Understanding Digital Sense of Place in Living Heritage Streets Through Multimodal Social Media Analysis: A Case Study of Songyang’s Ming–Qing Old Street

Abstract

Historic streets, as living heritage environments, preserve everyday cultural practices while facing increasing digital mediation in tourism and daily life. This study examines how a digital sense of place is constructed online in the Ming–Qing Old Street of Songyang, China. User-generated text and image data were collected primarily from Weibo, supplemented by user reviews from major travel platforms, including Dianping, Fliggy, Mafengwo, and Ctrip, and analysed through a multimodal framework. BERTopic was applied to identify thematic narratives in textual content, and ResNet-50 was used to classify visual scene elements in shared images, enabling an integrated interpretation of textual and visual representations. The results reveal four dominant dimensions of digital place perception: local food culture, living handicrafts, historic spatial fabric, and everyday atmosphere. Textual narratives emphasise emotional attachment and experiential interpretation, while visual representations highlight photogenic, performative, and shareable street scenes. The integration of these modalities forms a layered digital sense of place grounded in cultural continuity and daily life. The study demonstrates the value of multimodal social media analysis in understanding how living heritage streets are digitally represented and perceived, offering implications for sustainable heritage conservation, community-centred revitalisation, and data-informed cultural tourism management.

1. Introduction

With the proliferation of digital technologies and social media platforms, contemporary tourism practices and visitor experiences have undergone profound transformation. User-generated content (UGC) has become a primary channel through which tourists acquire information, form expectations, and narrate their experiences, thereby shaping destination images and influencing travel decisions [1,2]. Compared with traditional promotional materials, social media narratives—characterised by immediacy, affectivity, and situational expressiveness—enable more decentralised and plural representations of destinations.

These transformations are particularly salient in heritage tourism. Historic streets and cultural districts, once experienced primarily through embodied encounters, are now continuously reconstructed in digital environments as visitors document cultural atmospheres, everyday rhythms, and emotional impressions online [3,4,5]. Through multimodal expression, tourists increasingly act as “digital placemakers,” collaboratively participating in the production of place meanings and identities [6,7]. For living heritage environments, such mediated representations have growing implications for cultural sustainability, public engagement, and the long-term vitality of historic streets.

Against this backdrop, the classical notion of “sense of place” has expanded to incorporate digitally mediated interactions, giving rise to the emerging concept of digital sense of place. This concept highlights how place attachment, meaning, and identity are increasingly formed and negotiated through online practices [8,9]. Whereas traditional sense of place is primarily grounded in bodily presence, lived experience, and everyday interaction, digital sense of place refers to the formation, expression, and circulation of place meanings, emotional attachments, and spatial imaginaries through digitally mediated practices. In this study, it is understood not simply as the online extension of conventional place attachment but as a multimodal and platform-mediated construction of place meaning, shaped through textual narratives, visual representations, and participatory engagement on social media. Digital platforms therefore do not merely transmit information about places; they shape what becomes visible, shareable, and emotionally resonant in digital place perception. Within such multimodal and digitally mediated processes, different forms of expression foreground different dimensions of place: visual content tends to emphasise spatial appearance, atmosphere, and shareable scenes, while textual narratives more often articulate emotions, memories, and reflections on everyday experience. These layered and sometimes fragmented representations complicate issues of authenticity, interpretation, and sustainable heritage experience design [10,11].

Recent scholarship has advanced multimodal approaches to understanding digital place perceptions. Studies have examined emotional expressions, spatial perceptions, and visual preferences across tourism landscapes using topic modelling, deep learning–based image analysis, and integrated multimodal analytics [12,13,14,15,16,17,18]. GIS-based methods have further spatialised digital expressions, revealing intra-district variations, behavioural patterns, and spatial clustering in heritage settings [19,20,21,22]. These analytical innovations are supported by visualisation tools—such as heatmaps, temporal topic evolution, eye-tracking studies, and engagement metrics—that illuminate the dynamic ways tourists construct digital meaning [23,24,25]. While these studies have considerably expanded the methodological scope of digital place research, their analytical emphases remain uneven. Technically oriented approaches are effective in identifying emotional patterns, visual preferences, and spatial clustering, yet they often reduce place to an object of measurable perception rather than treating it as a lived and socially mediated process of meaning-making. At the same time, studies informed by digital placemaking and media-oriented perspectives have highlighted participation, circulation, and visibility in online environments but have less often examined how these dynamics intersect with everyday cultural practices and living heritage experience in small-scale historic streets. As a result, the relationship between multimodal representation and the socially grounded construction of digital sense of place remains insufficiently explained.

Beyond these methodological advances, current research still disproportionately centres on major cities and iconic destinations. Small-scale historic streets and living heritage environments—characterised by everyday cultural practices, local commercial life, and community-embedded heritage—remain underexplored [1,9,11], reflecting a persistent urban bias in current research and limiting the cross-cultural applicability of digital sense of place studies. Recent works in computational tourism analytics, digital heritage, and urban perception modelling highlight the need for more fine-grained, locality-specific studies capable of capturing both global patterns and situated cultural practices [26,27,28,29,30]. Against this background, how digital sense of place is constructed in living heritage environments remains underexplored. This is especially true for small-scale historic streets, where textual narratives and visual representations are closely entangled with everyday cultural practices, emotional attachment, and lived heritage experience.

To address these gaps, this study investigates the Ming–Qing Old Street of Songyang County, China, using multimodal social media content to examine how textual and visual expressions jointly construct a digital sense of place in a living heritage street. As a small-scale living heritage street in a county-level Chinese context, Songyang Ming–Qing Old Street preserves the historic streetscape and everyday commercial life. Moreover, the street has attracted sustained digital attention across social media and tourism platforms, where it is frequently represented through nostalgic and highly visual narratives. These characteristics make it a suitable case for examining how digital sense of place is constructed online. Integrating BERTopic-based textual modelling with ResNet-50 image recognition, the study analyses how tourists express emotions, articulate value perceptions, and construct place meanings in digital environments. Particular attention is paid to the distinct yet complementary roles of visual imagery and textual narratives in shaping place meaning. By establishing a multimodal analytical framework suited to small-scale living heritage streets, this research contributes new insights into digital heritage scholarship and offers practical implications for cultural communication, heritage management, and sustainable tourism planning [31,32].

This study contributes to digital heritage scholarship by:

(1)

constructing a multimodal methodological framework integrating textual and visual analytics;

(2)

revealing how textual narratives and visual representations jointly construct the digital sense of place of a small-scale living heritage street often overlooked in existing research;

(3)

offering practical implications for community-based conservation and sustainable cultural tourism.

2. Study Area, Data Sources, and Methods

2.1. Study Area

Songyang County is located in the western part of Lishui City, Zhejiang Province, eastern China. Situated in the mountainous region of southwestern Zhejiang, it is a county-level administrative unit characterised predominantly by hilly and mountainous terrain. Shaped by its natural topography and historical development trajectory, Songyang has remained relatively limited in urban scale. Its urban–rural spatial structure is comparatively intact, with traditional villages and historic streets continuing to occupy an important position within the regional spatial system. Long-term development in the county has been grounded in ecological conservation, the continuity of traditional settlements, and the preservation of local culture, resulting in a regional environment defined by an integrated system of traditional villages and a distinctive cultural ecology. Compared with rapidly urbanising areas, spatial renewal and tourism development in Songyang have followed a more gradual path, allowing a relatively strong continuity between historic spaces and everyday life to be maintained.

The Ming–Qing Old Street examined in this study is located in the central area of Xiping Subdistrict, the historic core of Songyang County (Figure 1). It constitutes a key street space within the traditional county-town layout. Extending from Taipingfang Road in the north to Changsong Road in the south and bounded by Xinhua Road to the west, the historic street covers an area of approximately 7.68 hectares. The overall spatial form is a typical linear street pattern, with a main axis of about 1.9 km in length. Buildings within the area are predominantly remnants from the Ming, Qing, and Republican periods, and the original street scale, building heights, and spatial fabric remain relatively well preserved. As an integral component of the county town’s historic spatial structure, the street functions not only as a site of cultural representation but also as an active space for everyday commerce and local services. It is an open street environment where residents’ daily activities and tourist visits intersect.

A defining characteristic of Songyang Ming–Qing Old Street is its strong feature of living heritage. Traditional businesses—including blacksmith workshops, scale-repair shops, Chinese herbal pharmacies, and barbershops—continue to operate on their original sites. These practices are not preserved as static displays but remain embedded in contemporary life through ongoing production, services, and social interactions. At the same time, changes in tourism development and consumption patterns have led to the gradual introduction of cafés, handicraft experience shops, and small creative studios. These emerging businesses coexist alongside traditional trades, forming a composite landscape of everyday living and consumption. The coexistence of old and new functions enables the street to sustain local life continuity while also undergoing incremental spatial renewal and functional transformation.

Through successive rounds of small-scale, gradual improvement, Songyang Ming–Qing Old Street has experienced upgrades to infrastructure and public cultural facilities. These interventions have sought to improve spatial quality while retaining the original street pattern and everyday atmosphere, allowing modern commercial and cultural elements to be incorporated without large-scale reconstruction. As a result, the street presents a spatial condition in which traditional and contemporary elements coexist, supporting diverse forms of daily use and tourist experience.

Within digital media environments, Songyang Ming–Qing Old Street has gradually developed a highly recognisable online image. On social media platforms, visitors and users frequently employ descriptors such as “lifestyle aesthetics,” “slow living,” “everyday vibrancy,” and “the integration of tradition and modernity.” These expressions tend to emphasise rhythms of daily life, spatial atmosphere, and lived scenes rather than focusing solely on architectural forms or visual landmarks [33]. They reflect visitors’ holistic perceptions of the street as an experiential living space [34]. It is precisely this combination of everyday life, living heritage practices, hybrid spatial experiences, and sustained digital visibility that makes Songyang Ming–Qing Old Street a suitable case for examining how digital sense of place in historic streets is constructed within online media environments.

2.2. Data Sources and Pre-Processing

This study draws on user-generated content (UGC) from major Chinese social media and tourism service platforms as its primary data source. UGC is widely recognised as an effective means of capturing tourists’ immediate perceptions, emotional expressions, and evaluative orientations following actual travel experiences. It has been extensively applied in studies of destination image, sense of place, and tourist behaviour. Compared with traditional survey or interview data, social media data offer several advantages, including large sample sizes, long temporal coverage, spontaneous expression, and the availability of multimodal information such as text and images. These characteristics make UGC particularly suitable for examining the formation of sense of place within digital media environments.

Data were collected from several widely used Chinese social media and travel-related platforms, including microblogging services, online travel communities, and tourism review platforms. These platforms differ in user composition, usage contexts, and content production logics, encompassing travel narratives, consumption reviews, photographic records, and affective expressions. Their combined use helps enhance the diversity and representativeness of the dataset. Using “Songyang Old Street” as the search keyword, a systematic retrieval was conducted across all platforms, covering the period from September 2016 to July 2025. A total of 2195 original text-and-image posts were obtained.

An initial screening procedure was conducted to ensure the relevance and reliability of the dataset. First, duplicate posts were removed. Second, advertisements, commercial promotions, and posts unrelated to the study area were excluded. Content relevance was determined by examining whether the posts explicitly referred to Songyang Ming–Qing Old Street or contained textual or visual descriptions associated with the street environment. Finally, manual verification was conducted to confirm that the remaining posts reflected experiences or observations related to the study site. After this screening process, 1438 valid textual posts were retained as the analytical sample.

To further ensure that the dataset primarily reflected visitor experiences rather than residents’ daily activities, posts were examined for textual and contextual cues indicating travel behaviour. Posts containing explicit travel narratives, visit descriptions, tourism recommendations, or references to short-term experiences (e.g., “visiting,” “travel,” “weekend trip,” or “recommended place”) were considered likely to represent visitor-generated experiences. In contrast, posts that appeared to reflect routine local life without clear indications of tourism-related experiences were not prioritised during the screening process.

Textual data were pre-processed using Jieba tokenisation, supplemented by the Harbin Institute of Technology stop-word list and a project-specific stop-word dictionary to remove irrelevant words and symbols. Key concepts were manually normalised to ensure semantic consistency (e.g., different expressions referring to the same local food were unified under a single term). Terms weakly related to place perception were further filtered out. The final clean corpus comprised 10,498 tokens and served as the input for subsequent topic modelling.

Visual data were extracted from the images associated with the 1438 valid posts. Following initial screening for link validity, image clarity, and content relevance, combined with manual verification, a final image dataset of 10,937 tourist photographs closely related to Songyang Ming–Qing Old Street was constructed. To maintain contextual consistency between textual and visual data, each image was linked to its original post, publication time, and platform source, allowing visual analysis results to be interpreted in relation to specific textual expressions.

During image pre-processing, all photographs were organised according to a unified standard. Image paths were stored in a consistent absolute directory structure, and image formats and dimensions were standardised to ensure stability and consistency during feature extraction. This procedure improved the performance and computational efficiency of the ResNet-50 model in scene recognition tasks and provided a reliable technical foundation for subsequent multimodal analysis.

2.3. Research Methods and Analytical Framework

This study adopts a multimodal mixed-methods approach to examine the construction of digital sense of place associated with Songyang Ming–Qing Old Street from both textual narratives and visual representations. Drawing on user-generated content from social media, the analytical framework integrates topic identification and scene perception to reveal how visitors articulate emotional experiences, value judgements, and spatial imaginaries of historic streets in digital environments (Figure 2).

Textual analysis was conducted using the BERTopic model for topic identification. BERTopic is a topic modelling approach based on Transformer architectures, which employs pre-trained language models such as BERT to convert texts into high-dimensional semantic embeddings. This enables the model to capture contextual meaning and semantic relationships within short and informal social media texts [35,36]. Compared with traditional topic models such as Latent Dirichlet Allocation (LDA), BERTopic demonstrates higher robustness and interpretability when applied to user-generated content.

Preliminary testing showed that LDA produced topics with limited differentiation, while BERTopic’s default HDBSCAN clustering algorithm tended to generate only a small number of broad, highly aggregated themes, which was insufficient for capturing the fine-grained structure of visitors’ perceptions targeted in this study. To obtain a more differentiated set of “micro-topics” for subsequent thematic consolidation, K-means was adopted within the BERTopic framework. Multiple K values were tested iteratively, ranging from relatively small to larger numbers of clusters. Lower K values tended to produce overly aggregated topics, whereas higher K values (e.g., approaching 28) led to substantial redundancy across clusters. The final choice of K = 20 was therefore made as a balance between thematic granularity, interpretability, and suitability for later manual merging of semantically related topics. Based on this setting, a bottom-up topic discovery strategy was adopted. The analytical procedure involved four main steps:

• Generating sentence-level semantic embeddings by inputting textual comments into a pre-trained Transformer model;
• Applying UMAP for non-linear dimensionality reduction to preserve key semantic structures;
• Applying K-means clustering to generate 20 fine-grained “micro-topics,” thereby enhancing thematic resolution and supporting subsequent thematic consolidation;
• Conducting qualitative interpretation and thematic consolidation to merge related micro-topics into four coherent and conceptually meaningful core themes.

This strategy improved both the granularity and interpretability of topic identification. In addition, by incorporating post timestamps, temporal analysis was conducted to track changes in theme prominence over time, allowing the exploration of the spatiotemporal dynamics of digital sense of place.

Visual analysis was carried out using the ResNet-50 convolutional neural network to perform scene recognition on 10,937 visitor photographs. ResNet-50 is a deep residual network comprising 50 convolutional layers, in which residual connections help mitigate vanishing gradient problems and enable stable training of deep architectures [17].

The main processing structure included the following components:

• An input layer receiving 224 × 224 RGB images with standardised normalisation to ensure data consistency;
• An initial convolutional layer (7 × 7 kernel, stride 2) for low-level feature extraction, followed by stacked residual blocks (1 × 1, 3 × 3, and 1 × 1 convolutions) that balance feature compression and expansion while capturing complex visual elements;
• A 3 × 3 max-pooling layer for down-sampling and enhanced translational invariance, followed by global average pooling to aggregate features into 2048-dimensional vectors and reduce overfitting risk;
• A fully connected layer mapping features to 365 scene categories, with Softmax used to output classification probabilities;
• Parameter optimisation through gradient-based learning supported by the residual network structure to improve recognition accuracy.

A pre-trained Places365-CNN model was used to classify all images into 365 predefined scene categories, including labels such as “restaurant,” “alley,” and “street market”. This scene-based visual interpretation provides a complementary perspective to textual analysis and enables an integrated understanding of how historic streets are visually represented and experienced in digital space.

3. Text-Based Analysis of Digital Sense of Place

3.1. High-Frequency Word Characteristics of Digital Sense of Place in the Historic Street

To identify the key elements that visitors focus on when engaging with Songyang Ming–Qing Old Street in digital media, a word-frequency analysis was conducted on the comment corpus after tokenisation and stop-word removal. The top 20 most frequently occurring words were extracted (

Table 1. High-frequency words related to the digital sense of place of Songyang Old Street.

Rank	Word	Frequency
1	History	278
2	Life	264
3	Ming–Qing	208
4	Tradition	199
5	Buy	160
6	Architecture	145
7	Recommend	133
8	Baixian Noodle Shop	128
9	Ancient town	120
10	Blacksmith shop	117
11	Braised salt chicken	114
12	Zhuangyuan pastry	110
13	Street	110
14	Atmosphere	104
15	Tourist	104
16	Barbershop	104
17	Along the River During the Qingming Festival	103
18	Blacksmithing	97
19	Like	91
20	Dengzhan pastry	90

). During this process, highly repetitive place-referential terms such as “Songyang Old Street,” “Old Street,” and “Songyang” were excluded as stop words to avoid analytical bias.

Overall, the distribution of high-frequency words reveals a clear multi-dimensional pattern in visitors’ digital expressions. On the one hand, terms such as “Ming–Qing,” “architecture,” and “tradition” appear frequently, indicating that the historical background, architectural periods, and cultural attributes of the street remain central reference points for visitors’ understanding of the area. On the other hand, experiential words such as “life,” “atmosphere,” and “like” also occupy prominent positions, suggesting that the street is not perceived merely as a historical relic, but as a space where everyday life continues to unfold.

At the same time, many high-frequency words point to specific and tangible local elements. These include local food-related terms such as Baixian Noodle Shop, braised salt chicken, Zhuangyuan pastry, and Dengzhan pastry; traditional service and craft-related places such as blacksmith shops and barbershops that remain in operation; and culturally symbolic references such as Along the River During the Qingming Festival. Rather than abstract concepts, these words are closely tied to concrete places, activities, and sensory experiences. This pattern indicates that visitors tend to articulate their overall perception of the historic street through identifiable, experience-based details in digital media.

Taken together, the high-frequency word characteristics suggest that the digital sense of place of Songyang Ming–Qing Old Street is not constructed solely around historical architecture. Instead, it emerges from the interplay of historic space, everyday life, local food culture, and living traditional businesses. Among these elements, food-related terms are particularly prominent, highlighting the important role of culinary experience in shaping visitors’ digital perceptions of the historic street.

3.2. Core Themes of Digital Sense of Place and Their Meanings

Building on the high-frequency word analysis, topic identification was further conducted using the BERTopic model. Under the predefined parameters, the model identified 20 semantically coherent micro-topics (Topic 0–Topic 19). Each topic is represented by a set of keywords with different weights, where higher weights indicate stronger semantic contributions to the topic (Figure 3).

Interpretation of the keyword sets and their corresponding comment texts allows an initial understanding of the thematic orientation of each micro-topic. For example, keywords such as “Yangjiatang Village” and “Zhongshuge Bookstore” in Topic 0 clearly point to surrounding rural tourism destinations connected to the old street. In contrast, several topics simultaneously include terms such as “history,” “blacksmith shop,” “atmosphere,” and “braised salt chicken,” indicating the coexistence of multiple experiential elements within a single theme.

Further examination shows substantial semantic overlap among many micro-topics. Rather than reflecting a limitation of the model, this overlap is closely related to how visitors articulate their experiences. Many comments naturally combine observations of historic architecture, traditional crafts, spatial atmosphere, and food experiences within the same narrative. A typical comment may describe Ming–Qing buildings, observe a working blacksmith shop, reflect on the everyday atmosphere, and recommend a local dish at the same time. Such multi-layered expressions lead to frequent co-occurrence of terms like “history,” “blacksmith shop,” “atmosphere,” and “braised salt chicken” across different topics. These words function as “bridging terms” that connect multiple themes and reflect the integrated nature of visitors’ lived experiences.

To improve interpretability while retaining thematic detail, micro-topics with high semantic similarity were consolidated based on cosine similarity measures, supported by visual inspection of the similarity heatmap and hierarchical clustering results (Figure 4 and Figure 5). Through this bottom-up aggregation process, the 20 micro-topics were ultimately merged into four macro-themes with clearer boundaries and richer conceptual meaning (

Table 2. Final themes and proportion of comments.

Final Theme	Topics	Number of Comments	Proportion (%)
Historic architecture and atmospheric experience	1, 4, 5, 9, 12, 17	473	34.00
Local food culture	2, 8, 10, 13, 14	450	32.35
Living crafts and visitor feedback	3, 6, 7, 11	255	18.33
Surrounding villages and emerging businesses	0, 16, 18, 19	213	15.31

).

Among the four macro-themes, historic architecture and atmospheric experience constitutes the core component of the digital sense of place of Songyang Old Street, accounting for the largest share of comments. Keywords within this theme focus on the Ming–Qing period, architecture, streets, time, and stories. This suggests that visitors are not merely observing physical structures, but engaging in a form of temporal experience. The preserved street layout and timber buildings provide an immersive historical setting, while terms such as “everyday atmosphere” and “sense of life” transform static architectural perception into emotionally resonant spatial memory. This pattern aligns closely with local conservation practices that emphasise maintaining authenticity, and visitors’ digital narratives revolve around this perceivable historical texture.

The local food culture theme reflects strong visitor interest in regional culinary traditions. As a long-established marketplace and dining hub, the old street’s local snacks—such as braised salt chicken, Zhuangyuan pastry, and Dengzhan pastry—serve as key entry points for experiencing everyday life and function as sensory anchors in visitors’ memories. In digital space, these foods are not simply consumed but frequently photographed and shared, becoming symbolic representations of place. Through recommendations and “check-ins,” visitors contribute not only to personal travel memories but also to the collective construction of local identity.

The theme of living crafts and visitor feedback highlights the emphasis on living heritage in Songyang. Traditional workshops such as blacksmith shops and cotton-fluffing stalls are valued not only as objects of visual observation, but as spaces of ongoing production and interaction. This non-performative, everyday continuity is perceived as an authentic form of local life and plays a key role in shaping a sense of place. At the same time, this theme incorporates visitors’ reflections on local lifestyles and degrees of commercialisation, adding nuance to the understanding of place complexity.

Although surrounding villages and emerging businesses account for a smaller proportion of comments, this theme plays an important role in extending the spatial and experiential boundaries of the old street. It indicates that the digital sense of place of Songyang Old Street is networked rather than self-contained. On the one hand, nearby villages such as Yangjiatang form complementary destinations for deeper exploration. On the other hand, the emergence of cafés and new-style beverage shops reflects local strategies centred on slow living and creative industries. The juxtaposition of new businesses within an old spatial setting attracts younger visitors and stimulates discussion around the blending of old and new, contributing significantly to the evolving narrative of place.

3.3. Temporal Dynamics of Core Themes

To capture the temporal evolution of the digital sense of place in Songyang Old Street, changes in the volume of posts associated with each of the four macro-themes were examined over time. Based on the distribution of micro-topics within each theme, a time-series line chart was generated to visualise their dynamic trajectories (Figure 6). The results show that fluctuations in thematic attention are closely aligned with on-site development strategies and event cycles, reflecting how visitors’ digital focus responds to seasonal patterns, specific events, and broader media effects.

The theme of historic architecture and atmospheric experience displays a clear cyclical pattern. Since 2018, attention to this theme has tended to rise in the first half of each year and decline in the second half, corresponding broadly to peak and off-peak tourism seasons. A noticeable drop occurred between late 2021 and early 2022, which coincides with a period of intensive infrastructure renovation within the old street. Construction activities during this phase may have temporarily constrained visitor experience and reduced content sharing. Following the completion of renovation works, attention to this theme recovered from 2022 onward and reached new peaks in 2023–2024, suggesting that conservation-oriented improvements enhanced the perceived authenticity and attractiveness of the street.

The local food culture theme also exhibits seasonal variation alongside a general upward trend. Mentions increased steadily from 2022 and peaked in mid-2023. This pattern reflects sustained visitor interest in local cuisine and is closely associated with the timing of traditional festivals such as the Dragon Boat Festival and the Mid-Autumn Festival. These events are often accompanied by the promotion of seasonal foods, including Zhuangyuan pastry and braised salt chicken, which in turn stimulates food-related sharing on social media. As a result, local cuisine becomes an important medium through which visitors experience everyday life and construct cultural memory.

In contrast, the theme of living crafts and visitor feedback shows a continuous upward trajectory from 2022 onward, reaching its highest level toward the end of the study period. This trend indicates that traditional crafts such as blacksmith workshops are increasingly perceived as authentic expressions of everyday local life, particularly through ongoing production and direct interaction. Growing attention to this theme is also linked to policy support and promotional efforts targeting long-established local businesses, which have enhanced their visibility in digital space.

The evolution of surrounding villages and emerging businesses is more strongly event-driven. For example, fluctuations in attention to the No. 9 Café are closely associated with a media filming event in 2023, demonstrating how celebrity exposure can rapidly redirect visitor attention in the short term. Such patterns highlight the sensitivity of this theme to external stimuli and its role in shaping episodic peaks in the digital representation of place.

4. Image-Based Analysis of Digital Sense of Place

4.1. Classification of Visual Scenes

After scene recognition of tourist photographs from Songyang Ming–Qing Old Street, the twenty most frequent visual scene categories were identified (

Table 3. Top 20 visual scene categories identified from the image dataset of Songyang Ming–Qing Old Street.

Rank	Scene Category	Percentage (%)
1	Bazaar/Outdoor	7.81
2	Alley	7.39
3	Medina (Historic Street Scene)	5.60
4	Delicatessen (Food Scene)	4.10
5	Repair Shop	3.48
6	Bakery/Shop	3.13
7	Slum/Traditional Housing	2.65
8	Shopfront	2.59
9	Coffee Shop	2.54
10	Bazaar/Indoor	2.30
11	Doorway/Outdoor	1.80
12	Barn Door/Traditional Gate	1.68
13	Restaurant Kitchen	1.67
14	Diner/Outdoor	1.53
15	Temple/Asia	1.39
16	Art Studio	1.29
17	General Store/Outdoor	1.14
18	Ice Cream Parlour	1.13
19	Jewellery Shop	1.10
20	Hardware Store	1.08

). These scenes are mainly associated with street spaces, markets, shops, dining settings, and handicraft workshops. Together, these everyday life and commercial environments account for approximately 55.4% of all recognised scenes, indicating that a relatively small number of core visual elements dominate visitors’ visual perceptions of the historic street.

Because the original 365 scene categories generated by the Places365 model are highly detailed and fragmented, they are not suitable for direct interpretation in the context of historic street environments. Therefore, the recognised scenes were further reorganised. Drawing on existing scene classification approaches [35,36] and considering the spatial and functional characteristics of Songyang Old Street, all recognised scenes were consolidated into seven broader categories. The classification scheme and descriptions are presented in

Table 4. Visual scene categories and descriptions.

Scene Category	Included Content	Description
Street and market spaces	Alleys, streets, shopfronts	Pedestrian movement and everyday street life
Dining and food scenes	Snack stalls, local dishes, beverages	Local cuisine and dining experiences
Traditional handicrafts and retail	Hardware shops, repair stores, grocery and fabric shops	Living crafts and small local businesses
Cultural and religious buildings	Museums, galleries, studios	Heritage buildings and cultural spaces
Residential and living spaces	Bedrooms, kitchens, lofts, guesthouses	Daily life and accommodation
Transportation and infrastructure	Parking areas, roads, signs, passages	Mobility and basic support facilities
Others	Warehouses, special or unclassified buildings	Special functions or unclassified spaces

.

4.2. Visual Expression of Digital Sense of Place

To further examine the visual dimension of digital sense of place, the frequencies and proportions of the seven scene categories were calculated based on the recognition results of 10,937 images (

Table 5. Frequency and proportion of visual scene categories.

Scene Category	Frequency	Percentage (%)
Street and market spaces	2268	36.13
Dining and food scenes	1505	23.98
Traditional handicrafts and retail	1037	16.52
Functional and interior spaces	503	8.01
Cultural and religious buildings	420	6.69
Residential and living spaces	284	4.52
Others	159	2.53
Natural and landscape spaces	67	1.07
Transportation and infrastructure	34	0.54

). This analysis reveals the relative importance of different visual elements in tourists’ digital representations.

Street and market spaces clearly dominate the visual representations. Scenes such as alleys, bazaars, and shopfronts appear most frequently, suggesting that visitors’ visual attention is primarily drawn to the overall spatial texture of the old street. These images highlight the continuity of the street layout, dense everyday activities, and strong sense of historical atmosphere.

Dining and food scenes constitute the second most prominent visual category, closely aligning with the findings from the text-based analysis. Restaurants, bakeries, and cafés, along with images of local foods such as weiyan chicken and zhuangyuan cakes, form a set of recognisable culinary symbols. The high visibility of food-related images underscores the role of food as a key attraction and as a visual marker of local vitality and identity.

Traditional handicraft and retail scenes further reflect the “living heritage” character of Songyang Old Street. Blacksmith shops, fabric stores, and small local retailers are frequently photographed. These scenes not only signal historical continuity but also provide visually engaging content for visitors. The focus on such spaces reinforces the perception of the old street as a place where traditional practices remain part of everyday life.

Other categories, although less prominent, enrich the overall visual structure of place perception. Functional and interior spaces and cultural buildings add depth to the visual experience, while residential scenes convey a sense of everyday authenticity. In contrast, the very low proportions of natural landscapes and transportation facilities indicate that visitors’ visual attention is strongly centred on the human-scale, built, and cultural environment rather than on natural or infrastructural elements.

Overall, the visual digital sense of place of Songyang Ming–Qing Old Street is mainly shaped by immersive street experiences, the symbolic presentation of local food, and the authentic display of traditional handicrafts. This visual structure corresponds closely with the preserved historical environment, active local commerce, and visitors’ visual sharing preferences in the digital media context. It provides quantitative evidence for understanding how the spatial vitality and image of the old street are constructed through visual representation.

5. Conclusions

5.1. Digital Sense of Place as an Integrated Experiential Process

The combined analysis of textual narratives and visual representations shows that the digital sense of place of Songyang Ming–Qing Old Street is not shaped by any single element. Instead, it emerges as an integrated structure centred on everyday lived experience. In textual expressions, food, handicrafts, and everyday atmosphere appear repeatedly. In visual content, street spaces, dining scenes, and traditional businesses dominate. Together, these representations suggest that visitors are not responding to isolated cultural features or spatial forms, but to scenes of life that are actively unfolding.

More specifically, the textual theme of local food culture corresponds closely to the visual prominence of dining and food scenes, indicating that culinary experience functions both as an emotional narrative anchor and as a highly shareable visual symbol of place. Similarly, the textual theme of living crafts and everyday production is echoed by the frequent appearance of traditional handicraft and retail scenes in photographs, showing that visitors value these spaces not only for their visual distinctiveness but also for the everyday production, interaction, and authenticity they embody. In this sense, textual narratives tend to articulate meaning and affect, while visual representations make these meanings tangible and communicable. Their combination helps explain how digital sense of place is constructed through the interplay of emotion, everyday practice, and visible spatial experience.

This pattern is closely related to the fact that Songyang Ming–Qing Old Street continues to function as a space for daily life and local commerce. Unlike highly touristic historic districts that are oriented mainly toward sightseeing consumption, Songyang Old Street has not been fully transformed into a display-oriented environment. A substantial degree of everyday continuity remains. As a result, digital expressions often begin with concrete experiences. Through eating, walking, observing, and interacting, visitors translate historic space into a lived place that can be perceived and remembered. This helps explain why local food and living crafts occupy such a central position in the digital sense of place.

It is also important to note that the “text–image tension” observed in this study should not be understood as a flaw or imbalance. Rather, it represents an inherent characteristic of digital place-making. Textual narratives tend to be inward-looking and emotional, emphasising temporality, atmosphere, and personal feelings. Visual sharing, by contrast, is more performative and outward-facing, highlighting spaces and activities that are easily seen and circulated. Together, these two modes create a multi-layered digital image of Songyang Old Street. This image combines emotional depth with communicative appeal. Instead of weakening the sense of place, this tension offers a useful lens for understanding how historic districts are constructed and experienced in digital media environments.

More broadly, the case of Songyang Ming–Qing Old Street suggests that the deep involvement of digital media in tourism does not necessarily lead to the loss of everyday life in historic districts. On the contrary, daily experiences can be reinforced and amplified through digital expression. This finding provides an alternative perspective for understanding digital sense of place in small cities and non-iconic destinations.

5.2. Practical Implications: Key Considerations for Historic District Renewal in the Digital Media Context

Based on the identified characteristics of digital sense of place in Songyang Ming–Qing Old Street, several practice-oriented implications can be drawn for historic district renewal under digital media conditions.

(1)

Historic district renewal should prioritise the perceptibility of everyday continuity rather than focus solely on visual image enhancement. The results indicate that visitors’ attention in digital media is directed less toward individual landmarks and more toward overall spatial rhythm, daily activities, and lived atmosphere. Renewal practices should therefore respect street scale, patterns of use, and everyday behaviour. Small-scale and incremental interventions are more effective in maintaining the perception of historic districts as lived spaces, rather than transforming them into display-oriented settings designed primarily for photography and check-ins.

(2)

Local food should be treated as a key medium in the construction of sense of place, rather than as a secondary tourism product. In Songyang Old Street, food-related elements appear prominently in both text and images, suggesting that culinary experience is a major entry point through which visitors understand and remember the place. In practice, supporting locally operated food businesses, strengthening the everyday and historical narratives behind local dishes, and guiding their cultural expression on digital platforms can help position food as a connector between daily life, local memory, and tourism experience.

(3)

The protection of living crafts and traditional service businesses should avoid excessive display or performance-based transformation. Visitors’ interest in blacksmith shops and barbershops stems from their continued integration into real production and everyday life. The value of these practices lies not in being watched but in their ongoing operation as ordinary activities. Renewal strategies should therefore focus on institutional support, spatial accommodation, and business environment improvement to sustain their normal functions, allowing “life in progress” to remain a core source of sense of place.

(4)

Historic district renewal should not be approached as an isolated spatial intervention but as part of a broader, networked experience system. Digital narratives indicate that the sense of place of Songyang Old Street is not generated in isolation. It is shaped through connections with surrounding villages and emerging new businesses. In practice, walking systems, thematic routes, and digital guidance tools can be used to link historic streets with nearby spaces. Such integration helps strengthen the continuity and coherence of sense of place at the regional scale.

5.3. Limitations and Directions for Future Research

Although this study provides a systematic analysis of the digital sense of place of Songyang Ming–Qing Old Street based on multi-platform social media data, several limitations should be acknowledged. First, user-generated content is produced voluntarily, and its contributors may be biassed in terms of age and media use habits. As a result, the data may not fully represent the perceptions of all visitor groups. The classification of users into tourists and locals was based on heuristic interpretation of their posted content, which may introduce potential sampling bias and affect the interpretation of the findings. Moreover, the Chinese social media platforms used in this study differ in user composition, communicative conventions, posting behaviours, content styles, and data accessibility, which may further influence the representativeness of the dataset and the patterns identified in the analysis. Data scraping constraints and manual screening procedures, while necessary for ensuring relevance and usability, may also have affected the final composition of the corpus. Second, the analysis focuses primarily on textual and visual content, while direct observation of on-site behaviour and embodied experience remains limited.

Future research could build on multimodal data analysis by incorporating surveys, interviews, or behavioural tracking data. This would allow for cross-validation between online expression and offline experience. Comparative studies across different types of historic districts would also help test the broader applicability of the findings presented in this study.