Semantic Communities and Boundary-Spanning Lyrics in K-pop: A Graph-Based Unsupervised Analysis

Oktay Karakuş; Can Eyüpoğlu

cs.SIFeb 13, 2026

Semantic Communities and Boundary-Spanning Lyrics in K-pop: A Graph-Based Unsupervised Analysis

Authors: Oktay Karakuş, Can Eyüpoğlu

Organizations: Cardiff University, School of Computer Science and Informatics Cardiff, UK

Abstract

We examine how lyric representation shapes computational similarity using four 384-dimensional multilingual MiniLM representations of 7,891 K-pop songs by 1,456 artists from charts spanning 2000 to October 2023. Whole-song truncated embeddings provide the strongest standardized separation between neighbouring and random song pairs (1.995), compared with 1.931 for chunked documents, although chunked representations show greater adjusted Rand agreement and line-based representations yield higher modularity. Using whole-song embeddings, we construct a weighted cosine nearest-neighbour graph ( $k=30$ ). A representative Louvain partition identifies 21 operational communities, without a corresponding robust plateau in the Markov Stability analysis. AI-assisted examination of selected lyrics and terms suggests overlapping patterns of memory, separation, longing, self-reflection, attraction, and performance. Community membership is associated with character-script composition (Cramér's $V=0.294$ ; permutation $p=0.001$ ). High-participation songs exhibit broader cross-community connectivity, while participation has weak negative associations with embedding-relative novelty. These findings show how representation choices and textual characteristics influence the network organization of a cultural corpus, subject to limitations from truncation, graph construction, and the absence of independent human semantic validation.

Figures & tables

Figure 1: Representation comparison under the matched

k=30

graph construction for four strategies: line mean, unique-line mean, chunked document, and whole-song truncated. The figure shows the standardized neighbour-versus-random cosine separation with its conditional 95% anchor-bootstrap interval, together with mean pairwise NMI and ARI between Louvain runs and standard weighted modularity. Whole-song truncated is the selected representation; this selection does not imply superiority on every metric. NMI and ARI are run-agreement means without uncertainty intervals, and the separation intervals are not formal pairwise significance tests.

Representation	Separation gap [95% CI]	NMI	ARI	Modularity (SD)
Line mean	0.690 [0.668, 0.711]	0.5779	0.4326	0.46498 (0.00669)
Unique-line mean	0.592 [0.570, 0.614]	0.6252	0.4906	0.46872 (0.00440)
Chunked document	1.931 [1.914, 1.948]	0.6195	0.5310	0.44927 (0.00506)
Whole-song truncated (selected)	1.995 [1.978, 2.011]	0.6226	0.5027	0.41523 (0.00338)

Table 1: Comparison of the four lyric-embedding representations under matched settings (Supplementary Table S1). Separation gap is the pooled-SD standardized difference between mean neighbour cosine similarity and mean random-pair cosine similarity, with a 95% anchor-song bootstrap interval (1,000 resamples) that is conditional on the song embeddings and sampled neighbours. Modularity is the mean (SD) across Louvain runs; NMI and ARI are means over 190 pairwise comparisons between runs and are not uncertainty intervals. Whole-song truncated, the selected representation, has the largest separation gap but not the highest NMI, ARI or modularity.

Figure 2: Semantic network atlas of the 7,891 songs (nodes) in the weighted cosine $k$ -nearest-neighbour graph ( $k=30$ ; 209,805 edges). Node and within-community edge colours identify the 21 operational Louvain communities of the seed-19 partition (C0–C20); grey edges connect different communities. Positions come from a force-directed layout and are diagrammatic, not quantitative semantic distances; the layout does not independently validate community separation.

Figure 3: Script composition of the 21 operational communities (C0–C20). Each row shows the within-community proportion of songs in four character-script categories (Hangul-dominant, mixed Hangul–Latin, Latin-dominant, other); circle area is proportional to that share on one common scale, and hollow circles indicate zero songs. The left column gives community song counts. Script categories are character-composition proxies, not validated language identification. The association between script composition and community membership is descriptive (

\chi^{2}(60)=2048.54

; Cramér’s

V=0.294

, bias-corrected

V=0.290

; permutation

p=0.001

) and does not imply causation.

Figure 4: Exploratory lexical signatures of the 21 operational communities (C0–C20), with community song counts. Each panel shows eight terms selected from the positive TF-IDF differences for that community. Word bounding-box area encodes the difference on one common global visual scale across all panels. Word positions carry no semantic meaning. These signatures are exploratory descriptions of the same embedding graph and partition, not validated semantic themes.

Group	$n$	Category	Descriptive label	Content evidence	Script profile	Confidence
C0	38	Unclear	Unavailable love and repeated versions	Longing for unavailable love; several versions recur.	100% Hangul; 0% mixed	Low
C2	243	Theme-led	Seasonal imagery and memory	Weather/season terms with separation and memory.	72% Hangul; 26% mixed	Medium
C6	350	Mixed	Reciprocal address and relationship change	Devotion and changing relationships with reciprocal address.	94% Hangul; 5% mixed	Medium
C8	372	Theme-led	Breakup grief and place-linked memory	Separation, tears, and places tied to past relationships.	84% Hangul; 15% mixed	Medium
C15	529	Form-led	Dance, performance, and hooks	Movement, audience address, and repeated performance hooks.	8% Hangul; 83% mixed	High
C16	555	Unclear	Everyday relationship talk	Common relationship vocabulary; examples vary.	62% Hangul; 37% mixed	Low

Table 2: Selected evidence-linked descriptions of six operational communities. Category labels abbreviate theme-driven, mixed content and language/form, primarily language/form, and heterogeneous/unclear; they summarize bounded AI-assisted review, not validated semantic labels or estimates of community purity. Script strata are character-composition proxies.

Metric	Boundary mean	Other mean	Mean difference	95% CI (difference)	Cliff’s $\delta$
External-strength fraction	0.7621	0.4832	0.2790	[0.2738, 0.2838]	0.955
Community entropy	0.7085	0.4545	0.2540	[0.2489, 0.2586]	0.979
External communities reached	11.0418	6.5801	4.4617	[4.2860, 4.6511]	0.788
Centroid novelty	0.1880	0.2002	$-0.0123$	[ $-0.0199$ , $-0.0047$ ]	$-0.098$
Local novelty	0.2258	0.2440	$-0.0182$	[ $-0.0232$ , $-0.0122$ ]	$-0.156$

Table 3: Boundary-group comparison at the q90 participation threshold (Supplementary Table S9). Boundary songs (

n=790

) have participation

\geq 0.832139

; all other songs (

n=7{,}101

) form the comparison group. Mean difference is boundary minus other. Intervals are song-level bootstrap 95% intervals for the mean difference, not for Cliff’s

\delta

. The comparison is descriptive and conditional on the similarity graph and partition; the three connectivity measures are related summaries of the same graph connections and are not independent validation.

Figure 5: Boundary-spanning architecture. (a) Two authentic high-participation examples: Purple Rain (C0;

P=0.906

) and 못난 내 사랑 (C17;

P=0.898

). Connections are coloured by the neighbouring community, edge widths are uniform, and radial positions are diagrammatic. These are two high-participation examples, not a high-versus-low comparison. (b) Cliff’s

\delta

for the boundary upper-decile group (

n=790

) versus all other songs (

n=7{,}101

) for external-strength fraction (

+0.955

), centroid novelty (

-0.098

) and local novelty (

-0.156

). Panel (b) describes group-level comparisons and does not show novelty measurements for the two songs in panel (a). No confidence intervals are shown for Cliff’s

\delta

Computational Literary Studies

Computational Literary Studies uses data analysis and language technologies to examine literary style, themes, narratives, influence, and change.

See all 49 papers

A Computational Framework for Modelling Organisation-Level Semantic Identity from Longitudinal Textual Data
Aug 27, 2026Brinda Murali Krishna, Oktay Karakuş, Can EyüpoğluComputational Social Science
School of Computer Science and Informatics, Cardiff University, UK · Department of Computer Engineering, Turkish Air Force Academy, National Defence University, Türkiye
Does Persona Make LLMs K-pop Fans? A Pilot Study of LLM-Based Online Concert Audience Agents
Jun 5, 2026Kirak Kim, Hyojin Kim, Yejin Son +2Multi-Agent LLM Systems LLM Agent Evaluation
Graduate School of Culture Technology, KAIST, Daejeon, South Korea · Department of Artificial Intelligence, Yonsei University, Seoul, South Korea
Voices of Civilizations: A Multilingual QA Benchmark for Global Music Understanding
Feb 28, 2026Shangda Wu, Ziya Zhou, Yongyi Zang +4Music Understanding Audio-Language Model Evaluation
Independent Researcher · The Hong Kong University of Science and Technology, Hong Kong SAR, China · The Chinese University of Hong Kong, Hong Kong SAR, China

Aug 27, 2026J/K move · Enter open · S save

A Computational Framework for Modelling Organisation-Level Semantic Identity from Longitudinal Textual Data

Brinda Murali Krishna, Oktay Karakuş, Can Eyüpoğlu

School of Computer Science and Informatics, Cardiff University, UK · Department of Computer Engineering, Turkish Air Force Academy, National Defence University, Türkiye

Computational Social Science

Open paper AskPDF arXiv

Can longitudinal texts produced by organisationally affiliated contributors reveal collective semantic differentiation and change without obscuring differences among those contributors? We develop a computational social science framework that combines established semantic and graph measures into interpretable organisational fingerprints and temporal profiles. We use organisation-level semantic identity to describe a corpus-dependent interpretation of recurring and changing features of affiliated textual output, rather than corporate culture or managerial intention. Drawing on an inherited semantic landscape, we examine 1,289 songs by 126 artists affiliated with four K-pop entertainment companies between 1997 and 2023. The evidence for differentiation is selective. After correction, all three novelty-related measures show artist-level differences, whereas the diversity and connectivity measures tested do not; only two company--community enrichments remain significant. Temporal trajectories reveal further distinctions between portfolios. Artist-grouped balanced accuracy reaches 0.334 against a chance baseline of 0.25, providing modest supporting evidence of affiliation-related signal rather than independent validation of identity. Robustness analyses support the presence of continuous semantic structure more strongly than precise rankings. Together, these findings show how collective regularities can coexist with shared content, contributor heterogeneity and change over time, while leaving cross-domain generality and broader organisational construct validity unresolved.

Abstract

Computational Literary Studies

Explore similar work

A Computational Framework for Modelling Organisation-Level Semantic Identity from Longitudinal Textual Data

Does Persona Make LLMs K-pop Fans? A Pilot Study of LLM-Based Online Concert Audience Agents

Voices of Civilizations: A Multilingual QA Benchmark for Global Music Understanding