Perceptual Attributes Identification and Importance–Performance Alignment Assessment of Urban Underground Complex: A Case Study in Chengdu Tianfu Square

Abstract

Background and objectives: The urban underground complex integrates comprehensive architecture and subterranean space, presenting physical and psychological complexities. To mitigate the negative experiences associated with this complexity, we need to explore, in depth, how the built environmental elements of underground complexes affect user perception to design a comfortable, healthy, and biophilic underground space. Methods: We selected Tianfu Square, a typical underground complex in Chengdu, China, as the empirical case. We identified 26 key environmental indicators affecting user perception and designed Kano and satisfaction questionnaires based on these indicators. A total of 476 questionnaires were distributed to collect data on users’ attitudes towards changes in element quality and their satisfaction levels. By integrating the Kano model and IPA method, we defined the perceptual attributes of elements based on their impact on user experiences and assessed the alignment between element attributes and satisfaction. Results: We categorized the elements into five perceptual types: attractive, one-dimensional, indifferent, must-be, and reverse attributes, and assigned corresponding importance levels. We then compared the importance and performance scores of these elements, evaluated their satisfaction alignment, identified elements needing optimization, and proposed improvement strategies. Implications: This research provides valuable guidance for designers on how various built environment elements in underground spaces influence user perceptions, with practical implications for urban underground complex planning and design.

1. Introduction

The metro development brings economic benefits [1], and underground space development develops in the direction of diversification, intensification, and systematization [2]. The underground space, initially an accessory space, has gradually evolved into a space with precise and recognizable architectural characteristics, offering independent development value and serving as a true urban space for the contemporary city’s expansion [3,4]. The urban underground complex is the product of underground space development to a particular stage, a large-scale comprehensive underground facility formed by an organic combination of commercial, transportation, and other functional units [5,6,7]. An underground space is generally defined as an enclosed environment situated below the earth’s surface. Consequently, unlike above-ground environments, it lacks direct access to outdoor open spaces [8]. Generally, the enclosed spatial configuration, deprivation of natural elements, multifaceted functional requirements, and other salient features characterize the underground complex. It stands as a notable example within the category of complex architectural designs, demanding significant investment costs and involving numerous development stakeholders across various architectural projects [9].

With the high-density development demand of the city and the progress of technology, the construction goal of the underground complex has changed from a space with a specific logistical function to a comprehensive place [10]. However, does the design result align with this shift in objectives? Due to the unique characteristics of underground spaces, users lack direct visual access to external events. This sense of confinement is often associated with a reduced perception of environmental control, which can lead to feelings of claustrophobia and repression [11]. In recent years, the interaction between humans and the environment has garnered significant attention from scholars in environmental psychology, who explore human cognition, emotion, and behavior towards the environment, making the correlation between people’s perception and the built environment a central theme in the study of underground space [12,13]. Studies have shown the air, sound, light, nature, transport, and spatial context of underground spaces, each of which exerted relationships with users’ physiological and psychological factors [14]. When these conditions create a significant disparity with traditional above-ground spaces, the public may feel frustrated and less willing to use underground spaces [15,16]. Therefore, to better understand how the built environment affects public perception in current underground complexes, we conducted a post-use evaluation using the Kano-IPA method. The goal is to identify key elements in the subsurface environment and explore their relationship with public satisfaction, assessing the extent of their impact and implementation alignment. This study provides designers with valuable feedback from users, reflecting current conditions and informing future design and practice.

In this study, we have innovatively applied the Kano-IPA method to the study of underground complexes. The Kano model is a framework used to capture user preferences and identify the perceptual features of elements. It defines and classifies customer needs by discovering the non-linear relationship between user perception and product quality [17,18] so as to better understand the priority of various needs, identify the key elements affecting comfort and pleasure, and make more informed improvement decisions for designers [19]. Importance–Performance Analysis (IPA) provides a two-dimensional view of quality, identifying strengths and weaknesses from users’ perspectives by simultaneously evaluating importance and performance [20]. The combination of the Kano model and IPA method is suitable for built environment assessment. The integration of these methods allows for determining importance levels through perception and visually presenting improvement priorities in an importance–performance matrix. We focused on Tianfu Square, a representative underground complex in Chengdu, to analyze the impact of critical spatial elements on user satisfaction. We categorized these elements into various types based on their distinct characteristics that influence satisfaction levels. Then, we assessed the importance and satisfaction associated with each spatial element, identifying discrepancies between their importance and performance. Ultimately, we highlighted the weaknesses of these elements and proposed targeted optimization strategies.

The paper is structured as follows: Section 2 outlines the research methodology employed in this study, including an introduction to the Kano-IPA method, the evaluation index, and the case studies along with their data sources. Section 3 presents the identification results of the Kano classification of elements, emphasizing the strengths and weaknesses of the built environment through an assessment of the alignment between elemental importance and user satisfaction. Finally, Section 4 concludes the study by discussing its findings and limitations.

2. Method

2.1. Kano-IPA Method

(1)

Kano Model

The Kano model, developed by Noriaki Kano [21] in Japan, examines the impact of functional quality on user satisfaction. The Kano model was initially applied in product development, marketing, and operations management [22,23]. It identifies and classifies attributes by refining the constituent elements of a product or service. This model is also applicable to the built environment, as it consists of component elements that determine spatial quality and can be similarly deconstructed and analyzed. Llinares et al. [24] proposed a method to define strategies for improving urban perception using the Kano classification, taking the urban design of Valencia as an example. Chen et al. [25] used the Kano model to identify and analyze the main attractive elements of art street. The perceptual characteristics of the elements are defined and classified according to their path of influence on satisfaction. As shown in Figure 1, the X-axis represents the degree of fulfillment of the element quality, while the Y-axis represents the level of user satisfaction. The varying relationships between the degree of fulfillment and user satisfaction correspond to five typical types of perceptual attributes: attractive, one-dimensional, must-be, indifferent, and reverse attributes.

▪

Attractive attribute: users are satisfied if the quality is present or upgraded but are not dissatisfied if it is absent. The relationship between attractive attributes and satisfaction is similar to an exponential function.

▪

One-dimensional attribute: user satisfaction increases with the fulfillment of the quality, and vice versa; this attribute is positively and linearly related to customer satisfaction. Users’ reaction depends linearly on fulfillment level only for one-dimensional requirements.

▪

Must-be attribute: this is the essential quality of a product/service, and users are highly dissatisfied in its absence. However, fulfilling this quality does not increase satisfaction since users take it for granted.

▪

Indifferent attribute: user satisfaction has nothing to do with the presence or absence of this quality, which is irrelevant.

▪

Reverse attribute: with the improvement of this quality, the user’s satisfaction gradually decreases, and vice versa. The one-dimensional and reverse attributes have a similar relationship with satisfaction but are opposite.

The attributes of an element dictate its improvement priority and the corresponding course of action. In terms of prioritization, the quality of elements should ensure that users do not experience dissatisfaction, followed by efforts to enhance user satisfaction. Following this logic, the primary focus should be on elements classified as must-be attributes, which are essential for preventing any perception of dissatisfaction. Next, attention should be directed toward attractive elements, which hold an average priority. While these elements do not contribute to dissatisfaction, they have the potential to enhance overall satisfaction. Finally, there are indifferent elements that possess the lowest priority. Regardless of their quality, these elements have a minimal impact on users’ perceptions.

The Kano model is usually used to analyze the data obtained from the Kano questionnaire, and its specific operation process is divided into three steps: Kano questionnaire, Kano evaluation table, and respondents’ judgment counted [17] (Figure 2).

Step 1. Invite users to fill out the Kano questionnaire, which contains a set of question pairs for each product attribute. The question set includes a functional and a dysfunctional form that captures customers’ responses when a product has or does not have a certain attribute.

Step 2. Judge each questionnaire using the Kano evaluation table. The questionnaire is deployed to a number of customers, and each answer pair is aligned with the Kano evaluation table, highlighting the individual customer’s perception of a product attribute.

Step 3. Count the results of each respondent’s judgment counted. The product attribute is classified according to a statistical analysis of the survey results of all respondents (the most frequent observation).

Calculate the better coefficient (SI) and worse coefficient (DSI) and bring them into the better–worse matrix (Figure 3) to reveal the final attribute of the elements. The better coefficient represents the degree to which satisfaction increases when designers decide to increase the factor quality, and the worse coefficient represents the degree to which satisfaction decreases when downgrading the factor quality. SI is defined by the proportion of performance attributes (one-dimensional and attractive attributes) in all attributes (Equation (1)); DSI is defined by the proportion of basic attributes (must-be and one-dimensional attributes) invalid data (Equation (2)) [23]. The value of SI is positive and the value of DSI is negative. The greater the absolute value of the coefficient, the greater the influence strength.

$$Better \left(SI\right)={\displaystyle \frac{A+O}{A+O+M+I }}$$

(1)

$$Worse \left(DSI\right)=-1\times {\displaystyle \frac{M+O}{A+O+M+I }}$$

(2)

Figure 3 is a four-quadrant matrix of the better–worse coefficient, a visual representation of the Kano classification elements. When the SI and DSI of an element are high and close to each other, the element is located in the first quadrant of the matrix, conforming to the characteristics of the one-dimensional attribute. When the DSI is obviously higher than the SI, the element is located in the second quadrant, which accords with the feature of the must-be attribute. When both DSI and SI are low, the element is located in the third quadrant, which is indifferent. When the SI is significantly higher than the DSI, the element is in the fourth quadrant and meets the attractive attribute.

(2)

Importance–Performance Analysis (IPA)

Importance–Performance Analysis (IPA) is a method used to prioritize enhancement which relies on user evaluations to assess the importance and performance of each product, service, or element [26,27]. In this study, the Importance–Performance Alignment of environmental elements in the underground complex was assessed using the IPA method. The specific method is to compare the standardized importance and performance values to evaluate the perceptual state of the elements. Ideally, the importance and performance ratings should match each other. Specifically, a higher perception of elements should correspond to increased satisfaction, indicating that users have had a positive experience with these environmental elements.

Based on the assessment, we classified the quality attributes into four quadrants: Quadrant I represents the “high importance & high performance” area. Quadrant II corresponds to the “high importance & low performance” area. Quadrant III denotes the “low importance & low performance” area. Quadrant IV signifies the “low importance & high performance” area. According to IPA theory, the recommended actions for these quadrants are “Keep up the good work”, “Concentrate here”, “Low priority”, and “Possible overkill” [28] (Figure 4). The IPA method is widely used in the evaluation of factors, services, and other elements, providing corresponding guidance strategies. It is also common to combine IPA with other methods to define importance, including combinations such as the analytic hierarchy process [29], structural Equation model [30], and fuzzy comprehensive evaluation [31].

(3)

Combination of Kano and IPA

In this study, we integrate the Kano model with IPA methodology, where the Kano model is employed to establish the importance criteria utilized in IPA analysis. Factor importance is gauged based on its influence on satisfaction levels. Factors with a significant impact on satisfaction are deemed highly important in our assessment. Combining the Kano and IPA models addresses the Kano model’s limitation of neglecting attribute performance and importance while providing a reliable quantitative basis for the importance values in IPA evaluation [32]. Many scholars in the field of architectural planning have combined these two methods to study the satisfaction of urban public areas [33], architectural environments [34], and landscape environments [35]. These results demonstrate the feasibility and effectiveness of this combined approach.

Unlike most previous studies on Kano and IPA that focus on using performance and importance evaluation methods to define Kano attributes, we emphasize the type characteristics of the analysis elements and the consistency of performance–importance. Therefore, we have two different evaluation systems: Kano evaluation and importance–performance evaluation. In our approach, the importance of IPA is defined by the Kano results. By incorporating the importance weights assigned to attributes in the Kano model, we assigned weights of 4, 3, 2, and 1 to each attribute. The determination of the element’s importance value involved the type weight W, better coefficient (SI), and worse coefficient (DSI), calculated using the following equation.

$$\mathrm{Importance}=Wt\times \left(SI+DSI\right)$$

(3)

After quantifying the importance of the elements, we compared the importance with the satisfaction score, which was rated using a 5-point Likert scale. The Likert scale is the most common and widely used tool to collect data on attitudes, especially in surveys and questionnaires [36]. Scores of 5, 4, 3, 2, and 1 corresponded to very satisfied, satisfied, moderately satisfied, dissatisfied, and very dissatisfied, respectively [37]. The performance formula is as follows.

$$\mathrm{Performance} ={\displaystyle \frac{1}{n}}{\displaystyle \sum }_{i=1}^n\mathrm{Satisfaction}$$

(4)

2.2. Index System

To construct the perception evaluation index system of the underground complex space for making questionnaires, we need to understand the concerns of users first. We collected 19 major problems commonly found in underground complexes through a literature review and random interviews. Then, we collated these problems into a multi-choice questionnaire, and 163 users participated in the online survey. We selected the problems with more than 20% frequency as the focus problems and proposed 26 corresponding perception indicators of the built environment for these focus problems (

Table 1. Focus issues in underground complexes acquired from the online survey, and the perception index corresponding to these issues.

Focus Issues	Frequency	Proportion	Corresponding Index
There is a lack of expected business types and a shortage of anchor stores.	92	56.44%	Diversity of commerce; quantity of each type of commerce
The scale of the building is unreasonable, too large or too small	89	54.60%	Scale of the building
The complex flow of the building makes people get lost	84	51.53%	Interior circulation; way-finding guidance
Goods are priced too high or too low	77	47.24%	Consumer orientation
Lack of public transportation accessibility	70	42.94%	Metro, bus, motorized, non-motorized, and other connecting methods
Inadequate indoor lighting and dim environments	65	39.88%	Lighting environment; natural lighting
It’s hard to find the entrance	61	37.42%	Number of entrances; entrance design
Poor ventilation, rather dull and hot	60	36.81%	Natural ventilation; temperature regulation
Traffic jams, people, and cars mixed	60	36.81%	Pedestrian facilities such as underpasses and footbridges
The environment is too noisy or cold	59	36.20%	Noisy environment
Depressing and cramped space	52	31.90%	Scale of the building; natural lighting; natural elements
Lack of personality and character in public space design	51	31.29%	Characteristic space design; impressive art nodes or architectural installations; personalization of the store; natural elements
Single-store design and lack of recognition	45	27.61%	The personalized design of the store
Inadequate promotion and advertising, lacking attractiveness	39	23.93%	Advertising; promotional activities
Inconvenient underground parking	36	22.09%	Motor vehicle connections; way-finding guidance; entrance design
Failure to improve the city’s image	33	20.25%	Marking structure
It’s crowded during rush hour	19	11.66%	Scale of the building; overall scale; interior flow line
Inadequate accessibility	15	9.20%	Facilities for the disabled
Insufficient urban public goods	8	4.91%	Public benefit events; civic open space

n = 163.

). Those indicators were initially divided into 5 dimensions: orientation, external transportation, space perception, service composition, and physical environment (

Table 2. Perception index of the underground complex.

1st Order Index	2nd Order Index
Orientation	A1	Consumer orientation
	A2	Scale of the building
	A3	Marking structure
Transportation	B1	Number of metro lines
	B2	Car transfer
	B3	Bicycle transfer
	B4	Connection to other buildings
Space perception	C1	Interior circulation
	C2	Way-finding guidance
	C3	Spatial scale
	C4	Characteristic space design
	C5	Artwork and installations
	C6	Personalized store
	C7	Natural lighting and ventilation
	C8	Plants and water features
	C9	Open-form entrance
	C10	Ground environment
Service composition	D1	Advertising and interactive activities
	D2	Diversity of commerce
	D3	Quantity of specialty retail
	D4	Quantity of integrated retail
	D5	Quantity of dining and leisure
	D6	Quantity of culture and sport
Physical environment	E1	Ambient background sound
	E2	Lamplight design
	E3	Indoor temperature

).

2.3. Case and Data

(1)

Empirical Case

We selected the Tianfu Square underground complex as our empirical case. Located in the urban center of Chengdu, Sichuan, Tianfu Square serves as an economic, cultural, and commercial hub, reflecting the city’s image [6]. Tianfu Square’s underground complex was built in 2013. It is positioned as an “urban living room” and is an urban commercial complex with experiential leisure as the theme. The first and second floors are commercial and underground pedestrian walkways, respectively. The third and fourth floors are the platforms of Chengdu Metro Lines 1 and 2 (Figure 5). Tianfu Square metro station is one of the most visited stations in Chengdu, with 1.9 million passengers passing through in July 2024. The mall sees an average monthly foot traffic of approximately 390,000 people and currently hosts 105 merchants [38].

(2)

Data Source

The data source is the questionnaire we conducted in the building to find random users. We have two sets of questionnaires. The first is the Kano questionnaire, which aims to categorize the attributes of the elements. The second is the satisfaction questionnaire, which aims to obtain user perception ratings to quantify the performance of the building. Due to the large number of questions in the questionnaire, we distributed the Kano questionnaire and the satisfaction questionnaire separately to minimize the difficulty for users in filling in the questionnaire and increase the efficiency of our work. Figure 6 shows the floor plan of Tianfu Square. We distributed questionnaires at the four marked points. We distributed 300 questionnaires of each type and recovered 210 Kano questionnaires (recovery rate is 70%) and 266 satisfaction questionnaires (recovery rate is 88.67%). Sample information is shown in

Table 3. Sample information of Kano questionnaire and satisfaction questionnaire.

Item		Kano Questionnaire (%) (N = 210) (%)	Satisfaction Questionnaire (%) (N = 266)
Gender	Male	55.64	51.90
	Female	44.36	48.10
Age	<18	15.04	24.29
	18–40	67.29	53.33
	>40	17.67	22.38
Education	Below Bachelor’s	33.46	67.14
	Bachelor’s	58.65	26.67
	Master’s and above	7.89	6.19
Arrive mode	Metro	63.16	69.52
	Bus	7.52	14.76
	Car	18.42	8.57
	Bike/Walk	10.90	7.14
Visit frequency	First time	53.96	33.33
	Occasional visit	34.34	49.52
	Frequent visit	11.70	17.14

.

2.4. Research Framework

From the perspective of the people’s perception of the underground complex environment, this study discusses the identification of perceptual attributes and the importance of performance synergy analysis of the urban underground complex. To achieve our research objectives, we developed the following framework (Figure 7).

The initial phase serves as the research basis, encompassing a fundamental collection and description of the perceptual attributes and satisfaction data of the urban underground complex. The specific implementation process uses the underground complex perception index system, designed based on the focal problem, to create satisfaction and Kano questionnaires to obtain users’ evaluations. This segment furnishes the primary dataset underpinning our research endeavors.

In the second part, the Kano-IPA model was employed in the second phase to delineate spatial elements’ perceptual attributes and synergy. Initially, leveraging the Kano questionnaire data, the Kano model was applied to assess and categorize the perceptual characteristics of elements. The better–worse coefficient method was employed to quantify the impact of perceptual factors on satisfaction and dissatisfaction, subsequently integrated into the Kano model. Elements were categorized as attractive, one-dimensional, must-be, indifferent, or reverse attributes based on their positioning within the Kano model. Varied attributes were assigned distinct importance weights. By amalgamating attribute weights with the influence intensity of each element on perception, the importance value of spatial elements was quantified. Afterward, a comparative analysis was undertaken between the importance value assigned to elements and their respective satisfaction performance. This analysis facilitated the identification of deficient elements within the underground complex that necessitated adjustment strategies, whether through augmenting or reducing input.

In the final part, the research findings were deliberated upon, and strategic recommendations were proffered. Building upon the insights gleaned from the Kano-IPA analysis, further elucidation on the satisfaction characteristics of elements was provided. By integrating these findings with on-site investigations of the structure, optimization suggestions were formulated for elements displaying subpar performance, with the goal of enhancing user satisfaction.

3. Results

3.1. Reliability and Validity

The satisfaction questionnaire is a continuous scale; therefore, its reliability and validity can be tested. We used SPSS 25.0 to test the reliability and validity of the satisfaction questionnaire results. First, the reliability was tested using Cronbach’s alpha, which estimates the internal consistency of the test. Cronbach’s alpha coefficients range from 0 to 1. The closer the value is to 1, the higher the consistency. The Cronbach’s alpha of this study is 0.623. It is higher than 0.6, which indicates that the internal consistency of the questionnaire is accepted [39].

Second, we used confirmatory factor analysis to test the correctness of the questionnaire dimension division. Exploratory Factor Analysis (EFA) is a key multivariate statistical technique that is widely used in the verification of psychological evaluation scale [40]. Results showed the KMO statistic to be 0.883 and Bartlett’s spherical test to be X2 = 28,885.17 and p = 0.000, which, taken together, indicated that the samples in this study were suitable for factor analysis. Factor analysis was performed to explore the dimensions of the questionnaire. After principal component analysis, all factors loading above 0.4 can be retained [41], and five components were extracted (

Table 4. Results of the KMO test and Bartlett’s test of Sphericity for the satisfaction data.

KMO Test and the Bartlett’s Test of Sphericity
KMO Measure of Sampling Adequacy		0.883
Bartlett Test of Sphericity	Approx. Chi-Square	2885.171
	df	325
	Sig.	0.000

). Those components can explain 58.093% of the total variance (

Table 5. Total variance explained based on the field questionnaire.

Factors	Total	% of Variance	Cumulative%	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	7.245	27.864	27.864	7.245	27.864	27.864	5.527	21.259	21.259
2	2.644	10.168	38.032	2.644	10.168	38.032	3.498	13.452	34.711
3	2.198	8.452	46.484	2.198	8.452	46.484	2.318	8.914	43.625
4	1.707	6.567	53.051	1.707	6.567	53.051	1.927	7.413	51.038
5	1.311	5.042	58.093	1.311	5.042	58.093	1.834	7.055	58.093

), which is above 50% and is acceptable [42]. The factor test tested the rationality of the evaluation index system constructed previously (

Table 6. Factor analysis.

Component	Name	Factor
1	Orientation	A1/0.772	A2/0.695	A3/0.580
2	Transportation	B1/0.798	B2/0.639	B3/0.675	B4/0.648
3	Space perception	C1/0.633	C2/0.749	C3/0.679	C4/0.716	C5/0.730
		C6/0.768	C7/0.683	C8/0.610	C9/0.533	C10/0.738
4	Service composition	D1/0.696	D2/0.797	D3/0.697	D4/0.733	D5/0.701
		D6/0.738
5	Physical environment	E1/0.742	E2/0.776	E3/0.716

Extraction Method: Principal component analysis. Rotation Method: Varimax rotation.

).

3.2. Attribute Identification

Figure 8 shows the statistics of the attribute proportion of each factor. E1 (ambient background sound) has the highest proportion of negative attributes (more than other attributes), which is directly judged as a negative factor. In the next step, we calculated SI and DSI values for other factors according to the proportion statistics of various attributes.

Table 7. Results of SI and DSI.

Indicator	SI	DSI	Indicator	SI	DSI
A1	0.621	−0.571	C7	0.590	−0.566
A2	0.332	−0.384	C8	0.667	−0.497
A3	0.560	−0.565	C9	0.606	−0.581
B1	0.497	−0.678	C10	0.566	−0.606
B2	0.560	−0.653	D1	0.709	−0.460
B3	0.420	−0.392	D2	0.658	−0.565
B4	0.642	−0.583	D3	0.348	−0.342
C1	0.715	−0.549	D4	0.522	−0.467
C2	0.397	−0.379	D5	0.602	−0.619
C3	0.524	−0.615	D6	0.599	−0.625
C4	0.615	−0.620	E2	0.375	−0.591
C5	0.708	−0.400	E3	0.474	−0.610
C6	0.619	−0.633

shows the calculation results.

The calculation results of SI and DSI confirm the path and intensity of each factor’s influence on satisfaction. We incorporated these results into the Kano model (Figure 9), where the mean values of SI and DSI serve as demarcation lines for the four quadrants of the Kano-type matrix. By analyzing the positioning of each element within this matrix, we can identify the distinct attribute types associated with each element based on their respective locations.

▪

Quadrant I: The elements in this region are one-dimensional elements. Their SI and DSI are both high, indicating that upgrading or downgrading significantly impacts satisfaction. The elements include A1 (Consumer orientation), A3 (Marking structure), B2 (Car parking), B4 (Connection to other buildings), C4 (Characteristic space design), C6 (Personalized of the store), C7 (Natural lighting and ventilation), C9 (Open-form entrance), C10 (Ground environment), D2 (Diversity of commerce), D5 (Quantity of dining and leisure), and D6 (Quantity of culture and sport).

▪

Quadrant II: The elements in this region are must-be elements, and their SI is not high, but their DSI is high, indicating that upgrading this element has a moderate impact on satisfaction; meanwhile, downgrading this element will significantly reduce satisfaction, including B1 (Number of the metro line), C3 (Spatial scale), E2 (Lamplight design), and E3 (Indoor temperature).

▪

Quadrant III: The elements in this region are indifferent elements. Their SI and DSI are not high, indicating that upgrading or downgrading the element has no significant impact on satisfaction, including A2 (Scale of the building), B3 (Bicycle parking), C2 (Way-finding guidance), D3 (Quantity of specialty retail), and D4 (Quantity of integrated retail).

▪

Quadrant IV: The elements in this area are attractive elements. Their SI is very high, but their DSI is not high, which indicates that upgrading the element can significantly improve satisfaction; however, degrading the element will not considerably reduce satisfaction. This includes C5 (Artwork and installations), C8 (Plants and water features), and D1 (Advertising, and interactive activities).

3.3. Importance–Performance Alignment Assessment

The perception and satisfaction of the factors should be matched together. The IPA can give the order of improvement. Factors of high importance or low satisfaction need to be satisfied first. The results show that the order of importance is as follows: physical environment > traffic > spatial perception > service composition > orientation, and the order of satisfaction is physical environment > service composition > external traffic > orientation > spatial perception (Figure 10).

We use a four-quadrant IPA matrix to explain the matching relationship between importance and satisfaction (Figure 11). We used mean values of importance and performance as the critical values of the four regions, a method similar to that of the Kano model mentioned above.

When the values of importance and performance align, it typically indicates a favorable relationship, suggesting that these elements do not require improvement. However, this alignment can manifest in two distinct scenarios. The first scenario occurs in the matrix’s first quadrant, where both importance and performance are relatively high. Examples of such advantageous elements include B1 (Number of metro lines) and E3 (Indoor temperature), which are the main sources of praise in Tianfu Square. For these elements, the recommended strategy is to prioritize maintenance. The second scenario is found in the third quadrant, where both importance and performance levels are low. Elements such as B3 (Bicycle parking) and C8 (Plants and water features) do not significantly influence user satisfaction, despite users expressing dissatisfaction with them. In this case, developers may consider adopting a status quo strategy, especially if budget constraints are a concern.

There are two scenarios in which the importance and performance of elements do not align. In the first scenario, where importance is high but performance is low, these elements fall into the second quadrant. Typical examples include C1 (Interior flow line) and C3 (Spatial scale). The mismatches of these elements can have a relatively negative impact on the overall evaluation of buildings. Consequently, these elements represent significant potential and warrant concentration here. Conversely, the second scenario involves low importance but high performance, which is situated in the fourth quadrant. For instance, C5 (Artwork and installations) and D3 (Quantity of specialty retail) exemplify this category, suggesting that these elements may be overkill. Therefore, investment and management allocated to these elements should be reduced.

4. Discussion and Conclusions

4.1. Discussion

The physical environment has the highest importance value, indicating that physical comfort is one of the primary conditions for a good experience in an underground complex. This conclusion is similar to the work of some other scholars [13,43,44]. The performance value of the physical environment is also high, which matches the importance level, and all elements belong to the type of high importance and high performance. Among them, the E1 (Ambient background sound) is the opposite factor because Tianfu Square is located in the city’s center, and there is a large volume of passengers brought by the subway. It will inevitably be noisy during peak hours. Public especially workers in the complex who are consistently exposed to the high noise levels are at high risk [45]. It is essential to implement noise reduction measures in the noise area, considering the installation of better acoustic absorbers in hotspot areas of the mall (dining, entrance, and recreational areas) to reduce the effect of increased sound levels [46]. E2 (Lamplight design) and E3 (Indoor temperature) are must-be elements, and the current satisfaction is high; therefore, we recommend that designers maintain the current status. To enhance user physiological comfort in underground space designs, it is essential to adopt environmental standards equivalent to those above ground. Furthermore, utilizing smart tools to monitor and adjust air quality, temperature, humidity, and lighting will contribute to creating a healthy underground environment [44].

Transportation has the second highest importance value, and its overall performance is also good, belonging to the type of high importance and high performance. Tianfu Square is strongly attracted to other urban areas, and as a landmark area of Chengdu, it has a robust traffic distribution function. B1 (Number of metro lines) is highly important, as it currently operates Lines 1 and 2 at Tianfu Square Station, facilitating both north–south and east–west transit across the city. Studies have shown that connecting rich commercial facilities around transit stops not only improves economic efficiency but also enhances travel security [12]. Tianfu Square is one of the busiest subway stations in Chengdu due to its strategic location and extensive underground commerce. However, stations with high passenger volume significantly impact network vulnerability [47]. Consequently, Chengdu Metro is constructing Express Line 18, which will integrate with Line 1 to enhance connectivity between the northern and southern parts of the city [48]. B4 (Connection to other buildings) is a critical design element due to its high importance value. In the past, underground spaces were developed more independently, whereas modern underground spaces emphasize convenient connectivity, marking a significant difference and advancement [49]. As an integral component of the urban environment, underground complexes necessitate a well-connected network to link surface and subsurface spaces. It is important to recognize that these spaces exist in three dimensions, with interconnections occurring both horizontally and vertically [50]. These conclusions align with Tang et al. [51], who posited that the key to designing underground complexes lies in using the transit system as the development axis under the guidance of the TOD model, thereby connecting with other regions or centers. The transit station should serve as the development source, while the underground pedestrian network should function as the development flow. The satisfaction of B2 (Car transfer) and B3 (Bicycle transfer) is relatively low. This is because the traffic sections around Tianfu Square are all the main roads, with a large traffic flow. Moreover, planners use underpass tunnels to divert traffic flow in different directions, which leads to difficulty in the transfer of vehicles. A study on shared transportation in smart cities highlights that modern transport must not only fulfill the basic function of movement but also be sustainable, highly accessible, and integrated with urban information and communication technologies [52]. Therefore, establishing transfer infrastructure for shared transportation around the city is an effective strategy. Additionally, visual communication and advertising for shared transportation are essential.

The importance value of space perception is also high, due to the lack of external forms in underground spaces. The exterior of the underground space is not visible; therefore, the interior experience is particularly important [50]. However, the performance of spatial experience in Tianfu Square belongs to the mismatch type of high importance and low performance, which could be better. The spatial elements of space perception need to be optimized. About 30% of the elements fall within the “Concentrate here” improvement areas, and 50% of the factors performed below average. Among them, C3 (Spatial scale) is a necessary element with high importance but low performance. Through our investigation, we found that temporary booths and constructions were added in some areas, resulting in congestion during peak passenger flow, and the overall height of the building was low, resulting in a relatively depressed feeling. The performance of C1 (Interior circulation) is also low because the flow line shape is irregular, it is not easy to form a mental map that people can easily remember the route, and some roads are narrow and multi-curved. To reduce the difficulty of way-finding, we propose adding iconic nodes because Xu et al. and Lu et al. believe that space environmental cue designs can be one of the most important means used for way-finding [53,54]. Another strategy is to create large spaces to increase visibility. In addition, elements such as C7 (Natural lighting and ventilation), C8 (Plants and water features), and C9 (Open-form entrance) are highly important and related to nature. Despite their good performance, they still warrant focused attention. The contact with nature is a significant difference between underground spaces and surface environments [55]. The contact with nature is a significant difference between underground space and surface environment. Many studies have shown that the introduction of biophilic elements in underground space can improve spatial quality. Jia et al. argue that the atrium of an underground building is crucial for creating a positive psychological environment. It enhances the space’s overall image through public and decentralized use and access to natural light and landscapes [34]. The introduction of some natural elements, such as plants and water, can also simulate the ground space. One study has shown that both the number of plants and plant types can moderate the relationship between the physical environments and occupant satisfaction [56]. The plant has aesthetic qualities that are indispensable for the design of attractive and pleasing places [57,58]. Some scholars further divide biophilic experiences into direct and indirect and categorize elements into nature in the space, natural analogs, and the nature of the space [59]. These perspectives show that biophilia is not limited to accessible or visible elements of nature. Therefore, the natural elements and biophilic design in underground complexes warrant further in-depth study.

The importance value of orientation and service composition is low, indicating that these two items will not have a significant impact on the user experience of the building. Among them, the orientation performance value is low, belonging to the type of low importance and low performance. Although their importance and performance match each other, some of the elements have obvious one-dimensional or attractive attributes that can significantly impact satisfaction. For example, A1 (Consumer orientation): Some users believe that Tianfu Square, a landmark Chengdu building, should reflect more regional characteristics and improve consumption. Therefore, if conditions permit, they should still be optimized and improved.

4.2. Conclusions

The urban underground complex is increasingly recognized as an innovative and prevalent architectural typology within urban environments, a providing potential solutions to various urban challenges. Our research focuses on the interaction between the built environment of the underground complex and users’ perceptions, revealing the perceptual attributes and states of spatial elements, which can help designers and developers understand the underground complex.

The main contribution of this study is two-fold. From a theoretical perspective, this study establishes a comprehensive research process for understanding the perceived characteristics of the built environment in underground complexes. This process includes extracting key perceived elements, designing an index system and questionnaire, identifying the Kano attributes of elements, and evaluating the performance-importance alignment of these elements. The study emphasizes the interaction between the built environment and the perceived environment of underground spaces, providing theoretical support for the functional development and psychological transformation of contemporary underground spaces. From a practical perspective, the quantitative classification of elements in the built environment of underground complexes helps designers and developers better understand the relationship between the built environment and user perception. The application of the Kano-IPA method in real projects can quickly identify elements needing improvement in post-use evaluations, providing a valuable references and inspiration for design practice.

This study has some limitations. First, the sample size is small, and the population profile lacks detail. Our future improvements include incorporating a broader range of empirical case types and regions, and increasing sources of perception data, such as social media. Additionally, the construction of underground complexes may be closely related to the TOD model. Therefore, incorporating research on public and shared transportation could significantly expand our understanding of the complexities of using such underground spaces.