Continuous sentiment scores for literary and multilingual contexts

2025-08-20

AI Summary

Traditional sentiment analysis methods struggle with metaphorical expressions, stylistic ambiguity, and cross-lingual/cross-temporal affective variation in literary texts, while also exhibiting limited performance on low-resource languages; moreover, discrete label outputs from Transformer-based models hinder fine-grained affective modeling. This paper proposes a continuous sentiment scoring method based on concept vector projection—the first application of this technique to multilingual literary sentiment analysis—yielding real-valued sentiment scores rather than categorical labels. The model jointly optimizes multilingual literary semantic representations with human annotation distributions. Evaluated on English and Danish literary corpora, it significantly outperforms existing tools, achieving high alignment with human judgments (Pearson’s *r* > 0.92). The resulting continuous scores enable robust fine-grained modeling of sentiment arcs across narrative time.

Technology Category

• Natural Language Processing: Sentiment Analysis, Stylistic Analysis, and Argument Mining
• Cognitive Modeling & Cognitive Systems: Affective Computing
• Application Domains: Humanities & Computational Social Science

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Sentiment Analysis is widely used to quantify sentiment in text, but its application to literary texts poses unique challenges due to figurative language, stylistic ambiguity, as well as sentiment evocation strategies. Traditional dictionary-based tools often underperform, especially for low-resource languages, and transformer models, while promising, typically output coarse categorical labels that limit fine-grained analysis. We introduce a novel continuous sentiment scoring method based on concept vector projection, trained on multilingual literary data, which more effectively captures nuanced sentiment expressions across genres, languages, and historical periods. Our approach outperforms existing tools on English and Danish texts, producing sentiment scores whose distribution closely matches human ratings, enabling more accurate analysis and sentiment arc modeling in literature.

Problem

• Analyzing nuanced sentiment in literary texts
• Overcoming limitations of categorical sentiment labels
• Addressing underperformance in low-resource languages

Innovation

• Continuous sentiment scoring via concept vector projection
• Trained on multilingual literary data for nuanced expression
• Outperforms existing tools with human-aligned distribution accuracy