Call for Participants: The 1st AraFinNLP Shared Task 2024
Shared-task URL: https://sina.birzeit.edu/arbanking77/arafinnlp/ Registration: https://forms.gle/rAZkRuiiEdjzTDedA
We are excited to announce the first Arabic Financial NLP (AraFinNLP) shared task, which aims to advance Arabic Financial Natural Language Processing (NLP). In today's dynamic financial landscape, the Arab world is witnessing significant growth in its stock markets and financial sectors, highlighting the pivotal role of Financial NLP in this context. AraFinNLP shared task is part of the 2nd ArabicNLP conference, which is co-located with ACL 2024 in Bangkok, Thailand, scheduled for August 16, 2024.
Shared Task Description
The AraFinNLP shared task comprises two subtasks:
Subtask 1: Multi-dialect Intent Detection
Subtask-1 involves Cross-dialect Intent Detection in the banking domain. This subtask is centred around developing models that can accurately classify customer intents from queries in various Arabic dialects. Participants will be provided with a dataset containing queries in English and their corresponding MSA and Palestinian translations. The challenge lies in training models that not only understand MSA and Palestinian but are also adept at interpreting an array of Arabic dialects, such as Gulf, Levantine, and North African. This subtask aims to foster the creation of NLP tools that can seamlessly interact across the diverse linguistic landscape of the Arab world, enhancing customer experience and operational efficiency in the banking sector. The evaluation process for this subtask, combined with the second subtask of Dialectical Translation for Intent Detection, will be designed to comprehensively assess the effectiveness of these models in a real-world, multi-dialectal context.
Subtask 2: Cross-dialect Translation and Intent Preservation
Subtask-2 focuses on the translation from MSA to various Arabic dialects: Palestinian, Saudi, Tunisian, and Moroccan, with the Palestinian dialect provided as a training set. The objective is to retain the original intent in the translated dialects, ensuring that the intent detection is as effective as the MSA examples. This subtask will test the ability of models to adapt MSA banking queries into dialectal Arabic while preserving the semantic integrity, crucial for accurate intent classification, despite the potential complexities introduced by dialectal variations. In Subtask-2, participants will receive a parallel dataset containing banking queries in both MSA and Palestinian Arabic, and their associated intents. The subtask at hand is to accurately translate these queries into the following Arabic dialects: Palestinian, Saudi, Tunisian, and Moroccan. Participants will use the intent labels to examine if the resulting translations' intents are preserved and aligned with the original MSA queries.
For more details please visit: https://sina.birzeit.edu/arbanking77/arafinnlp/
Important Dates
- February 24, 2024: Shared task announcement. - March 1, 2024: Release of training and development datasets - April 5, 2024: Registration deadline. - April 10, 2024: Test set made available. - April 22, 2024: Codalab Test system submission deadline. - May 17, 2024: Shared task papers due date - June 17, 2024: Notification of acceptance - July 1, 2024: Camera-ready papers due - August 16, 2024: ArabicNLP conference
Organising Committee
- Mo El-Haj, Lancaster University, United Kingdom - Houda Bouamor, Carnegie Mellon University, Qatar - Saad Ezzini, Lancaster University, United Kingdom - Ismail Berrada, Mohammed VI Polytechnic University, Morocco - Sanad Malaysha, Birzeit University, Palestine - Mohammed Khalilia, Birzeit University, Palestine - Mustafa Jarrar, Birzeit University, Palestine - Sultan Almujaiwel, King Saud University, Saudi Arabia
Registration
Registration is now open up until 5 April 2024: https://forms.gle/MuQb9H1Gq2nQoFUg8
We look forward to your participation in the AraFinNLP shared task.
Best wishes,
AraFinNLP Organising Committee
--------------------------------
Dr Mo El-Haj Senior Lecturer in NLP Director of Admissions (SCC) Co-Director of UCREL NLP Group
Natural Language Engineering (NLE) Journal Editorial Board https://www.cambridge.org/core/journals/natural-language-engineering
Advisory Board of the Natural Language Processing Book Series https://benjamins.com/catalog/nlp
School of Computing and Communications, Lancaster University https://www.lancaster.ac.uk/staff/elhaj @DocElhaj
You may receive emails from me outside what are your typical office hours. I do not expect you to respond to my email outside your working hours.