Dates
-----------
All deadlines are 11:59 PM AoE time.
- Workshop Paper Due Date: Feb 14, 2025
- Notification of acceptance: Mar 7, 2025
- Camera-ready papers due: Mar 14, 2025
- Workshop date: May 3, 2025
(NAACL 2025 will take place in New Mexico, USA from Apr 29 to May 4th 2025.)
Workshop history
------------------
Recent advancements in large-scale language models like ChatGPT have expanded the horizons of natural language processing. These models excel at understanding complex language patterns and incorporating extensive knowledge. However, they have limitations such as struggling with rare information, generating incorrect content, and lacking access to real-time data due to fixed parameters. To address these issues, knowledge-augmented NLP methods have emerged. These methods offer a plug-and-play approach to enhance model performance by explicitly integrating external knowledge inputs. This bridges the gap between statistical learning from text patterns and comprehensive world knowledge. In essence, while large language models have made significant progress, their limitations make it crucial to integrate external knowledge. Knowledge-augmented NLP methods offer a promising solution, improving efficiency and effectiveness in language processing by empowering models to understand and utilize diverse knowledge inputs.
This workshop is the 4th edition of the workshop focusing on knowledge-driven NLP, aiming to highlight the importance of the application of knowledge in NLP methods, and to explore its role in large language models (LLMs). The workshop will discuss a range of topics including the benefits of knowledge enhancement, the latest developments, challenges faced, and practical application scenarios. The previous three workshops were held at AAAI 2023, KDD 2023 and ACL 2024, respectively, attracting over a hundred paper submissions and more than 300 participants each. This event is organized by scholars from 7 institutions in the field, reflecting its widespread attention and importance in the academic community.
List of Topics
------------------
1. All about knowledge data: automatic or human-driven approaches for gathering, curating, and constructing knowledge corpus / database.
2. All about knowledge acquisition: web-based retrieval, tool / API usage, structured knowledge and other techniques for sourcing related knowledge.
3. All about knowledge utilization: pre-training, fine-tuning with knowledge, plug-and-play knowledge injection, knowledge editing and updating in LMs
4. Knowledge in diverse modalities and novel applications: knowledge-augmented visual/code/math LMs, or applications such as chemistry, education and etc.
5. Language model as parametric knowledge base: distillation and knowledge elicitation from LLMs and applications such as generative retrieval, recommendation and etc.
Submission Guidelines
--------------------------------
We welcome 2 types of papers: archival workshop papers, non-archival papers. Only archival-track papers will be included in the workshop proceedings. All submissions should be in PDF format and made through the OpenReview website set up for this workshop (https://openreview.net/group?id=aclweb.org/NAACL/2025/Workshop/KnowledgeNLP). In line with the ACLs conference policy, camera-ready versions of papers will be given one additional page of content.
Regular papers are limited to a total of 8 pages, excluding all references and optional appendix. All submissions must be in PDF format and formatted according to the new Standard ACL Conference Proceedings Template. Following this ACL conference submission policy, reviews are double-blind, and author names and affiliations should NOT be listed. Submitted papers will be assessed based on their novelty, technical quality, potential impact, and clarity of writing. For papers that rely heavily on empirical evaluations, the experimental methods and results should be clear, well executed, and repeatable. Authors are strongly encouraged to make code publicly available.
For the archival track: Accepted papers in the archival track will be published in the official ACL anthology, categorized under ACL 2024. This provides authors with the opportunity for their work to be recognized and cited as part of the ACL proceedings.
For the non-archival track: In the non-archival track, there are no dual submission restrictions or anonymity requirements. We advise authors to consult the policies of the respective conferences, such as ACL, for specific guidance on dual submissions and anonymity.Accepted papers in the non-archival track will be made accessible on the workshop website but will not be included in the ACL proceedings. This allows authors to share their research while adhering to the policies of the corresponding conferences.
Furthermore, we invite submissions of papers on relevant topics that are currently under review or intended for submission to other venues, including the NAACL 2025 main conference and findings. Submissions must adhere to the standard format for workshop papers and will not be part of the workshop proceedings. These papers will undergo a review process by our workshop reviewers.
Organizers
----------------
Weijia Shi, University of Washington
Wenhao Yu, Tencent AI Seattle Lab
Akari Asai, University of Washington
Greg Durrett, UT Austin
Meng Jiang, University of Notre Dame
Hannaneh Hajishirzi, University of Washington / AI2
Luke Zettlemoyer, University of Washington / Meta
Read more:
https://knowledge-nlp.github.io/naacl2025