SEntFiN 1.0: Entity‐aware sentiment analysis for financial news

Abstract

Fine-grained financial sentiment analysis on news headlines is a challenging task requiring human-annotated datasets to achieve high performance. Limited studies have tried to address the sentiment extraction task in a setting where multiple entities are present in a news headline. In an effort to further research in this area, we make publicly available SEntFiN 1.0, a humanannotated dataset of 10,753 news headlines with entity-sentiment annotations, of which 2,847 headlines contain multiple entities, often with conflicting sentiments. We augment our dataset with a database of over 1,000 financial entities and their various representations in news media amounting to over 5,000 phrases. We propose a framework that enables the extraction of entity-relevant sentiments using a feature-based approach rather than an expression-based approach. For sentiment extraction, we utilize 12 different learning schemes utilizing lexicon-based and pretrained sentence representations and five classification approaches. Our experiments indicate that lexicon-based N-gram ensembles are above par with pretrained word embedding schemes such as GloVe. Overall, RoBERTa and finBERT (domain-specific BERT) achieve the highest average accuracy of 94.29% and F1-score of 93.27%. Further, using over 210,000 entity-sentiment predictions, we validate the economic effect of sentiments on aggregate market movements over a long duration.

Enhancing portfolio management using artificial intelligence: literature review

“…Comparatively, Zhao et al ( 2020 ) proposed a RoBERTa as a pre-trained model, which exploits different fine-tuning methods for sentiment analysis and critical entity detection in online financial texts. SEntFiN 1.0 is the most recent publicly available example of a human-annotated dataset of news headlines containing multiple entities (Sinha et al, 2022 ). The authors concluded that deep bidirectional pre-trained language models such as domain-specific BERT fine-tuned to SEntFiN outperform state-of-the-art learning schemes significantly.…”

Investigate How Market Behaves: Toward an Explanatory Multitasking Based Analytical Model for Financial Investments

“…1) SentFiN 1.0: Entity-aware sentiment analysis for financial news: [33] This human-annotated dataset contains 10,753 news headlines with entity-sentiment annotations, of which 2,847 headlines contain multiple entities, often with conflicting sentiments. For our work purpose, we have dropped the conflicting sentiments.…”

A Review of Datasets for Aspect-based Sentiment Analysis

“…For instance, healthcare datasets could facilitate sentiment analysis of patient feedback, improving healthcare service quality. ABSA can help identify investor sentiment towards dif-ferent aspects of a company or its financial performance, potentially predicting stock market trends (Sinha et al, 2022;de França Costa and da Silva, 2018;Ong et al, 2023;Hridoy et al, 2021). Education datasets could uncover student sentiments towards specific aspects of the learning environment, leading to targeted improvements (Alassaf and Qamar, 2020).…”