This task focuses on detecting causal effects in financial disclosures using a hybrid QA approach, involving extractive and generative QA with datasets in English and Spanish.
Website: https://www.lllf.uam.es/wordpress/fincausal-25/fnp-2025/
This task involves querying and reasoning from a database for financial QA, requiring models to generate code for querying and reasoning based on questions. The dataset for this task is from [1].
Website: https://dbqr-qa.github.io
[1] Rungsiman Nararatwong, et al. 2024. DBQR-QA: A Question Answering Dataset on a Hybrid of Database Querying and Reasoning. In Findings of the Association for Computational Linguistics: ACL 2024.Â
Participants will evaluate the capability of LLMs to make investment decisions in cryptocurrency trading. Performance will be assessed using FinMem, with data for Bitcoin (BTC) and Ethereum (ETH) provided for model fine-tuning.
Website: https://coling2025cryptotrading.thefin.ai
Participants will develop or fine-tune LLMs to navigate complex regulatory texts and industry standards through various tasks, enhancing the ability of LLMs to interpret and apply regulations and industry standards in the financial sector.
Website: https://coling2025regulations.thefin.ai/
This task challenges participants to identify and explain instances of false financial news using the FIN-FACT dataset. Evaluation metrics include Accuracy, Precision, Recall, Micro-F1, ROUGE, BERTScore, and BARTScore.
Website: https://coling2025fmd.thefin.ai/