<![CDATA[Background: The large-scale digitization of Indonesian literary works has produced extensive textual corpora that challenge conventional close-reading approaches and call for systematic, data-driven methods capable of capturing thematic, semantic, and affective patterns in fiction. Objective: This study aims to examine how text mining and semantic modeling can reveal lexical salience, intertextual relations, and narrative emotion in Indonesian fiction across different thematic orientations. Method: Using a quantitative corpus-based design, the study analyzes 36 Indonesian literary texts published between 1980 and 2022 through TF–IDF–based lexical analysis, document-level semantic embeddings with cosine similarity and clustering, and sentence-level sentiment analysis. Results: The findings show distinct lexical signatures that differentiate thematic clusters, coherent semantic groupings reflecting intertextual proximity, and sentiment trajectories dominated by neutral-to-negative polarity with strategically placed affective peaks across narrative progression. Implication: These results demonstrate that computational methods can empirically support literary analysis without displacing interpretive criticism. Novelty: The study integrates lexical, semantic, and affective modeling within a unified framework for Indonesian fiction, offering a scalable and replicable approach to digital literary studies.]]>
Copyrights © 2026
