Domain-Specific Sentiment Lexicons Induced from Labeled Documents

Abstract

Sentiment analysis is an area of substantial relevance both in industry and in academia, including for instance in social studies. Although supervised learning algorithms have advanced considerably in recent years, in many settings it remains more practical to apply an unsupervised technique. The latter are oftentimes based on sentiment lexicons. However, existing sentiment lexicons reflect an abstract notion of polarity and do not do justice to the substantial differences of word polarities between different domains. In this work, we draw on a collection of domainspecific data to induce a set of 24 domain-specific sentiment lexicons. We rely on initial linear models to induce initial word intensity scores, and then train new deep models based on word vector representations to overcome the scarcity of the original seed data. Our analysis shows substantial differences between domains, which make domain-specific sentiment lexicons a promising form of lexical resource in downstream tasks, and the predicted lexicons indeed perform effectively on tasks such as review classification and cross-lingual word sentiment prediction.

Dimensions of Quality: Contrasting Stylistic vs. Semantic Features for Modelling Literary Quality in 9,000 Novels

“…More recent works have tested the effect of alternative features, such as sentiment analysis on reader experience (Drobot, 2013;Kim and Klinger, 2018;Brooke et al, 2015;Jockers, 2017;Reagan et al, 2016). Studies relying on sentiment analysis usually draw scores from lexica (Islam et al, 2020) or human annotations (Mohammad and Turney, 2013), to outline the sentiment arcs of narrative texts (Jockers, 2017), and have shown a correlation between reader appreciation and sentiment (Maharjan et al, 2017(Maharjan et al, , 2018. Hu et al (2021) and Bizzoni et al (2022b) modelled persistence, coherence, and predictability of sentiment arcs using fractal analysis, a method to study the dynamics of complex systems (Hu et al, 2009;Gao and Xu, 2021), finding correlations with reader appreciation (Bizzoni et al, 2021).…”

Human Emotion and Sentiment in Natural Language Understanding and Generation using Large Language Models with Limited to No Labeled Data

“…Using lexicons to classify texts is a typical solution offered for this issue. [142]. However, the lack of accuracy of lexiconbased techniques in text classification, especially in comparison to fully supervised machine learning-based methods is well known [143].…”