Ontology is a conceptual, semantic-level framework or a concrete artifact provided for a specific purpose (Guarino and Pierdaniele, 1995). It defines classes, properties, and relations among the members of classes that are relevant to model the knowledge of a domain (Gruber, 2009). This enables ontology to represent complex relationships between knowledge entities, integrate heterogeneous data resources, infer implicit knowledge, support knowledge integration, and enhance interoperability. Several studies have applied ontology to the knowledge organization of biography, mainly focusing on the construction principles and practical implementation of biography ontology.

Three methodologies guide biographical ontology construction:

- Character-centric. The ontologies encapsulate educational backgrounds, professional competencies, and interpersonal relationships. Adhi and Widodo (2019) developed an ontology focused on characters, detailing their relatives, partners, and skills. Motta et al (2009) developed an ontology for musicians, containing person, works, and time-related entities. Leskinen and Hyvönen (2020) introduced a model comprising seven entities: person, relationship role, family relationship role, label, occupation, event, and place.

- Event-driven. The ontologies utilize temporal and spatial dimensions to forge connections. Yeh’s (2018) Biographic Knowledge-based Story Ontology (BKOnto) defined four entities and fifteen properties centered on events. Zou and Park (2018) developed the Event-based Ontology for Historical Description (EOHD), which encompassed six entities, such as spatiotemporal and object change events. Krieger and Declerck’s (2015) Ontology for Biographical Knowledge (BIO) represented actions within the happening, abstract, and object classes.

- Integrative approach. Araújo et al (2018) designed an ontology for the Museum of the Person (OntoMP) that correlated people and events, detailing ancestry, progeny, and professional endeavors through 30 concepts and 11 properties, including education, migration, and accidents.

The practical applications of biography ontology include: (a) semantic annotation and association. The BiographyNet Schema (Ockeloen et al, 2013) extracted characters, historical events, places, and eras from 125,000 biographies to generate a semantic knowledge base. The Bio CRM Ontology modeled characters’ roles, time spans, social relationships, and historical events in the University of Oxford’s Early Modern Letters Online (EMLO) project; (b) data analysis and visualization. The Bio CRM ontology served as a descriptive framework for 13,100 Finnish biographies on the BiographySampo website (https://seco.cs.aalto.fi/projects/biografiasampo/en/), facilitating life map visualization, field retrieval, knowledge inference, and demographic data computation. Additionally, it enriched and analyzed 13,000 biographies in the Finnish National Biography Semantic Network and 19,137 biographies of Nossi alums through semantic enrichment, quantitative analysis, and visualization and; (c) knowledge mining. The Ontology for Biographical Knowledge (BIO) (Krieger and Declerck, 2015) supported cross-linguistic data mining and integration in the EU TrendMiner project by organizing and semantically linking individuals’ biographies and life experiences.

Digital storytelling, which originated at the Centre for Digital Storytelling in the 1990s (Lambert, 2013), refers to the methods and practices of using digital technologies and media for storytelling (Nosrati and Detlor, 2022). Advances in digital technologies and media integration have broadened the application of digital storytelling to cultural heritage (Valtolina, 2016), education (Wu and Chen, 2020), museums (Perouli, 2021), and social memory (Tie et al, 2023, pp. 73). Digital storytelling involves cross-media, multimedia, and interactive storytelling, characterized by narrativity, mediality, and interactivity (Miller, 2019). As a digital humanities method, digital storytelling promotes knowledge association, reconstruction, discovery, presentation, and dissemination through semantic web, computational, and multimedia technologies (Fu et al, 2023, pp. 1525). Digital storytelling enhances user experience and understanding, as well as value creation.

Current research in digital storytelling concentrates on its elements, frameworks, and assessment. Two dominant theories outline digital storytelling elements: the three-element theory—data, storytelling, and visualization (https://powerbi.microsoft.com/en-us/data-storytelling/?cdn=disable)—and the five-element theory—central idea, exploratory points, linear structure, storytelling techniques, and visualization (Dykes, 2015). Frameworks refer to methodologies for producing and presenting digital stories, including: ① Product design method, which mirrors the product design process and segments digital storytelling into ideation, investigation, reporting, and refinement (Planer and Godulla, 2021); ② Data-centered method, emphasizing data analysis insights and categorizing the storytelling process into digitization, datafication, and aestheticization—converting physical entities to digital, extracting data value, and disseminating through media and technology; ③ Storytelling-centered method, which involves selecting storytelling elements, choosing scenes, designing plots, creating stories, and artistically presenting them (Meier, 2022). In addition, several methods are used to assess the quality of digital stories and define them as successful or unsuccessful. These methods focus on plot design, the pacing of storytelling, audio soundtrack and emotional content of the digital stories. Factors such as emotional content of the digital stories, and interactivity, quality of images and visual impact were also given attention (Malkawi et al, 2019).

In summary, existing research primarily conceptualizes biographical knowledge with the core concepts of characters and events; however, the definitions of entities and properties are simple. Moreover, there has been insufficient attention to other types of historical materials related to the biographical subject, such as photographs, manuscripts, and certificates. The current biography ontologies are primarily applied to semantic association, data analysis, and knowledge mining, yet their role in advancing biographical narrative and presentation remains limited. Accordingly, this study employs the Autobiography as a case study, articulating its requirements for knowledge description and extracting biographical knowledge elements in order to construct an ontology that enhances the granularity of knowledge representation and interrelates diverse material types. Building on this foundation, the study integrates thematic clustering and visualization techniques to explore the methodological pathway for biography digital storytelling, involving the thematic clustering of biographical content, the linkage of storytelling threads, and the visualization of storylines. The goal is to foster innovation in the narrative structures of biographies, enhance the user experience, and enrich users’ comprehension of the evolution of library science in China.

The Autobiography was selected as the primary storytelling material for this study. Autobiography, published by the National Library of China Publishing House in 2020, was derived from Mr. Peng Feizhang’s oral accounts, as transcribed and edited by Peng et al (2020). Drawing on in-depth interviews with Mr. Peng, the collators collected and digitized a range of physical materials, such as photographs, letters, certificates, and manuscripts. After verifying the historical accuracy of the materials, they organized them chronologically, resulting in a comprehensive manuscript of approximately three hundred thousand words, including ninety images.

Peng Feizhang is widely recognized as a distinguished Chinese library scientist, educator, and bibliographer with significant contributions spanning several decades. Beginning his career as a junior academic during the early years of the People’s Republic of China, Peng later studied in the Soviet Union. Upon returning to China, he became the inaugural professor of library science, and subsequently held multiple pivotal roles, including serving as the first dean of China’s first School of Library and Information Science, the country’s first doctoral supervisor in information science, and the convener of the disciplinary review group of the State Council Academic Degrees Committee. Mr. Peng’s contributions have profoundly influenced the academic landscape and institutional development in library science.The digital storytelling of his Autobiography not only captures the development of Chinese librarianship but also serves as a valuable means of studying the history of library science education and promoting the ideals embodied by exemplary figures. With the publisher’s permission, the complete verbatim manuscript of the Autobiography was obtained and utilized in this study.

This study characterizes the storytelling materials along three dimensions: content, type, and utilization.

The construction of the biography ontology is informed by mature methodologies, notably the seven-step approach (Noy and McGuinnes, 2004) and the IDEF5 method (Integrated Definition for Ontology Description Capture) (Koji et al, 2004). The construction of biography ontology encompasses five key steps:

5.1.4 Designing Classes and Properties

• Definition of core and extended classes. Drawing on the analysis in Sections 5.1.2 and 5.1.3, the core classes of the ontology are identified as foaf:Agent, shl:OfficialEvent, shl:Temporal, shl:Place, and Resource. Additionally, 19 extended classes are established, referencing Friend of a Friend (FOAF), shlgen, Time, The Bibliographic Ontology (BIBO), and The Common European Research Information Format (CERIF), to further extend and complement the core classes.

• Design of data properties. Data properties are used to characterize data, providing detailed descriptions of the core and extended classes. The data properties are defined by referencing established ontologies pertaining to people, organizations, time, places, and bibliographic data. For entities lacking appropriate existing ontologies or vocabularies, a custom approach is undertaken to address specific application requirements. Ultimately, 53 data properties are defined to comprehensively describe the ontology’s classes, as shown in Fig. 2.

Fig. 2.

Classes and data properties of biography ontology.

• Definition of object properties. Object properties capture the relationships between various classes, reflecting the semantic connections present among biographical knowledge entities. Drawing on established ontologies such as FOAF, CBDB, BIBO, shlgen, shlmarc, ORG, REL, and CIDOC-CRM, we incorporate these ontologies alongside custom extensions designed to meet the unique requirements of biographical knowledge. Ultimately, 46 object properties are defined, as shown in Table 1.

Table 1. Object properties of the biography ontology.

Type	Object property/Relation	Definition
Spatio- temporalrelation	time:hasTime, time:hasTemporalDuration	Temporal characteristics of agents, resources, and events.
	shl:nativePlace, shl:burialPlace, shl:region, crm:tookPlaceAt, createdAt/publishedAt, hasBeenTo	Geographical characteristics of agents, resources, and events.
Person- Organisation relation	rel:parentOf, rel:siblingOf, rel:childOf, rel:friendOf, rel:employedBy/rel:employerOf, rel:mentorOf, rel:classmateOf, rel:grandchildOf, rel:worksWith, rel:mentorOf, rel:spouseOf, rel:acquaintanceOf	Relations among relatives, colleagues, and partners.
	org:hasMember, org:headOf, org:subOrganizationOf	Organizational structures: members, leaders, sub-institutions.
Action relation	studyAt	Educational affiliations between individuals and organizations.
	workAt	Employment affiliations between individuals and organizations.
	rel:collaboratesWith	Cooperative interactions among agents.
	org:originalOrganization, org:resultingOrgani- zation	Organizational transformations: mergers, splits, renamings, relocations.
	crm:wasInfluencedBy	Interactions among events, agents, and resources.
	shl:creatorOf	Agents’ activities: writing, photographing, designing websites and audiovisual content.
	bibo:editor	Agents publishing and disseminating scholarly and digital media content.
	publisherOf	Agents publishing and disseminate monographs, essays, websites, audiovisual documents.
	bibo:translator	Agents translating scholarly texts.
	bibo:director	Agents organizing and directing audiovisual resources.
	sign	Contracts executed among agents.
	leaderOf	Collaborations between agents and academic projects.
	crm:hadParticipant	Interactions between agents and events.
	isDescribedBy	Descriptions linking events and resources.
	crm:occurredInThePresenceOf	Materials featured in events.
	concurWith	Concurrent events.
	followedBy	Sequential events: before and after.
	causedBy	Causal relationships among events.
	postedIn	Associations between academic papers and journals.
	outcomeOf	Collaborations on academic papers and monographs.

This phase entails extracting storytelling elements—agents, time, locations, events, and resources—utilizing NLP technology, grounded in the core, extended classes, and data properties defined within the biography ontology.

Semantic association involves a detailed exploration of biographical knowledge and a comprehensive understanding of biographical content. It aims to sequentially organize the text by linking storytelling elements to form story units, serving as the basis for constructing storytelling plots and threads. Ontology offers a detailed and structured framework for knowledge modeling and representation. This section details the associations among storytelling elements through the object properties defined in the biography ontology. The storytelling elements and their interrelationships are illustrated in Fig. 3.

Fig. 3.

The classes and their relationships of the biography ontology.

LDA is a generative probabilistic model of a corpus. The basic idea is that documents are represented as random mixtures over latent topics, where each topic is characterized by a distribution over words (Blei et al, 2003). By giving the topic probability distribution of each document in the corpus, LDA enables text mining, topic discovery and clustering. In this study, the data is stored as a corpus, which is preprocessed as described in Section 5.1.2 after data cleansing and segmentation. Each natural paragraph is treated as a separate document, resulting in a total of 1169 documents. The terms in the documents are converted into a word frequency matrix, and LDA is applied to model the topic distribution. The analysis yielded a different number of topic clustering results and their Coherence Score, as shown in Fig. 4.

Fig. 4.

The Coherence Score corresponding to different number of topics.

The Coherence Score is a key metric for determining the number of topics. The underlying principle is that a high frequency of keyword co-occurrence within each topic corresponds to a high Coherence Score, indicating strong intra-topic correlation, high text interpretability, and effective clustering. The Coherence Score for various clustering results is computed using the U_mass method, with the findings presented in Fig. 4. The analysis reveals that the highest Coherence Score occurs when the number of topics is set to 5, indicating the optimal clustering configuration. Consequently, the Autobiography is segmented into five topic modules.

As an indispensable component of narrative, the theme represents the central idea of the storytelling text. It can be presented through storytelling threads, linking storytelling units and forming a coherent storyline. Following the thematic clustering analysis, the Autobiography is divided into five thematic modules, which form the storytelling threads. The associated story units are then systematically organized and linked across distinct categories. This process encompasses the following two primary steps.

First, the intertopic distance map (Fig. 5, left) and the most relevant terms for topics (Fig. 5, right) are presented based on the clustering results. In the intertopic distance map, higher inter-theme similarity values indicate that the circles representing the topics are closer together, signifying greater similarity in the content of those topics. The underlying principle is that each topic is represented as a vector, and the cosine of the angle between two topic vectors is computed to derive the inter-topic similarity. A higher similarity value reflects more substantial thematic similarity. Fig. 5 shows that the similarities between topics 1, 2, and 3 are relatively high and displays the most relevant terms for each topic, aiding in the determination of topic content. Due to space limitations, only the relevant terms for topic 1 are presented.

Fig. 5.

Topic clustering results for the Autobiography.

Second, determine the content of each topic module, and organize the storytelling threads. In conjunction with the content of the Autobiography, five thematic modules are identified: childhood and adolescence, working, teaching, studying abroad, scientific research, and family. Based on this categorization, five storytelling threads are developed to link story units: ① Working experience, including employment at Wuhan University, promotion to professor; ② Studying abroad, including the events such as arrival at the Union of Soviet Socialist Republics (USSR), studying under Prof. Alexander Davidovich Eichengolts, and completing the thesis defense; ③ Teaching and academic research, including starting enrolment of Masters students, proposing the research object of bibliography, publishing the first academic paper, and completing project of the National Social Science Fund of China; ④ Childhood and adolescence, covering events such as birth, attending primary school, high school and Boone Library School; ⑤ Marriage and family, encompassing marriage, birth of a child, and celebration of a diamond wedding with wife. Story units linked by storytelling threads can be accessed by theme, enhancing the logical flow and plot of biography digital storytelling.

The storyline is designed to integrate complex data in a visually intuitive manner (Chao and Zhang, 2019, pp. 70) and to organize storytelling elements chronologically. For the event elements extracted in Section 4, significant events are initially screened, then sorted and classified based on chronological order and thematic clustering to construct a timeline. The associated storytelling elements are then interconnected to form a cohesive storyline. This approach provides a comprehensive and intuitive representation of the biographical subject’s life experiences (see the left side of Fig. 6). Finally, the storytelling texts are composed in the first person for each node in the storyline, enabling users to access detailed information about the characters, organizations, time, locations, and the causes and consequences of the events by clicking on the nodes. This facilitates users to foster a deep and intuitive understanding and empathy towards Mr. Peng’s experience. Due to space constraints, this study uses the node “Wuhan University got doctoral degree programs in information science” (illustrated by the yellow arrow on the left of Fig. 6) as an example, as shown on the right of Fig. 6.

Fig. 6.

Biography storyline from temporal dimension.

Each node within the storyline is developed using storytelling elements—such as persons, organizations, time, location, and events—thereby forming the storyline based on the associations between these elements. For brevity, we use the node “Wuhan University got doctoral programs in information science” as an example to display storytelling, as shown in Fig. 7. The node revolves around Mr. Peng, with the blue box representing the core event, white boxes symbolizing related events, numbers within the boxes signifying the stages of events, and lines denoting causal, sequential, and co-occurring relationships.

Fig. 7.

A node of Storyline.

The storyline illustrated in Fig. 7 contains the following key events: In February 1985 (Phase I), Mr. Peng was appointed as a member of the Second Discipline Review Group of the Academic Degrees Committee of the State Council, and was assigned to the Chinese Language and Literature Group due to the lack of a separate review group for the discipline of library science, and thus he resolved to strive for a separate disciplinary review group and doctoral degree programs; In March 1986 (Phase II), he wrote The Report on the Application for Doctoral Degrees in Library Science, which was forwarded to Wu Heng at the Academic Degrees Committee via Yan Yimin, and then to the Minister of Education, He Dongchang, who provided a response. In June 1990 (Phase III), the Academic Degrees Committee of the State Council formed a distinct academic review group for library and information science (Phase III(1)) and held its fourth meeting in Beijing (Phase III(2)). During this meeting, Mr. Peng, along with six other experts—Ni Bo, Du Ke, Meng Guangjun, Guan Yixian, Yan Lizhong, and Wang Tingjiong—discussed key issues regarding master’s and doctoral programs in library and information science. In November 1990 (Phase IV(1)), with the approval of the Academic Degrees Committee of the State Council, Wuhan University was granted the right to offer doctoral programs in information science (Phase Ⅳ(1)), and Peking University was granted the right to offer doctoral programs in library science (Phase IV(2)). In 1991, Wuhan University commenced the enrollment of doctoral students in Information Science (Phase V). Fig. 7 links events chronologically and associates relevant storytelling elements through the classes and properties defined within the biography ontology.

Linking story units and associating storytelling elements with threads facilitate the extension of these units towards the core of the Autobiography. This approach supports the deep mining of storytelling materials, the identification of relevant historical resources, and the formation of a storytelling perspective on the development of library science education. To construct the storyline from multiple thematic dimensions, we divide the Autobiography into five thematic modules: adolescence and childhood, studying abroad, working experience, teaching and academic research, and marriage and family. This segmentation was derived from storytelling threads generated through thematic clustering in Section 5, with the modules arranged based on the similarity between thematic clusters (as depicted on the left side of Fig. 8).

Fig. 8.

Biography storyline from thematic dimension.

On the left side of Fig. 8, users can select a module to view the corresponding storyline. To illustrate, upon selecting “working experience”, the upper right section of the interface displays a first-person storytelling text about Mr. Peng’s career, incorporating relevant characters, institutions, and temporal and spatial elements. Deeper insights into a specific event can be further examined in the lower right corner of the interface, where the specific course of action is outlined, while images, manuscripts, and other relevant resources can be accessed. Due to space constraints, the author uses the example of “Wuhan University got doctoral degree programs in Information Science” (as shown on the right side of Fig. 8). Storylines from the thematic dimension are integrated into the storytelling threads, representing biographies across a variety of subjects with granularity and precision, allowing users to quickly identify and gain profound insights into the biographical subject’s life experiences.