Spatiotemporal narrative represents a critical focus in literary studies, concerned not only with how time and space function as active narrative devices but also with their profound implications for interpreting and structurally analyzing classical texts. In traditional Chinese literature, time and space serve as more than mere backdrops; they are core narrative elements that carry cultural memory, manifest situational logic, and connect plotlines, thereby directly shaping textual meaning and interpretive possibilities.

While theoretical foundations such as Bakhtin's "chronotope" and narratological theories of time (Zheng 2021) have been established, practical analyses of classical texts often treat temporal and spatial dimensions in isolation or remain confined to abstract textual interpretation (Wei et al. 2022). This makes it difficult to capture their dynamic interrelations and holistic configuration within narrative progression. Existing methodologies, largely tailored to modern genres (Yang / Luo 2014), lack a broadly applicable analytical framework capable of revealing the intricate spatiotemporal interactions in premodern literature—especially for texts characterized by implicit structures and intertwined temporal-spatial information. There is thus a pressing need for methodological innovation.

Event Knowledge Graphs (EKGs), a core technology in digital humanities (Rospocher et al. 2016), offer a promising pathway for the structured representation and in-depth mining of classical literary knowledge. For classical Chinese texts with often implicit narrative logic and fragmented information, EKGs can effectively uncover latent event chains (Liu et al. 2024), reconstruct character trajectories, and reveal semantic connections that traditional philological methods struggle to capture. This provides a new analytical framework for understanding the narrative structure and internal logic of classical works.

Addressing this gap, this study employs EKGs as a methodological tool, leveraging their intuitive, dynamic, and extensible characteristics to enable the visualization and structured analysis of spatiotemporal narratives in classical texts. Focusing on the Ming dynasty travelogue The Travel Diaries of Xu Xiake—a representative work noted for dense temporal markers, detailed geographical records, and multi-layered events—this research proposes a novel spatiotemporal narrative analysis framework based on EKGs. The framework constructs a structured element system centered on Xu Xiake's life trajectory, forming a "time-space-person-event" relational network and developing an EKG that integrates multidimensional associations among time, space, and emotion. Through fine-grained event extraction and relational modeling, the study systematically reveals the interactive mechanisms between these dimensions, overcoming the limitation of treating them in isolation. This work also contributes methodological insights for deeper exploration of the narrative wisdom and structural features of classical Chinese literature.

The specific contributions of this research are threefold:

Development of an EKG Construction Methodology for Classical Texts: We fine-tune a BERT model for Named Entity Recognition (NER) and implement the graph on the Neo4j platform. After evaluating BERT-Ancient-Chinese, GuwenBERT, and SikuBERT, GuwenBERT demonstrated superior performance and was adopted for large-scale entity extraction. To address challenges posed by classical Chinese—such as lack of punctuation, flexible syntax, and semantic compactness—we implement a character-level BMEWO tagging scheme. Event extraction identifies triggers and links entities into structured representations. Events are categorized into three dimensions (time, space, emotion), with emotion serving as the key connective element. Relation types (e.g., "occurs in," "arrives at," "experiences," "feels") are defined to construct a multi-relational network. The final interactive EKG in Neo4j visualizes the unfolding of Xu Xiake's journeys across temporal, spatial, and emotional domains.

Introduction of an Innovative Multidimensional Analytical Framework: We propose a framework that integrates time, space, and emotion, positioning emotion as the central thread linking the other two dimensions. Time is conceptualized at three levels: natural time (seasons, weather) influencing travel conditions and mood; social time (dynastic years, eras) embedding the narrative within broader contexts; and personal time (life stages) reflecting biographical and inner development. Space is divided into natural space (mountains, rivers) and social space (temples, inns), the latter revealing social structures and interactions. Emotion functions as the "semantic glue," binding time and space; mapping emotional expressions onto specific temporal markers and locations enables the reconstruction of Xu Xiake's evolving psychological landscape.

Implementation of a Visual Spatiotemporal Narrative Analysis Method: Utilizing the constructed EKG, we implement a method that visualizes spatiotemporal contexts and dynamically analyzes emotional variations. This enables quantitative analyses of: Xu Xiake's emotional tone across life stages; travel patterns under different natural time conditions; narrative functions of specific locations; structure of his social network; influence of social time and space on behavior; and evolution of emotional states under changing natural and social conditions.

This study establishes, for the first time, a systematic protocol for constructing EKGs for classical travel literature. It breaks through the limitations of traditional isolated analyses by explicitly revealing the implicit multidimensional event networks within the text. The proposed EKG framework not only provides a methodological paradigm for developing generalized EKGs for travel literature but also offers a valuable reference for the structural analysis and knowledge discovery of other narrative classical texts.

Figure 1 excerpt from the EKG of The Travel Diaries of Xu Xiake

Figure 2 Schematic Diagram of the Spatiotemporal Narrative Framework

Figure 3 Demonstration of the Spatiotemporal Narrative Analysis Process (Selected Sections)

Liu, Weili / Zhang, Haitao / Zhang, Xinrui et al. (2024): “三元空间视域下重大突发事件的事件图谱研究 [Research on Event Graph of Major Emergencies from the Perspective of Ternary Space]”, in: 图书情报工作 [Library and Information Service] 68, 10: 135–146. DOI: 10.13266/j.issn.0252-3116.2024.10.012.

Rospocher, Marco / Van Erp, Marieke / Vossen, Piek et al. (2016): “Building event-centric knowledge graphs from news”, in: Journal of Web Semantics 37-38: 132–151. DOI: 10.1016/j.websem.2015.12.004.

Wei, Tong / Sang, Yuchen / Shi, Rui (2022): “基于知识重构的年谱时空可视化呈现——以《朱熹年谱长编》为例 [Spatio-temporal Visualization of Chronological Biography Based on Knowledge Reconstruction: A Case Study of Zhu Xi Nianpu Changbian]”, in: 中国图书馆学报 [Journal of Library Science in China] 48, 02: 62–75.

Yang, Li / Luo, Wenlin (2014): “论《给樱桃以性别》中的时空叙事话语与主题叙事 [On the Spatiotemporal Narrative Discourse and Thematic Narrative in Sexing the Cherry]”, in: 当代外国文学 [Contemporary Foreign Literature] 35, 03: 98–109.

Zheng, Jiwen (2021): 村上春树小说的时空体特色研究 [A Study on the Chronotope Features of Haruki Murakami's Novels]. Ph.D. thesis, 南京师范大学 [Nanjing Normal University] https://d.wanfangdata.com.cn/thesis/D02589846 [28.04.2026].