–353 (2021).

Google Scholar

Lai, H. H., Lin, Y. C. & Yeh, C. H. Form design of product image using grey relational analysis and neural network models. Comput. Oper. Res. 32, 2689–2711 (2005).

Article Google Scholar

Chen, Z. An elderly-oriented design of HMI in autonomous driving cars based on rough set theory and backpropagation neural network. IEEE Access 12, 26800–26818 (2024).

Article Google Scholar

Tsuchiya, T. et al. A method for learning decision tree using genetic algorithm and its application to Kansei engineering system. In 1999 IEEE International Conference on Systems, Man, and Cybernetics 6, 279–283 (IEEE, 1999).

Nagamachi, M. Kansei engineering as a powerful consumer-oriented technology for product development. Appl. Ergon. 33, 289–294 (2002).

Article PubMed Google Scholar

Zou, J. & Li, P. Modelling of litchi shelf life based on the entropy weight method. Food Packag. Shelf Life 25, 100509 (2020).

Article Google Scholar

Tang, W. Y. et al. Research on multi-objective optimisation of product form design based on Kansei engineering. J. Eng. Des. 2024, 1–26 (2024).

Google Scholar

Wang, W. et al. Multilevel comprehensive evaluation and decision making of ergonomics. Discrete. Dyn. Nat. Soc. 2015, 689203 (2015).

Article Google Scholar

Zhang, S. et al. Cognitive matching of design subjects in product form evolutionary design. Comput. Intell. Neurosci. 2021, 8456736 (2021).

Article PubMed PubMed Central Google Scholar

Opricovic, S. & Tzeng, G. H. Compromise solution by MCDM methods: A comparative analysis of VIKOR and TOPSIS. Eur. J. Oper. Res. 156, 445–455 (2004).

Article Google Scholar

Tzeng, G. H. & Huang, J. J. Multiple attribute decision making: Methods and applications (CRC Press, 2011).

Zanakis, S. H. et al. Multi-attribute decision making: A simulation comparison of select methods. Eur. J. Oper. Res. 107, 507–529 (1998).

Article Google Scholar

Behzadian, M. et al. A state-of the-art survey of TOPSIS applications. Expert Syst. Appl. 39, 13051–13069 (2012).

Article Google Scholar

Lin, Y. C., Chen, C. C. & Yeh, C. H. Intelligent decision support for new product development: A consumer-oriented approach. Appl. Math. Inf. Sci. 8, 2761–2770 (2014).

Article Google Scholar

Wang, M., Cheng, X. & He, Z. Research on multiple affective responses design of product based on Kansei engineering and TOPSIS-AISM. Math. Probl. Eng. 2022, 6945986 (2022).

Google Scholar

Quan, H. et al. Personalized product evaluation based on GRA-TOPSIS and Kansei engineering. Symmetry 11, 867 (2019).

Article Google Scholar

Vapnik, V. The nature of statistical learning theory (Springer, 2013).

Burges, C. J. C. A. tutorial on support vector machines for pattern recognition. Data Min. Knowl. Discov. 2, 121–167 (1998).

Article Google Scholar

Vapnik, V., Golowich, S. & Smola, A. Support vector method for function approximation, regression estimation and signal processing. Adv. Neural Inf. Process. Syst. 9, 281–287 (1996).

Google Scholar

Yang, C. C. & Shieh, M. D. A support vector regression based prediction model of affective responses for product form design. Comput. Ind. Eng. 59, 682–689 (2010).

Article Google Scholar

Kennedy, J. & Eberhart, R. Particle swarm optimization. In Proceedings of ICNN'95 4, 1942–1948 (IEEE, 1995).

Yang, C. C. Constructing a hybrid Kansei engineering system based on multiple affective responses: Application to product form design. Comput. Ind. Eng. 60, 760–768 (2011).

Article Google Scholar

Yuan, B. et al. Applying latent Dirichlet allocation and support vector regression to the aesthetic design of medical nursing beds. J. Comput. Inf. Sci. Eng. 23, 051014 (2023).

Article Google Scholar

Shieh, M. D., Li, Y. & Yang, C. C. Comparison of multi-objective evolutionary algorithms in a hybrid Kansei engineering system for product form design. Adv. Eng. Inform. 36, 31–42 (2018).

Article Google Scholar

Davis, M. & Putnam, H. A computing procedure for quantification theory. J. ACM 7, 201–215 (1960).

Article Google Scholar

Abdi, H. & Williams, L. J. Principal component analysis. Wiley Interdiscip. Rev. Comput. Stat. 2, 433–459 (2010).

Article Google Scholar

Ishihara, S. et al. An analysis of Kansei structure on shoes using self-organizing neural networks. Int. J. Ind. Ergon. 19, 93–104 (1997).

Article Google Scholar

Kang, X. & Nagasawa, S. Integrating Kansei engineering and interactive genetic algorithm in Jiangxi red cultural and creative product design. J. Intell. Fuzzy Syst. 44, 647–660 (2023).

Google Scholar

Wu, Y. & Zhang, H. Image style recognition and intelligent design of oiled paper bamboo umbrella based on deep learning. Comput. -Aided Des. Appl. 19, 76–90 (2021).

Article Google Scholar

Tharangie, K. G. D. et al. Insight into Kansei color combinations in interactive user interface designing. In Human-Computer Interaction. New Trends 735–744 (Springer, 2009).

Deng, L. & Wang, G. Quantitative evaluation of visual aesthetics of human-machine interaction interface layout. Comput. Intell. Neurosci. 2020, 9815937 (2020).

Article PubMed PubMed Central Google Scholar

Naim, N. F. M. & Hwang, M. Emotional correlation test from binary gender perspective using Kansei Engineering approach on IVML prototype. J. Ind. Eng. Manuf. Lett. 23, 68–74 (2023).

Google Scholar

Qu, Q. X. Kansei knowledge extraction based on evolutionary genetic algorithm: An application to e-commerce web appearance design. Theor. Issues Ergon. Sci. 16, 299–313 (2015).

Article Google Scholar

Gretalita, N. W. M., Suzianti, A. & Ardi, R. Visual usability design of financial personal assistant application. In 2017 International Conference on Information Technology 142–147 (IEEE, 2017).

Download references

The authors would like to thank all individuals who contributed to the development of this paper. This research has been funded by the National Social Science Foundation of China Major Program, 17ZDA155.

School of Design and Art, Beijing Institute of Technology, Beijing, China

Zikai Wang & Fang Li

You can also search for this author inPubMed Google Scholar

You can also search for this author inPubMed Google Scholar

Z.K.W. conducted the main research, data collection, and manuscript writing. F.L. supervised the entire study, provided guidance on research design and analysis, and critically reviewed and revised the manuscript. All authors read and approved the final manuscript.

Correspondence to Fang Li.

The authors declare no competing interests.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

Wang, Z., Li, F. Design of panoramic virtual museum interactive interface based on entropy weight TOPSIS and PSO-SVR. npj Herit. Sci. 13, 201 (2025). https://doi.org/10.1038/s40494-025-01760-5

Download citation

Received: 26 November 2024

Accepted: 29 April 2025

Published: 21 May 2025

DOI: https://doi.org/10.1038/s40494-025-01760-5

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative