DANTEStocks presents an innovative dataset comprising tweets from the stock market for NLP research, enabling emotion analysis and entity recognition.
This paper introduces DANTEStocks, a pioneer multi-layered-annotated corpus for fostering research in Natural Language Processing (NLP) of user-generated content in Portuguese. Comprising 4,048 tweets (X posts) from the stock market domain written in Brazilian Portuguese, the corpus was annotated with three standoff layers of information, that is to wit, emotion categories (following Plutchik’s wheel of emotions); part-of-speech and dependency relations (under the Universal Dependencies framework); and named entities (according to the second HAREM’s taxonomy). The DANTEStocks corpus has been constructed within the context of the POeTiSA Project, which aims at increasing the amount of linguistic resources (lingwares) and fostering the development of tools and NLP applications for Brazilian Portuguese. In this article, we address the design of the several annotation tasks carried out in DANTEStocks, reporting on the available annotation results.
No takes yet. Share an insight, caveat, or question.
Felippo et al. (2025) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: