Abstract—Lexicon-based sentiment analysis often faces challenges due to the limited coverage and adaptability of existing sentiment lexicons, particularly in domain-specific or resource-scarce contexts. Constructing new lexicons from scratch is labor-intensive and rarely feasible for large-scale applications. To address this gap, we introduce the Enhanced Sentiment Analysis using Extended Corpus and Co-Occurrence Graph (ECG) framework, which leverages co-occurrence graph construction, triadic closure theory, and PageRank-inspired weighting to propagate sentiment values from a small seed set to a broader vocabulary. Our study focuses on nouns extracted from the Harry Potter corpus, which are typically underrepresented in standard sentiment resources. Experimental results demonstrate that ECG achieves effective sentiment propagation within 1500 iterations, reaching stable convergence. Evaluation using the Coefficient of Variation (COV) indicates improved sentiment differentiation, with broader coverage compared to benchmark lexicons such as SentiWordNet, Valence Aware Dictionary and Sentiment Reasoner (VADER), and National Research Council (NRC). For instance, ECG consistently assigned interpretable sentiment values to narrative-specific entities (e.g., “Slytherin”, “Voldemort”), which were often missed by conventional lexicons. These findings highlight ECG’s scalability, interpretability, and suitability for low-resource scenarios. Beyond literary analysis, the framework can be extended to other domains such as social media, customer feedback, and specialized corpora. Overall, ECG represents a resource-efficient alternative that bridges the limitations of static lexicons and enhances sentiment inference in diverse contexts.
 
Keywords—sentiment analysis, co-occurrence graph, sentiment propagation, triadic closure, lexical expansion


Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).