We propose a focused FinEval 2026 shared task, Multilingual Financial Language Understanding. The goal is to keep the challenge operationally lightweight while covering three complementary abilities: 

• 1) Entity extraction in Greek, 

• 2) Pragmatic intent recognition in Japanese,

• 3) Cross-document multilingual financial QA. 

The task is intentionally narrower than a general multilingual benchmark: each component targets a phenomenon that is practically relevant in financial communication and difficult to solve through generic translation or surface fluency.

Participants may take part in one, two, or all three subtasks. We sincerely welcome students, researchers, and engineers passionate about financial technology, LLMs, and language agent design to participate in this challenge!

Task Description

Based on Plutus, models perform Greek financial numeric NER (`GRFinNUM`) and textual NER (`GRFinNER`) over company filings. The task evaluates whether systems can recognize financial quantities, accounts, organizations, and domain-specific entities in morphologically rich Greek text.

Dataset Description

GRFinNUM / GRFinNER: GRFinNUM/NER targets numeric spans and categories; GRFinNER targets textual financial entities. Both are annotated by expert native Greek speakers with finance and linguistic expertise.

• Dataset Download: Link

Evaluation Metrics

Entity F1, computed over predicted spans and entity categories.

Task Description

Based on EBISU, models perform Japanese Financial Implicit Communication Recognition (`JF-ICR`) over company Q&A. Given a Japanese financial question and company response, the model predicts implicit intent such as commitment, non-commitment, risk signaling, or refusal.

Dataset Description

JF-ICR: 94 expert-annotated Japanese company Q&A instances from public corporate presentation, investor-relations, and shareholder-meeting materials. Each example pairs a financial question with a Japanese company response and asks the model to classify the response's implicit communicative intent.

• Dataset Download: Link

Evaluation Metrics

Japanese ICR: accuracy over implicit communication labels.

Task Description

Based on MultiFinBen/PolyFiQA, models answer cross-document financial questions using SEC financial-report statements and contemporaneous multilingual news. 

Dataset Description

PolyFiQA-Easy / PolyFiQA-Expert: the QA examples combining SEC filing statements with news in English, Chinese, Japanese, Spanish, and Greek. Reports are reduced to core statements, including comprehensive income, consolidated balance sheets, and cash flows. Examples were expertly constructed and validated for relevance and consistency.

• Dataset Download: Link

Evaluation Metrics

ROUGE-1 against expert references, with evidence-grounding diagnostics for whether answers align with required financial-statement evidence.

• (Released)   All Public Datasets release: 28 July 2026

• (Pending)     Testset release and test phase start: 4 August 2026

• (Pending)     Test phase end: 20 August 2026

• (Pending)     Leaderboard release: 22 August 2026

• (Pending)     Systems paper submission due: 25 August 2026 

• (Pending)     Notifications of Acceptance: 1 September 2026

• (Pending)     Camera-ready Paper Deadline: 10 September 2026

• (Pending)     Workshop Date: October 28 2026

We invite academic institutions, industry professionals, and individuals to participate in this exciting competition. Please register your team using the following link:

Multilingual Financial Language Understanding Challenge @ EMNLP-2026 

Please choose a unique team name and ensure that all team members provide their full names, emails, institutions, and the team name. Every team member should register using the same team name. We encourage you to use your institutional email to register.

Solution Submission Link:

TBD 

System Paper Submission:

The ACL Template MUST be used for your submission(s). Accepted papers proceedings will be published at ACL Anthology. 

• The paper title format is fixed: "[Model or Team Name] at the Multilingual Financial Language Understanding Challenge: [Title]".

• The reviewing process will be single-blind. Accepted papers will be published in the ACL Anthology.

• Shared task participants will be asked to review other teams' papers during the review period. 

• Submissions must be in electronic form using the paper submission software linked above.

• At least one author of each accepted paper should register and present their work in person at the FinNLP Workshop. Papers with “No Show” may be redacted. Authors will be required to agree to this requirement at the time of submission. It's a rule for all EMNLP-2026 workshops. 

Challenge Results

TBD

For questions about shared tasks, contact aclsigfintech@gmail.com.