I. Introduction

Long-form web novels distributed through digital platforms have emerged as one of the central forms of contemporary popular narrative. In particular, Chinese xuanhuan (玄幻) and xianxia (仙俠) web fiction is characterized by massive textual scale, often extending to millions of Chinese characters, as well as long-term serialized publication structures. These narratives unfold through repetitive, expansive, and cumulative mechanisms. Rather than relying solely on character-centered storytelling, such works repeatedly organize narrative development around the acquisition and consumption of specific resources, the advancement of cultivation realms, and the continual reorganization of power structures.

However, previous literary studies have primarily focused on narrative content, character typology, and genre conventions, while remaining limited in quantitatively analyzing the temporal and structural transformations that occur within long-form web fiction. In particular, Battle Through the Heavens (鬥破蒼穹, serialized from April 2009 to July 2011) represents one of the most influential long-form xuanhuan novels in Chinese online literature, consisting of 1,623 chapters and approximately 5.87 million Chinese characters. The core mechanism of the novel lies not simply in interpersonal relationships, but in a progression system centered on the acquisition and fusion of scarce resources, especially the “Heavenly Flames” (異火, Yi Huo). The protagonist Xiao Yan’s advancement through cultivation realms does not occur through the natural flow of time, but through the stepwise acquisition and integration of specific resources. Throughout this process, character networks, factional structures, spatial mobility, and combat configurations are continuously reorganized.

Focusing on these characteristics, this study models the narrative of Battle Through the Heavens as an “evolving progression network” that transforms over time. To achieve this, the study proposes a double bipartite analysis framework that integrates two independent 2-mode networks: Character–Flame and Flame–Realm. In addition, rather than adopting chapter sequence as a temporal axis, this study reconstructs temporality around the protagonist’s cultivation-level progression in order to more effectively capture the internal rhythm of growth and the structural turning points of the narrative.

The objectives of this study are twofold. First, it seeks to visualize the temporal narrative structure of long-form web fiction through dynamic network methodology. Second, it aims to structurally demonstrate that the progression mechanism of Battle Through the Heavens operates as a resource-mediated progression narrative. Through this approach, the study expands the methodological connection between web fiction studies and digital humanities, while proposing a new analytical model for examining temporality in large-scale serialized narratives.

II. Theoretical Background and Previous Studies

Unlike traditional heroic narratives, Chinese xuanhuan and xianxia web novels are structured around quantified progression systems and stepwise upgrade mechanics. In these works, character growth is not represented primarily through psychological maturation or accumulated life experience, but through the acquisition of specific resources, abilities, equipment, and cultivation realms. In particular, the realm system operates similarly to the level systems of MMORPGs, enabling readers to perceive progression intuitively through numerical and hierarchical advancement.

Previous studies have described these characteristics through concepts such as “upgrade narratives” (升級流) or “gamified narrative structures.” Some studies have examined the relationship between repetitive growth structures and platform-based serialization systems, while others have focused on system fiction, data-driven ability structures, and interface-centered narrative forms. However, most prior studies have remained centered on textual interpretation, and relatively few have analyzed how progression structures dynamically evolve over time within the network structure of actual works.

Battle Through the Heavens represents one of the most systematic implementations of this progression mechanism. In the novel, resources such as the Heavenly Flames function not merely as setting elements, but as core mechanisms that drive realm advancement and the restructuring of power relations. Accordingly, this study conceptualizes the novel’s growth structure as a “resource-mediated progression narrative” and analyzes it within a dynamic network framework.

Within literary studies, network analysis has been widely employed to visualize character relationships and analyze centrality, clustering, and connectivity. Following Franco Moretti’s concept of distant reading, digital humanities scholarship has increasingly attempted quantitative analysis of large-scale literary corpora. In particular, co-occurrence-based character networks have become a representative methodology for identifying social structures within novels.

However, web fiction differs from conventional literary forms in that character relations, factional structures, spatial movement, and resource distributions continuously transform over extended serialization periods. Consequently, temporality becomes a central analytical dimension. Nevertheless, much existing research remains confined to static network analysis, reducing entire narratives to fixed relational structures and thereby failing to adequately reflect dynamic narrative reorganization over time.

Recent digital humanities research has increasingly explored dynamic networks, temporal graphs, and narrative evolution. Yet most studies treat narrative progression as simple chapter-based segmentation or focus only on changes in interpersonal relations. In contrast, in progression-oriented web novels such as Battle Through the Heavens, realm advancement itself functions as a structural turning point within the narrative. Existing temporal models are therefore insufficient for fully explaining the rhythm of progression in such works.

To address these limitations, this study reconstructs the temporal axis around the protagonist’s cultivation progression stages. This approach reflects the fact that the progression mechanism of web fiction is organized not simply through event sequences, but through repeated cycles of cultivation and breakthrough. Furthermore, by constructing a multilayered network that incorporates not only character relations but also resource nodes (Heavenly Flames) and progression nodes (Realms), this study proposes a more sophisticated model for narrative evolution in long-form web fiction.

III. Data Construction and Research Methodology

In order to structurally analyze the progression narrative of Battle Through the Heavens, this study constructed an integrated network model connecting Character–Flame–Realm relations. The analysis was conducted on the entire corpus of 1,623 chapters (approximately 5.87 million Chinese characters). During preprocessing, a manually curated character dictionary containing 302 characters was constructed, while Heavenly Flames, cultivation realms, and progression events were extracted and normalized.

Specifically, the study unified aliases and alternative appellations of identical characters, variant spellings of Heavenly Flame names (e.g., 八荒破滅炎 / 八荒破滅焱), and inconsistencies in realm notation. Cases of realm-related noise (e.g., 鬥氣 vs. 鬥氣大陸) were also separately processed to ensure data consistency. Chapter titles containing character names, flame names, or realm terminology were excluded to avoid analytical distortion. In addition, only characters whose in-text appearance frequency corresponded consistently with progression frequency through cross-validation with the character dictionary were included in the analysis.

From the preprocessed dataset, this study extracted 23 progression triples based exclusively on directly verifiable narrative events within the text. Each triple follows a causal structure in which “a character acquires, inherits, or fuses with a specific Heavenly Flame, resulting in advancement to a higher cultivation realm.” Based on these triples, the analytical network was refined into three categories of nodes: 8 Character nodes, 12 Flame nodes, and 13 Realm nodes, for a total of 33 nodes.

Methodologically, this study does not employ a conventional character co-occurrence network, but instead adopts a 2-mode network structure centered on the novel’s core progression mechanism: the relationship between Heavenly Flame acquisition and realm advancement. The first network is a Character × Flame network that maps which Heavenly Flames are acquired, inherited, or utilized by specific characters. The second network is a Flame × Realm network that visualizes how each Heavenly Flame is associated with specific stages of realm progression and breakthrough. These two 2-mode networks are then integrated through shared Flame nodes to reconstruct a unified progression graph linking Character–Flame–Realm relations.

Edges within the integrated graph are categorized into three causally directed types based on triple semantics. First, “acquires” edges represent non-competitive acquisition or inheritance of resources, including innate possession, master-disciple transmission, bloodline awakening, and inherited cultivation techniques. Second, “triggers” edges encode competitive acquisition events, such as battles, final confrontations, or resource struggles, in which absorption of an external Heavenly Flame directly causes realm advancement. Third, “progression” edges represent breakthroughs achieved solely through cultivation without resource mediation, including Xiao Yan’s early-stage progression (Dou Qi → Dou Zhe → Dou Shi → Da Dou Shi) and certain environmentally conditioned breakthroughs.

Edge types are assigned at the triple level. In resource-mediated triples (“acquires” and “triggers”), the two directed edges (Character → Flame and Flame → Realm) share the same semantic type. In progression triples, a single directed edge (Character → Realm) bypasses the Flame layer. Within the dataset, the 23 progression triples consist of 15 acquires triples (30 edges), 3 triggers triples (6 edges), and 5 progression triples (5 edges), yielding a total of 41 directed edges.

This semantically grounded classification visualizes the structural asymmetry of progression mechanisms at the graph level, distinguishing between non-competitive inheritance, competitive struggle, and unmediated cultivation. Notably, all three “triggers” triples and all five “progression” triples are concentrated exclusively on the protagonist Xiao Yan, quantitatively demonstrating that active struggle and self-driven cultivation are structurally centralized in a single character.

In addition, this study constructed two parallel network models in order to cross-validate resource centrality at two analytical levels. First, a 2-mode co-occurrence network (Top-5 Heavenly Flames × 302 characters) measures social proximity through chapter-level co-occurrence frequency between characters and Heavenly Flames. The analysis shows that 244 out of 302 characters (80.8%) co-occur at least once with one of the top five Heavenly Flames, while 164 characters surpassed the threshold of three cumulative co-occurrence chapters and were included in visualization.

Second, the CFR (Character–Flame–Realm) causal network explicitly models resource-mediated progression structures by extracting verifiable Character–Flame–Realm relations in triple form.

To quantify resource centrality, this study introduces type-specific influence metrics. Conventional formulas such as:

Influence(v) = degree(v) × log(frequency(v) + 1)

apply identical measurements to all node types and therefore fail to reflect role differentiation among heterogeneous nodes. Accordingly, this study defines distinct formulas according to the functional role of each node category:

Character_Influence(C) = ( |F(C)| + |R(C)| ) × log(freq(C) + 1)

*F(C): flames connected to C, R(C): realms connected to C

Flame_Influence(F) = |C(F)| × log(freq(F) + 1)

*C(F): characters connected to F

Realm_Influence(R) = |C(R)| × log(freq(R) + 1)

*C(R): characters connected to R

These formulas are based on the topological distinction that Character nodes function as multi-domain agents operating across both Flame and Realm layers, Flame nodes function as social hubs connecting characters, and Realm nodes function as convergence points of progression trajectories.

Finally, this study introduces ArcPhase as an auxiliary visual channel that compresses temporal flow into the static graph structure. ArcPhase divides the narrative into twelve qualitative temporal stages (P0–P11), representing the occurrence timing of each progression triple through color encoding. The phases are defined according to major narrative turning points within the novel, ranging from P0 (Innate/Origin) to P11 (Ending). A unidirectional color gradient from bright yellow (#F2C200) to deep crimson (#2A070C) encodes temporal progression, narrative intensity, resource hierarchy, and cultivation-stage escalation simultaneously.

Unlike conventional color classification methods, ArcPhase is applied not to nodes but to progression triples themselves. Thus, even the same Heavenly Flame may belong to different phases depending on which character acquires it and at what narrative moment. This event-centered encoding visually emphasizes that resource acquisition and realm advancement constitute two inseparable aspects of a single progression event.

IV. Results

Figure 1: Top-5 Heavenly Flame–Character Co-occurrence Network

The Top-5 Heavenly Flame–Character Co-occurrence Network quantitatively demonstrates the social centrality of Heavenly Flames within the narrative. Among the 302-character dictionary, 244 characters (80.8%) co-occur at least once with one of the top five Heavenly Flames, while 164 characters (54.3%) surpassed the threshold of three cumulative co-occurrence chapters and were included in the visualization. These findings indicate that approximately four-fifths of all characters exist within the textual gravitational field of the Heavenly Flames, suggesting that the flames function not merely as narrative props but as structural anchors organizing character placement.

The network also automatically visualizes the political and factional structures of the novel. Clusters formed around the five Heavenly Flames correspond closely to specific narrative factions, including late-stage Hun Clan and Ancient Clan groups around Jing Lian Yao Huo, antagonistic pursuit factions around Yun Luo Xin Yan, Yao Lao-related groups around Gu Ling Leng Huo, Jia Nan Academy and early-stage factions around Qing Lian Di Xin Huo, and final-battle factions around Xu Wu Tun Yan.

In particular, a persistent core cast of 21 characters—including Xiao Yan, Yao Lao, Zi Yan, Da Zhang Lao, Cai Lin, Xun Er, and Xiao Yi Xian—co-occurs with all five Heavenly Flames. Although these characters account for only approximately 7% of the entire cast, they function as witnesses to all major flame-related events, while the remaining 281 characters appear only within specific flames or narrative phases. Quantitatively, this demonstrates that the narrative structure is organized as a concentric system in which a stable core cast surrounds the Heavenly Flames while peripheral characters are continuously replaced.

Table 1: Quantitative Analysis of Network Density and Centrality Before and After Heavenly Flame Acquisition

The quantitative analysis of network density and centrality before and after major Heavenly Flame acquisitions further demonstrates abrupt structural transformations within the narrative. At each major acquisition point, network density and centrality indicators increase dramatically. In particular, the sharp increase in betweenness centrality during P7 demonstrates Xiao Yan’s role as a critical bridge connecting geographically fragmented factions. Likewise, the substantial rise in eigenvector centrality during P9 indicates that Xiao Yan evolves beyond simple relational expansion into a strategic hub connected to the highest-level power nodes of the fictional world.

Figure 2: Cumulative Influence Trajectories of Major Characters, Heavenly Flames, and Realms Across Narrative Progression

The cumulative influence graphs further reveal the progression-centered structure of the narrative. Xiao Yan’s cumulative influence overwhelmingly surpasses that of all other entities. Although the curve rises gradually during the early chapters, it exhibits stepwise surges whenever new Heavenly Flames are acquired or cultivation realms are surpassed. After the Central Plains phase, the curve rises almost vertically, ultimately reaching an exceptionally high influence score of 248. In contrast, all other characters remain below 20, quantitatively demonstrating the novel’s overwhelming concentration on the protagonist’s progression.

The Heavenly Flame influence graph shows that Qing Lian Di Xin Huo begins influencing the narrative around Chapter 200, establishing the foundation of Xiao Yan’s growth trajectory. Later-stage Heavenly Flames, especially Jing Lian Yao Huo and Xu Wu Tun Yan, display explosive increases in influence immediately upon introduction. This indicates that acquisition of increasingly powerful Heavenly Flames dramatically amplifies not only Xiao Yan’s combat ability but also his structural influence over the entire narrative world.

The Realm graph visualizes both cultivation hierarchy and narrative power inflation. Beginning from the low-level Dou Qi stage, the narrative gradually shifts toward higher-level realms such as Dou Huang and Dou Zun during the middle stages. The most striking transformation occurs during the Ancient Realm Final Battle phase, when the supreme cultivation realm Dou Di appears and rapidly rises to an influence value of 22.2. This quantitatively demonstrates the decisive narrative importance of the legendary cultivation stage pursued by all cultivators throughout the story.

Taken together, these findings demonstrate that Xiao Yan’s path toward becoming Dou Di through the acquisition of increasingly powerful Heavenly Flames is not a linear developmental process, but rather a canonical example of explosive stepwise progression driven by major narrative events. Combined with the Character–Flame–Realm causal network, these results provide integrated evidence that Heavenly Flames function as the central organizing mechanism of the narrative at both the levels of social proximity and causal progression.

Figure 3: Character–Flame–Realm (CFR) Causal Network

The CFR (Character–Flame–Realm) causal network enables modeling not merely hierarchical structures but the dynamic progression structure of the narrative itself. Characters such as Hun Tian Di and Tuo She Ancient Emperor, who had already reached the highest cultivation levels before the beginning of the narrative and therefore do not undergo progression events during the main storyline, were excluded from the analysis. This exclusion represents not merely data cleaning but a methodological decision consistent with the study’s focus on dynamic progression mechanisms rather than static distributions of power.

Among the eight validated core characters, a strong positive correlation emerges between textual appearance frequency and progression-event frequency. Xiao Yan alone accounts for 10 out of 21 total progression events (47.6%), while Yao Lao, Xun Er, Zi Yan, and Hai Bodong each account for two events. This quantitatively demonstrates that the narrative operates as a “single super-hub” progression structure centered overwhelmingly on the protagonist.

Analysis of the twelve Flame-layer nodes reveals a discrepancy between the official Heavenly Flame rankings within the fictional world and their actual narrative mediation frequency. For example, Xu Wu Tun Yan, officially ranked second among the Heavenly Flames, mediates only one progression event despite being mentioned 294 times in the text. Nevertheless, it possesses degree centrality of 4 because it simultaneously mediates the final-stage advancement of both Xiao Yan and Hun Tian Di to L12. Similarly, Yun Luo Xin Yan, despite having the highest textual frequency (655 mentions), mediates only one progression event and maintains degree centrality of 2.

By contrast, Gu Ling Leng Huo, officially ranked eleventh, exhibits degree centrality of 4 by mediating progression events for both Xiao Yan and Yao Lao, thereby forming a dual-owner structure. Jing Lian Yao Huo, officially ranked third, likewise functions as a dual-owner mediation node with degree centrality of 4. Meanwhile, Qing Lian Di Xin Huo, officially ranked nineteenth, mediates only Xiao Yan’s advancement to L6, but serves as the narrative ignition point because it constitutes the protagonist’s first Heavenly Flame absorption event.

This asymmetry suggests that the narrative design prioritizes stage-specific instrumental logic over absolute resource ranking. Furthermore, analysis of the Realm axis shows that L11 (Dou Sheng) functions as the largest convergence node within the graph, receiving five incoming edges from four different characters mediated through multiple Heavenly Flames. In contrast, the early and middle cultivation stages (L1–L9) are overwhelmingly dominated by Xiao Yan’s individual progression trajectory.

From the perspective of edge-type distribution, all six progression events occurring at L1–L5 and L10 are classified as “progression” edges achieved solely through cultivation without Heavenly Flame mediation, and all belong exclusively to Xiao Yan. Conversely, all five incoming edges converging on L11 are mediated through Heavenly Flames, visually demonstrating that late-stage realm advancement becomes entirely dependent on external resource acquisition and absorption mechanisms.

V. Conclusion

This study constructed a Character–Flame–Realm (CFR) causal network for Battle Through the Heavens and analyzed the progression narrative of long-form web fiction through a digital humanities methodology. Unlike previous studies centered on character co-occurrence, this research adopted the progression event itself—namely, “resource acquisition and realm advancement”—as the primary analytical unit, thereby structurally visualizing the progression mechanism of the novel.

The analysis demonstrates that the growth structure of Battle Through the Heavens is organized around a “single super-hub” centered on Xiao Yan, with progression events heavily concentrated on the protagonist. The narrative’s development is structured not primarily around interpersonal relationships, but around scarce resources such as the Heavenly Flames. Core resources including Jing Lian Yao Huo, Gu Ling Leng Huo, and Xu Wu Tun Yan function not merely as setting devices, but as structural hubs mediating realm advancement and narrative transformation.

Furthermore, the discrepancy between official Heavenly Flame rankings and actual narrative influence indicates that power structures within the novel are governed less by nominal hierarchy than by narrative placement logic. Most importantly, this study demonstrates a form of “triple isomorphism” in which protagonist, Heavenly Flame, and cultivation realm mutually reinforce one another to produce the overall progression structure. The narrative of Battle Through the Heavens is therefore not simply a story of individual power escalation, but a progression system driven by the repeated integration of Heavenly Flame acquisition and realm advancement.

In methodological terms, this study proposes a new digital humanities framework for analyzing the temporal progression structures of long-form web fiction by reconstructing temporality around cultivation-stage advancement and integrating two 2-mode networks—Character–Flame and Flame–Realm—through a double bipartite analysis framework. This approach may be extended not only to Chinese xuanhuan and xianxia web fiction, but also to broader forms of digitally mediated narratives structured around gamified progression systems.