FiNCAT-2: An enhanced Financial Numeral Claim Analysis Tool

In recent times, investors tend to use digital financial content available online while making investment-related decisions. Numerals presented in such content may be claims of executives for alluring them. Numerals that represent claims i.e. ‘in-claim’ numerals may not necessarily be real facts. In...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Software impacts 2022-05, Vol.12, p.100288, Article 100288
Hauptverfasser: Ghosh, Sohom, Naskar, Sudip Kumar
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:In recent times, investors tend to use digital financial content available online while making investment-related decisions. Numerals presented in such content may be claims of executives for alluring them. Numerals that represent claims i.e. ‘in-claim’ numerals may not necessarily be real facts. Investors should not get swayed by the information presented using ‘in-claim’ numerals. We developed an open-source tool, FiNCAT-2 that distinguishes ‘in-claim’ numerals from those which are not (i.e. ‘out-of-claim’ numerals). It consists of a logistic regression model trained using contextual BERT embeddings (768-dimensions) of the target numeral. It can be accessed from GitHub, Google Colab and HuggingFace Spaces. •FiNCAT-2: is an enhanced Financial Numeral Claim Analysis Tool.•It helps investors to detect numerals in financial texts which are claims and not necessarily facts.•This tool has been permanently hosted on HuggingFace Spaces.•It can also be accessed from Google Colab.•Link to the demonstration video: (https://www.youtube.com/watch?v=5GmtiOKWSwc).
ISSN:2665-9638
2665-9638
DOI:10.1016/j.simpa.2022.100288