Thursday, 8 December 2022
The 1st Workshop on Mathematical Natural Language Processing
at EMNLP 2022
The 2024 edition of MathNLP will be co-located with LREC-COLING 2024
A one-day workshop on Mathematical Language Processing
The articulation of mathematical arguments is a fundamental part of scientific reasoning and communication. Across many disciplines, expressing relations and interdependencies between quantities (usually in an equational form) is at the centre of scientific argumentation. One can easily find examples of mathematical discourse across different scientific contributions and textbooks. Nevertheless, despite its importance, the application of contemporary neuro-symbolic models for performing inference over mathematical text remains under-explored, especially when compared with the advances in natural language processing and domain-specific text mining (e.g. biomedical text).
Call for papers
We welcome contributions of new, previously unpublished papers which could be either long (6-8 pages) or short (2-4 pages). All submissions should be formatted in the ARR style and will be peer-reviewed by multiple reviewers. The authors' identities must be concealed to enable double-blind peer-review, and all the conflicts of interests must be declared in advance.
MathNLP is an archival workshop. Accepted papers will be published in the proceedings.
We are particularly interested (but are not limited to) in works related to the following topics:
Neural/Neuro-symbolic architectures to support mathematical natural language inference;
Equational embeddings;
Translation from natural mathematical language to formal languages (and vice-versa);
Linguistic analysis of mathematical discourse and argumentation relations in the context of mathematical text;
Probing mathematical understanding of state-of-the-art models;
Adaptation of NLP tasks for mathematical discourse;
NLP applied to mathematics education;
Premise selection over mathematical text;
Understanding and typing of variables in mathematical text;
Retrieval of equations/formulas/expressions based on textual queries;
Retrieval of textual context based on equational queries.
Important dates
(Anywhere on Earth)
Submission deadline (extended): 1 October, 2022 10 October, 2022
Notification of acceptance: 25 October, 2022 30 October, 2022 31 October, 2022
Camera-ready paper deadline: 5 November, 2022
Workshop date: 8 December, 2022
Keynote Speaker
Dr Ashwin Kalyan, Allen Institute for Artificial Intelligence (AI2)
LLMs-as-a-Service: Harnessing the power of Foundation Models for Challenging Reasoning Problems.
Abstract:
AI systems built on top of foundation models achieve state-of-the-art performance on a wide range of tasks making them the one of the most versatile and dependable AI technology. However, even for these systems, hard reasoning problems — ones that require mathematical and algorithmic reasoning in addition to more general skills like language understanding, commonsense reasoning and computer vision — pose a significant challenge. First, I will discuss the successes and limitations of state-of-the-art LLMs on hard reasoning problems like fermi problems and challenging math word problems — encouraging the broader AI community to address this challenge in AI reasoning. Next, I propose “LLMs-as-a-Service”, a compositional and neuro-symbolic strategy to develop the next generation of AI solutions that achieve best-of-both-worlds — harness the capacity of powerful foundational models while at the same time overcoming their shortcomings in producing well-reasoned, consistent answers.
Short Bio:
Ashwin Kalyan is a scientist connecting AI, innovation and research. He led and contributed to research projects and technologies that have resulted in new perspectives of integrating AI systems with practice (e.g. neuro-symbolic approaches for program synthesis, novel decoding strategies for language models) that have impacted industry practices in addition to the wider research community. Currently, he is a researcher at the Allen Institute of Artificial Intelligence where he investigates the abilities and limitations of foundation models, especially in the context of hard reasoning problems that require mathematical and algorithmic reasoning. He has authored 20+ publications in top-tier AI conferences (e.g. NeurIPS, ICML, CVPR, ACL, EMNLP) and was recognized by the prestigious JP Morgan PhD Fellowship. He obtained his PhD from Georgia Institute of Technology and prior to that, B.Tech from National Institute of Technology Karnataka.
He started the “student researcher” program at AI2, a research apprenticeship initiative that nurtures scientific talent by providing aspiring researchers (including undergraduate and PhD students) a peek into cutting-edge AI research. He serves as the technical advisor for Youth for Creativity and Excellence (YCEF), a privately funded non-profit organization that promotes scientific, cultural and creative pursuits in India.
Organizers
Marco Valentino - IDIAP Research Institute and University of Manchester
Deborah Ferreira - The MathWorks
Andre Freitas - IDIAP Research Institute and University of Manchester
Sean Welleck - University of Washington and Allen Institute for Artificial Intelligence
Moritz Schubotz - University of Wuppertal
Programme Committee
Mokanarangan Thayaparan
Jordan Meadows
Akiko Aizawa
Somak Aditya
Matthew Finlayson
Takuto Asakura
Sponsors
