On-the-Ground Application of Cloud Evaluation: Big Data Reveals Experiential Effectiveness of Industrial Heritage Revitalization

Featured Application

This research proposes a dynamic, data-driven evaluation system for continuously monitoring and enhancing the performance of revitalized industrial heritage sites by tracking user satisfaction over time and providing a methodology for timely adjustments to project strategies.

Abstract

Post-occupancy evaluation is a critical mechanism for ensuring the sustained success and continuous improvement of industrial heritage revitalization initiatives. The quality of the visitor experience plays a key role in determining a project’s long-term vitality. This study focuses on assessing user satisfaction with a revitalized industrial heritage site by employing web crawling and data mining techniques to systematically collect and analyze user-generated reviews from major online platforms. Using the 1933 Old Millfun in Shanghai, China, as an example, this research identifies six core evaluation dimensions derived from extensive user commentary: project accessibility, cultural legibility, aesthetic distinctiveness, commercial appeal, facility completeness, and sense of security. These dimensions are integrated into a comprehensive analytical framework, with the Fuzzy Comprehensive Evaluation (FCE) method applied to quantitatively assess the site’s performance across each category. By combining qualitative sentiment data with quantitative evaluation techniques, the data-driven presentation provides nuanced insights into the evolving user experience. The research results contribute to the development of a replicable and scalable paradigm for measuring user experience in industrial heritage revitalization and highlights the potential of digital platforms as valuable tools for heritage site management and continuous optimization.

1. Introduction

As a vital component of cultural heritage within the broader context of urban regeneration, industrial heritage reflects the collective memory of human civilization and the evolution of cities. Its preservation and adaptive reuse have become increasingly important in the era of stock-oriented urban development. In recent years, academic research has largely focused on two main areas: the evaluation of industrial heritage values and the development of reuse models alongside supporting policy frameworks. Within this scope, scholars have examined various topics, including methods for heritage value assessment, development of social and economic significance of heritage reuse, and the formulation of adaptive reuse strategies [1,2,3,4,5].

In the early stages of industrial heritage revitalization, many projects tend to yield substantial short-term benefits, including economic gains, cultural enrichment, and enhancement of the city’s image. These initial successes often attract significant public attention and investment from diverse stakeholders. However, over time, numerous projects encounter challenges such as declining public engagement, underutilization, and even abandonment, despite substantial initial investments [6,7]. A key underlying issue is the misalignment between original design concepts or functional orientations and the evolving socio-cultural landscape. As a result, these regenerated spaces often fail to meet the dynamic needs of contemporary society and the increasingly diverse preferences of users [8]. While considerable theoretical progress has been made in evaluating heritage value and developing reuse strategies, user experience—particularly in terms of users’ perceptions and satisfaction—remains a relatively underexplored area [9,10,11]. This limited attention to user feedback and behavioral data often leads to declining user engagement over time, ultimately weakening operational performance and undermining the long-term sustainability of such projects [12].

Today, most industrial heritage revitalization projects continue to function as public spaces [13]. Unlike conventional public amenities, these spaces serve a dual purpose: providing accessible public infrastructure while preserving and conveying industrial culture to future generations. Accordingly, industrial heritage sites should meet expectations not only for physical comfort and spatial quality but also for cultural engagement. Users evaluate typical public space elements—such as accessibility, vibrancy, amenities, and environmental quality—while also seeking deeper connections to industrial culture, including historical aesthetics, technological narratives, and shared social memory [14,15,16,17,18,19,20]. Effectively identifying and integrating these subjective user perceptions into the design and planning process remains a pressing and valuable area for further research.

The advent of the big data era has transformed performance evaluation methods in architectural projects. Digitalization and Internet of Things (IoT)-driven feedback mechanisms now enable space users to report issues in real time through mobile applications, indoor touchscreens, and similar interfaces [21,22]. These technologies optimize resource allocation, improve service responsiveness, and support evidence-based decision-making [23,24]. With continued technological progress, data-driven tools are increasingly recognized as indispensable for quantitative analysis and scientific urban planning [25]. Globally, the benefits of digitalization are evident, as tourism platforms such as TripAdvisor, Google Travel, Dianping, and Booking.com integrate user feedback systems. These serve a dual purpose: first, by providing project operators with real-time feedback for decision-making; and second, by offering prospective users reliable reference points. Nevertheless, the statistical methodologies underlying these systems remain rudimentary. Given its multifaceted and multi-objective nature, this field presents significant potential for further development.

Traditionally, questionnaires have been the primary tool for gathering user feedback. However, this method faces several limitations, including issues of timeliness, response bias, and difficulty in capturing longitudinal data in the past time [26,27]. Respondents’ attitudes and experiences may change over time, yet one-time surveys often fail to reflect these evolving perceptions in post-occupancy evaluations. Moreover, professionally designed questionnaires typically rely on closed-ended questions, which can restrict insights into users’ emotions, preferences, and motivations. In contrast, contemporary internet platforms offer an alternative channel for capturing more authentic, unsolicited user feedback. Advances in information filtering and content moderation now enable these platforms to exclude unconstructive or irrelevant comments, enhancing the overall quality of user-generated content [28,29]. Additionally, the digital “memory” of the internet supports retrospective analysis, allowing researchers to track user sentiment and engagement over extended periods. As such, these platforms provide project managers and researchers with powerful tools for collecting long-term user experience data in a more natural, dynamic, and continuous manner.

With the platform providing a robust data source, the next step is to employ appropriate methodologies for data analysis. This research hopes to analyze a substantial volume of user comments, identifying strengths and weaknesses across different aspects of the project and thereby offering feedback to decision-makers for timely corrective action. Public user reviews are largely qualitative in nature, and it is impractical to address each individual subjective evaluation separately. Consequently, aggregating information from a large number of individuals requires fuzzy processing to ensure that the output remains both relevant and reasonable. The Fuzzy Comprehensive Evaluation (FCE) method aligns precisely with these analytical requirements. At its core, FCE addresses the question of which grade an object belongs to under fuzzy and uncertain criteria [30]. It is particularly effective in quantifying qualitative information and managing the fuzziness inherent in subjective judgments [31,32,33,34]. Moreover, applying FCE in this evaluation process relies solely on user feedback data, thereby reducing interference from external factors. By contrast, the Analytic Hierarchy Process (AHP) fundamentally depends on pairwise comparisons, assigning scores from 1 to 9 to calculate the weight of each influencing factor [35]. This comparison process requires input from additional sources, such as expert assessments. Therefore, employing FCE for data statistics is more suitable for this study, as it better meets the requirements for objective and reliable data outputs.

Building on this premise, the present study utilizes user-generated online reviews as the primary data source to evaluate satisfaction with industrial heritage revitalization projects. Based on fuzzy mathematical principles, a scoring system is developed using a large volume of online review data to quantify user satisfaction. The 1933 Old Millfun in Shanghai, China, is selected as a representative case for empirical analysis. Review data are sourced from “Dazhong Dianping (Dianping)” (a completely free and open travel review website), one of China’s leading consumer review platforms, established in 2003 and recognized for its large user base and extensive data repository [36]. Using web scraping techniques, user reviews from the past four years are systematically collected. The text-mining software: KH Coder, was employed to extract high-frequency keywords and sentiment trends. These keywords are organized into thematic categories, which are further integrated into a multi-dimensional evaluation system. Finally, fuzzy mathematics is applied to quantitatively analyze user satisfaction and track performance trends over time. The analysis identifies both strengths and weaknesses across various dimensions, offering actionable insights for the future development and optimization of industrial heritage reuse.

2. Materials and Methods

Evaluating public engagement in industrial heritage revitalization is a complex and multidimensional task [6,37]. Researchers often face challenges in capturing the full range of influencing factors—particularly the subjective experiences of users, which are inherently difficult to quantify and interpret [38]. However, when a substantial volume of user feedback is collected, patterns of concern and engagement tend to stabilize within evaluations of the same project [39,40]. This consistency provides a foundation for constructing a more comprehensive and reliable set of evaluative dimensions to assess user engagement and satisfaction. Within this structured framework, user comments can be systematically categorized and quantified. By applying fuzzy mathematical theory, subjective data points are integrated into a coherent, multidimensional evaluation model. This study thus proposes the development of an evaluative paradigm that not only deepens understanding of public experience but also enables real-time monitoring. The model is designed to support the ongoing optimization of industrial heritage revitalization by identifying emerging issues and facilitating timely interventions.

First, data collection was carried out using extensive comments posted by individuals on public online platforms. Next, high-frequency terms related to the project were extracted and organized into influencing factors. Finally, the Fuzzy Comprehensive Evaluation (FCE) method was applied to integrate these influencing factors with the number of comments, thereby enabling the transition from qualitative insights to quantitative analysis. It is important to note that the influencing factors extracted from comments are not fixed across projects; rather, they evolve as users respond to different contexts. Accordingly, the three key elements in FCE analysis—the factor set, weight set, and comment set—also adapt dynamically. This approach employs a consistent methodology while allowing flexibility in data analysis, making it well-suited to industrial heritage regeneration projects with diverse emphases and thereby demonstrating strong universality. The specific methodology is as follows:

2.1. Data Collection

The study utilized a large amount of publicly available comments posted by users on the Dianping website to understand visitors’ satisfaction with industrial heritage revitalization projects. As an open-access website that supports data acquisition for research purposes, Dianping provides a valuable source of authentic user feedback. Comment data were extracted in reverse chronological order using web crawling techniques, enabling the systematic collection of user reviews within a certain period of time. Given the unstructured, free-form nature of the content, these reviews offer critical insights into user perceptions from both psychological and experiential standpoints.

From the perspective of environmental and architectural psychology, users tend to comment on aspects of a space that are particularly salient or memorable. This aligns with the concept of “selective attention”, which refers to the cognitive tendency to focus on information that confirms or challenges existing beliefs and expectations, while disregarding less relevant stimuli [41,42]. Selective attention plays a key role in shaping spontaneous user reviews, as users are unlikely to mention baseline conditions unless they deviate from expectations. For example, according to Maslow’s Hierarchy of Needs [43], basic physiological factors—such as air quality, rest areas, and environmental comfort—are fundamental to human survival and often go unnoticed unless compromised [44]. As a result, user-generated content tends to emphasize features that significantly exceed or fall short of expectations, making it a useful filter for identifying critical experiential factors in post-occupancy evaluations.

Despite the open and participatory nature of online platforms, which can introduce challenges such as subjectivity, misinformation, off-topic remarks, and promotional content, platform-level moderation mechanisms help to mitigate these risks. On Dianping, all user comments undergo manual review by platform administrators before publication. This screening process filters out inappropriate, irrelevant, or inaccurate content, thereby ensuring a baseline level of reliability and thematic coherence. While moderation cannot completely eliminate data noise, it significantly enhances the dataset’s validity for research purposes.

The combination of open-access, user-centered content, the filtering effect of selective attention, and platform-level moderation yields a unique and rich data source for evaluating user satisfaction and spatial performance. These reviews reflect not only the functional dimensions of post-occupancy experiences but also users’ emotional and psychological responses, providing a robust empirical foundation for constructing a multidimensional evaluation framework.

2.2. Data Extraction

After collecting a sufficient volume of user-generated comments, the dataset underwent a rigorous filtering process to ensure its integrity and analytical value. This step involved systematically identifying and eliminating invalid entries, including duplicated reviews—often generated through copy–paste behavior or repetitive content from the same user—along with promotional advertisements, irrelevant or off-topic remarks, and entries that were incomplete or unintelligible. The goal of this filtering phase was to retain only meaningful user feedback, thereby enhancing the overall quality and reliability of the dataset for subsequent analysis.

Following data cleaning, KH Coder—a robust tool widely employed in the fields of linguistics, social science, and user experience research—was utilized to conduct word frequency analysis. This quantitative content analysis technique involved parsing the cleaned dataset to identify the most frequently occurring terms. High-frequency keywords reflect recurring topics or aspects that users commonly engage with, indicating areas of heightened interest, concern, or satisfaction. These terms were then subjected to a semantic grouping process, wherein similar or related words were clustered into broader thematic categories. For instance, terms such as “crowded,” “entrance,” and “traffic” could be aggregated under a theme like “accessibility,” while words such as “style,” “design,” and “historic” might fall under “aesthetic and cultural expression.” This thematic mapping allows for a structured interpretation of otherwise fragmented user narratives, transforming qualitative feedback into analytically tractable dimensions. These keyword clusters thus serve as proxies for users’ core concerns and experiential priorities, offering insight into what matters most to visitors in their evaluation of industrial heritage spaces.

To deepen the analytical insight, semantic sentiment analysis was performed on the comment segments associated with each identified thematic category. This process aimed to assess the emotional valence—whether positive or negative—of user statements linked to specific project features. Advanced text mining and natural language processing techniques were applied to evaluate the polarity of user expressions, capturing both overtly emotional language (e.g., beautiful, disappointing) and more subtly evaluative terms within context. The sentiment analysis was conducted with attention to linguistic nuances, such as negations or modifiers, to enhance accuracy. Sentiment polarity scores were then mapped to the corresponding evaluation dimensions, enabling the visualization of satisfaction trends across key aspects of the user experience.

2.3. Data Analysis: FCE

To transform qualitative insights into quantifiable data suitable for modeling, a structured analytical framework was developed. This framework consists of three key components: a factor set (defining the evaluation dimensions), a weight set (indicating the relative importance of each dimension), and a comment set (representing quantified user responses). Together, these components enable the application of FCE techniques, facilitating a nuanced and multidimensional assessment of user satisfaction with the industrial heritage revitalization project. By linking subjective user perceptions with objective performance metrics, this approach yields a more holistic and actionable evaluation model.

2.3.1. Factor Set

Following the analysis of user-generated comment data, $n$ high-frequency phrases were extracted, each representing a distinct aspect of the user experience. These $n$ phrases were then categorized and thematically grouped into $m$ broader evaluation dimensions, where ($n$ > $m$). This classification process enabled the construction of a hierarchical evaluation framework, with $m$ representing the primary (first-level) dimensions and $n$ forming the secondary subcategories within each dimension.

The resulting framework provides a structured approach to understanding user feedback by organizing diverse experiential elements into a coherent system of evaluation. It also forms the foundation for subsequent quantitative analysis using multi-criteria evaluation techniques. A detailed breakdown of this two-tier classification structure is presented in

Table 1. Fuzzy Comprehensive Evaluation Method model.

Factor Set		$\mathbf{Weight} \mathbf{Set} \left(\mathit{w}\right)$	Comment Set
$\mathbf{First}-\mathbf{Level} \mathbf{Categorization} \left(\mathit{R}\right)$	$\mathbf{Sec}\mathbf{ond}-\mathbf{Level} \mathbf{Categorization} \left(\mathit{r}\right)$
${\mathit{R}}_{\mathit{m}1}$	$r_{m1n1}$	$w_{m1n1}$	Positive emotions (%)
	$r_{m1n2}$	$w_{m1n2}$	Positive emotions (%)
	$r_{m1n3}$	$w_{m1n3}$	Positive emotions (%)
	…	…	…
${\mathit{R}}_{\mathit{m}2}$	$r_{m2n1}$	$w_{m2n1}$	Positive emotions (%)
	$r_{m2n2}$	$w_{m2n2}$	Positive emotions (%)
	$r_{m2n3}$	$w_{m2n3}$	Positive emotions (%)
	…	…	…
${\mathit{R}}_{\mathit{m}3}$	$r_{m3n1}$	$w_{m3n1}$	Positive emotions (%)
	$r_{m3n2}$	$w_{m3n2}$	Positive emotions (%)
	$r_{m3n3}$	$w_{m3n3}$	Positive emotions (%)
	…	…	…
…	…	…	…

, which outlines the first-level evaluation dimensions and their corresponding second-level factor categories.

2.3.2. Weight Set

To quantify the relative importance of each evaluation factor, corresponding weights are assigned based on the frequency of keyword occurrences. Specifically, the weight ($w$) of each second-level factor is determined by the frequency ($v$) with which its associated high-frequency keyword appears in the user comment dataset (see Table 1). The underlying assumption is that a higher frequency of mentions indicates greater user concern or attention, and therefore, the factor should carry more influence in the overall evaluation.

For example, consider the second-level factor $r_{m1n1}$, its corresponding weight $w_{m1n1}$ is calculated using the following proportional formula:

$$w_{m1n1}={\displaystyle \frac{v_{m1n1}}{v_{m1n1}+v_{m1n2}+v_{m1n3}+\dots }}$$

(1)

This frequency-based weighting method offers a data-driven and objective approach to capturing the salience of each factor in users’ spatial experiences. It ensures that the aggregated evaluation reflects not only the presence of specific qualities but also the intensity of user attention devoted to them.

2.3.3. Comment Set

Due to the psychological phenomenon known as selective attention, individuals tend to focus on aspects of an experience that are either particularly positive or negative, while overlooking neutral or unremarkable elements [41,42]. As a result, user-generated evaluations—especially in open-comment formats—often exhibit what is known as the extreme effect [45]. This cognitive bias leads users to express either high satisfaction or strong dissatisfaction, while moderate or ambivalent responses are relatively uncommon.

To quantify sentiment data within this context, a categorical scoring system is employed in which each evaluation is classified as either positive or negative. The sentiment score for each evaluation factor is then calculated as the proportion of evaluators falling into a given sentiment category relative to the total number of valid responses. This proportional method allows for the transformation of qualitative emotional expressions into measurable data, making them suitable for integration into the FCE framework.

2.3.4. FCE Calculation Model

Through the above data, fuzzy calculations (Equation (2)) are performed according to the Fuzzy Comprehensive Evaluation Method to obtain the project’s score values in different dimensions, thereby reflecting the user satisfaction results.

$$f\left({\mathit{R}}_m\right)=r\ast w=\left[\begin{array}{c}\begin{array}{c}\begin{array}{c}{r}_{mn1}\\ r_{mn2}\\ \begin{array}{c}{r}_{mn3}\\ \begin{array}{c}\dots \end{array}\end{array}\end{array}\end{array}\end{array}\right]*\left[\begin{array}{c}\begin{array}{c}{w}_{mn1}\\ w_{mn2}\\ \begin{array}{c}{w}_{mn3}\\ \begin{array}{c}\dots \end{array}\end{array}\end{array}\end{array}\right]$$

(2)

3. Empirical Analysis

3.1. Case Selection

As of 2024, a total of 379 sites across China have been officially designated as industrial heritage sites, with Shanghai leading the list by hosting 30 sites—the highest among all cities (Figure 1) [46]. Among these, the Shanghai 1933 Old Millfun stands out as a landmark example of adaptive reuse in industrial architecture. Originally constructed in 1933, it is notable as the only large-scale former slaughterhouse in China whose architectural structure remains fully intact.

The main building is a four-story, 30,000 m² monolithic structure, characterized by a central atrium and a radial circulation system, with its four quadrants interconnected by a distinctive network of central bridges (Figure 2). In 2007, the building was repurposed into a creative industrial park while retaining its original architectural layout. The revitalized site now houses a diverse range of functions, including retail outlets, restaurants, cultural exhibitions, theatrical performances, and small-scale commercial enterprises. The revitalization project emphasized the aesthetic and historical value of the original concrete structure, integrating industrial materials into contemporary interior design. This approach highlights the spatial and structural logic of the former slaughterhouse while aligning with modern concepts of industrial heritage aesthetics and adaptive reuse (Figure 3) [47]. The 1933 Old Millfun thus serves as a paradigmatic case of industrial heritage revitalization, embodying both architectural continuity and functional innovation.

3.2. Data Acquisition and Analysis

The study employed the “Scrapy crawler framework”, developed in Python, to collect user reviews of the 1933 Old Millfun from the Dianping website. The dataset covered the period from 2020 to 2024 and comprised 3643 valid entries. Using KH Coder, high-frequency words were extracted from the collected reviews. These high-frequency words were then processed using the Latent Dirichlet Allocation (LDA) model to identify potential thematic terms. After screening, the high-frequency words were organized into 15 thematic keywords, which served as secondary factors in the evaluation system. Based on these 15 keywords, further categorization yielded six overarching dimensions—Project Accessibility, Cultural Legibility, Aesthetic Distinctiveness, Commercial Appeal, Facility Completeness, and Sense of Security—which were defined as first-level classification criteria (

Table 2. Online commentary high-frequency word categorization.

Factor Set		High-Frequency Words
First-Level Categorization $\left(\mathit{R}\right)$	Second-Level Categorization $\left(\mathit{r}\right)$
Project accessibility	Walking	Walking, Subway, Parking spaces, Driving
	Public transport
	Parking
Cultural legibility	Preservation of historical elements	Slaughterhouse, Shanghai Municipal Council, Cow Road, Corridor Bridge, Information Board
	Introduction to Historical Background
Aesthetic distinctiveness	Architectural style	Industrial style, Maze, Light and shadow space, Umbrella-shaped columns, Spiral staircase, Photograph, Filming locations
	Spatial layout
	Architectural construction
Commercial appeal	Diversity	Conferences, Weddings, Retail, Studios, Gastronomy
	Vitality
Facility completeness	Rest facilities	Number of toilets, Cleanliness, ‘Starbucks’, Convenience store, Seating area
	Basic amenities
	Environmental hygiene
Sense of security	Architectural components	Narrow, Steep, Gloomy, Terrifying
	Spatial atmosphere

). In parallel, the number of review entries associated with each thematic keyword, along with their corresponding positive sentiment counts, was statistically analyzed (

Table 3. Online commentary data statistics.

Factor Set		Number of Comments Mentioned $\left(\mathit{v}\right)$	Proportion of Positive Evaluations (%)
First-Level Categorization $\left(\mathit{R}\right)$	Second-Level Categorization $\left(\mathit{r}\right)$
Project accessibility	Walking	181	45
	Public transport	266	74
	Parking	93	61
Cultural legibility	Preservation of historical elements	501	100
	Introduction to Historical Background	377	32
Aesthetic distinctiveness	Architectural style	2899	92
	Spatial layout	2067	87
	Architectural construction	1466	91
Commercial appeal	Diversity	1022	41
	Vitality	901	32
Facility completeness	Rest facilities	89	59
	Basic amenities	95	47
	Environmental hygiene	47	90
Sense of security	Architectural components	27	47
	Spatial atmosphere	20	0

).

Project Accessibility

The project benefits from a well-developed urban bus network, with more than 10 routes available within a 500 m radius. Despite its location in a central urban area with multiple parking facilities, visitor feedback suggests only moderate satisfaction with accessibility, reflected in an average score of 61/100. As China’s economic hub, Shanghai offers an extensive transportation network, with the metro system serving as the preferred mode of travel for most out-of-town visitors [48,49]. However, the 1933 Old Millfun is located roughly 1 km from the nearest metro station, requiring a 15 min walk—an inconvenience for those already fatigued from long-distance travel. This relatively indirect connection contributes to a low pedestrian accessibility rating of 45/100.

Cultural Legibility

As an industrial heritage renovation, the 1933 Old Millfun prioritizes the conservation and expression of historical and cultural value, which are central to the project’s identity [6]. It remains China’s only fully preserved slaughterhouse, with its distinctive spatial and structural design attracting strong public interest, especially among visitors eager to engage with its industrial past. Architect Chongxin Zhao adopted a ‘subtractive’ approach, retaining the building’s original spatial logic and structural integrity while removing later decorative overlays, such as colorful paint and cladding. This minimalist restoration exposed raw concrete surfaces, enhancing the rugged historical atmosphere and deepening spatial and sensory engagement [50]. While these efforts earned praise for authenticity, many visitors noted the absence of explanatory signage for architectural components, limiting their understanding of the site’s historical and cultural layers. Consequently, despite positive perceptions of preservation quality, the project received a cultural legibility score of just 32/100.

Aesthetic Distinctiveness

The 1933 Old Millfun is widely recognized for its exceptional aesthetic character, resulting from the complete preservation of its architectural form and the rare typology of a slaughterhouse. Its basilica-like structure, combined with a functionally integrated industrial layout, distinguishes it from typical factories, where form follows function [51]. Visitors consistently highlighted the space’s visual impact, describing it as “maze-like”, “labyrinthine”, and a “play of light and shadow”. These qualities contributed to a visual satisfaction rating of 92/100, with the spatial layout scoring 87/100. Architectural features such as “flower columns”, “beamless floors”, and “spiral staircases”—absent from conventional civil architecture—stimulated curiosity and enhanced the sensory experience, leading to a curiosity satisfaction for construction rating of 91/100.

Commercial Appeal

As a mixed-use creative industrial park, commercial vibrancy is crucial to the project’s success in revitalizing the heritage site. The 1933 Old Millfun accommodates restaurants, retail outlets, studios, exhibition halls, and conference venues within a compact 30,000 m² building. However, visitor feedback indicates low satisfaction with commercial activity. Several units have ceased operations, leaving visible vacancies and inactive zones. This underutilization has diminished the site’s vitality, with commercial diversity and overall vibrancy receiving positive evaluations from only 41% and 32% of respondents, respectively.

Facility Completeness

Facility provision is generally well received. Public seating areas are distributed across the third and fourth floors, while essential amenities—such as a convenience store and a Starbucks café—are positioned on the ground floor for ease of access. These facilities are frequently used according to visitor reviews. Although the complex spatial layout has been described as a “maze”, restroom provision is adequate, with at least two restrooms per floor located at opposite corners. Most circulation and viewing areas are open-air, yet environmental cleanliness remains high, achieving a cleanliness score of 90/100.

Sense of Security

Originally designed for industrial livestock processing, the building incorporates tall concrete barriers and one-way circulation paths intended to control animal movement [50]. These features now inadvertently enhance visitors’ sense of security. However, narrow, steep staircases—once optimized for efficiency—pose accessibility challenges for elderly visitors, children, and those with mobility issues. Cultural perceptions of slaughterhouses in China, where death and decay are sensitive topics, also evoke mixed emotional responses. A small portion of visitors (20 recorded cases) reported feelings of unease or slight fear, heightened by the monochromatic concrete interiors. While this austere aesthetic reinforces the architectural atmosphere, it also contributes to discomfort for certain audiences.

3.3. The Calculation of FCE

Based on the mention frequency $v$ of the influencing factors obtained in Table 3, the calculation method in Section 2.3, “Weight set”, can be used to quantify the FCE model as shown in

Table 4. Data for Fuzzy Comprehensive Evaluation.

Factor Set		Weight Set $\left(\mathit{w}\right)$	Proportion of Positive Evaluations (%)
First-Level Categorization $\left(\mathit{R}\right)$	Second-Level Categorization $\left(\mathit{r}\right)$
Project accessibility	Walking	33.5%	45
	Public transport	49.3%	74
	Parking	17.2%	61
Cultural legibility	Preservation of historical elements	57.1%	100
	Introduction to Historical Background	42.9%	32
Aesthetic distinctiveness	Architectural style	45.1%	92
	Spatial layout	32.1%	87
	Architectural construction	22.8%	91
Commercial appeal	Diversity	53.1%	41
	Vitality	46.9%	32
Facility completeness	Rest facilities	38.6%	59
	Basic amenities	41.1%	47
	Environmental hygiene	20.3%	90
Sense of security	Architectural components	57.4%	47
	Spatial atmosphere	42.6%	0

.

According to Equation (2) of FCE, the comprehensive evaluation results for the Project accessibility, Cultural legibility, Aesthetic distinctiveness, Commercial appeal, Facility completeness, and Sense of security dimensions can be obtained as follows:

$$\mathrm{P}\mathrm{r}\mathrm{o}\mathrm{j}\mathrm{e}\mathrm{c}\mathrm{t} \mathrm{a}\mathrm{c}\mathrm{c}\mathrm{e}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}:f\left({\mathit{R}}_1\right)=r\ast w=\left[\begin{array}{c}\begin{array}{c}\begin{array}{cc}45& 55\\ 74& 26\\ \begin{array}{c}61\end{array}& 59\end{array}\end{array}\end{array}\right]\ast \left[\begin{array}{c}\begin{array}{c}33.5\mathrm{\%}\\ 49.3\mathrm{\%}\\ \begin{array}{c}17.2\mathrm{\%}\end{array}\end{array}\end{array}\right]=\left(62.04 37.96\right)$$

(3)

$$\mathrm{C}\mathrm{u}\mathrm{l}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{a}\mathrm{l} \mathrm{l}\mathrm{e}\mathrm{g}\mathrm{i}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}:f\left({\mathit{R}}_2\right)=\dots =(70.80 29.10)$$

(4)

$$\mathrm{A}\mathrm{e}\mathrm{s}\mathrm{t}\mathrm{h}\mathrm{e}\mathrm{t}\mathrm{i}\mathrm{c} \mathrm{d}\mathrm{i}\mathrm{s}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{n}\mathrm{e}\mathrm{s}\mathrm{s}:f\left({\mathit{R}}_3\right)=\dots =(90.17 9.83)$$

(5)

$$\mathrm{C}\mathrm{o}\mathrm{m}\mathrm{m}\mathrm{e}\mathrm{r}\mathrm{c}\mathrm{i}\mathrm{a}\mathrm{l} \mathrm{a}\mathrm{p}\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{l}:f\left({\mathit{R}}_4\right)=\dots =(36.78 63.22)$$

(6)

$$\mathrm{F}\mathrm{a}\mathrm{c}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y} \mathrm{c}\mathrm{o}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}\mathrm{t}\mathrm{e}\mathrm{n}\mathrm{e}\mathrm{s}\mathrm{s}:f\left({\mathit{R}}_5\right)=\dots =(60.37 39.63)$$

(7)

$$\mathrm{S}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{e} \mathrm{o}\mathrm{f} \mathrm{s}\mathrm{e}\mathrm{c}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{t}\mathrm{y}:f\left({\mathit{R}}_6\right)=\dots =(27.00 73.00)$$

(8)

So far, the evaluation results corresponding to each dimension can be obtained. On this basis, by further using the second-level operation of FCE (

Table 5. Fuzzy Comprehensive Evaluation for 1933 Old Millfun.

Factor Set	Weight Set $\left(\mathit{w}\right)$	Proportion of Positive Evaluations (%)
First-Level Categorization $\left(\mathit{R}\right)$
Project accessibility	5.4%	62.04
Cultural legibility	8.7%	70.80
Aesthetic distinctiveness	64.0%	90.17
Commercial appeal	19.1%	36.78
Facility completeness	2.3%	60.37
Sense of security	0.5%	20.00

), a comprehensive evaluation of the 1933 Old Millfun can be obtained, with the six dimensions as the main influencing factors. The calculation result is as follows:

$$1933 \mathrm{O}\mathrm{l}\mathrm{d} \mathrm{M}\mathrm{i}\mathrm{l}\mathrm{l}\mathrm{f}\mathrm{u}\mathrm{n}:f\left(\mathit{R}\right)=R\ast w= \left[\begin{array}{c}\begin{array}{c}\begin{array}{cc}62.04& 37.96\\ 70.80& 29.10\\ \begin{array}{c}90.17\end{array}& 9.83\\ 36.78& 63.22\\ 60.37& 39.63\\ 27.00& 73.00\end{array}\end{array}\end{array}\right]\mathrm{*}\left[\begin{array}{c}\begin{array}{c}5.4\mathrm{\%}\\ 8.7\mathrm{\%}\\ \begin{array}{c}64.0\mathrm{\%}\\ 19.1\mathrm{\%}\\ 2.3\mathrm{\%}\\ 0.5\mathrm{\%}\end{array}\end{array}\end{array}\right]=(75.73 24.27)$$

(9)

4. Result Analysis

According to the maximum membership principle in FCE, the evaluation results of 1933 Old Millfun for the six dimensions can be presented in Figure 4.

The project achieved a score of 62.04/100 for accessibility. Most visitors arrived via public transportation or on foot. While the distance from the nearest metro station remains a notable inconvenience, accessibility is partially offset by the availability of surrounding bus routes. However, such conditions are largely determined by urban infrastructure and macro-level planning, which lie beyond the scope of a single project. Within the site, parking management offers significant room for improvement. Currently, most parking spaces are occupied for extended periods by employees of small enterprises within the complex, who typically arrive early and depart late. This long-term occupation leaves few spaces for short-term visitors. A practical solution would be to reallocate staff parking to a nearby public parking lot, approximately 250 m away, while reserving a limited number of visitor-only spaces near the main entrance to improve accessibility.

In terms of cultural legibility, the project can be regarded as a partial success. Efforts in historical preservation are commendable: both the exterior façade and interior spaces retain high authenticity and historical integrity. However, interpretive communication remains insufficient. Given the richness of the site’s historical and architectural elements, more contextual information and explanatory displays are needed to convey its cultural significance. The building offers a profoundly immersive historical experience—where “every step tells a story”—yet interpretive aids akin to museum settings could further enhance understanding and emotional engagement, potentially improving the current satisfaction score of 70.80/100.

Aesthetic distinctiveness emerged as the project’s strongest attribute, attracting the largest share of reviews and earning a score of 90.17/100. This can be attributed to the building’s original 1933 design by British architect Balfours and the meticulous preservation achieved during renovation [52]. The result exemplifies restorative regeneration in industrial heritage, where architectural integrity and spatial aesthetics are faithfully retained and celebrated.

In contrast, commercial appeal scored considerably lower at 36.78/100, reflecting significant visitor dissatisfaction. Visitor expectations center on the diversity and quality of commercial offerings, shaped by precedents such as Beijing’s 798 Art Zone and Shanghai’s M50 Creative Park [53]. At 1933 Old Millfun, the current mix of commercial offerings shows limited synergy with the site’s distinctive historical and architectural character. Visitor feedback suggests that most people are primarily attracted by the visual appeal of the architectural spaces. However, the project’s current commercial offerings lack distinctive character and often conflict with the unique architectural environment. This mismatch substantially diminishes the inherent competitiveness of the on-site retail outlets. Establishments such as Starbucks, convenience store, Western-style restaurants, and Hong Kong-style cafés—ubiquitous throughout Shanghai—struggle to differentiate themselves in this setting. A more strategic integration could align commerce with heritage—for example, specialty meat-based restaurants referencing the building’s former slaughterhouse function, or vintage shops and photography studios capitalizing on its spatial uniqueness to create immersive experiences. Furthermore, the building’s underutilized spaces could be activated to address the surrounding area’s lack of supporting amenities (the surrounding area is mainly composed of old residential areas with few commercial and service facilities), thereby enhancing both service capacity and commercial attractiveness.

The sense of security dimension was among the least discussed, appearing in only 47 of the 3643 comments. Concerns centered on the atmospheric quality of the space—some visitors reported feelings of unease or even fear, linked to the austere material palette, low lighting, and historical associations with slaughter. These reactions are largely subjective and psychological rather than indicative of actual safety risks.

From an integrated perspective, the comprehensive evaluation across six key dimensions yields a final score of 75.73/100. This result reflects the application of FCE, which synthesizes multi-dimensional data while accommodating subjective variation in user feedback. Through its weighted approach, less salient factors—such as the infrequently mentioned sense of security—are appropriately de-emphasized, ensuring a balanced and representative assessment. The methodology not only strengthens objectivity and comparability in post-occupancy evaluation but also provides a quantitative framework for benchmarking similar industrial heritage reuse projects.

5. Discussion and Conclusions

This study adopts a user-centered approach, leveraging extensive unstructured user-generated content from online platforms to develop a user experience evaluation framework grounded in FCE. Drawing on 3643 valid user reviews of the Shanghai 1933 Old Millfun collected over the past four years, semantic analysis software was employed to extract high-frequency keywords, which were then synthesized into six key evaluation dimensions: Project Accessibility, Cultural Legibility, Aesthetic Distinctiveness, Commercial Appeal, Facility Completeness, and Sense of Security. By converting qualitative feedback into structured quantitative indicators, fuzzy set theory was applied to calculate the satisfaction performance of each dimension, enabling a precise diagnosis of the project’s operational strengths and weaknesses over time. A second-level fuzzy synthesis was subsequently conducted, yielding a comprehensive user satisfaction score for the project.

The methodological core of this study lies in its capacity to transform subjective public perceptions into a structured system of factor sets, weight sets, and comment sets, thereby bridging the gap between qualitative experience and quantitative evaluation. Crucially, the visitor-informed factor identification approach employed here offers strong transferability across diverse industrial heritage reuse contexts. This adaptability makes it suitable for comparative assessments and iterative improvements in similar projects. Moreover, the adoption of a standardized, data-driven scoring system facilitates inter-project benchmarking and provides a practical tool for ongoing monitoring, policy feedback, and strategic optimization in the sustainable development of industrial heritage sites.

Although this research adopts a user-centric perspective, public users have certain limitations. Individual users often focus on personal experiences and lack the ability to evaluate projects from a holistic or professional standpoint, as they are neither architects nor operators. In some cases, user perceptions may conflict with the original intent of industrial heritage regeneration or emphasize issues that are unavoidable or difficult to resolve. For example, a few visitors to the 1933 Old Millfun expressed dissatisfaction with the spatial ambiance. Modifying the project to accommodate such isolated preferences would compromise the unique value and character of the regenerated industrial heritage. This is where the statistical methodology of FCE proves particularly valuable: by filtering out noise from individual feedback, to mitigate the limitations inherent in public users.