In recent years, the field of natural language processing has considerably benefited from larger-scale models, better training strategies, and greater availability of data, exemplified by GPT-3 etc. It has been shown that these pre-trained language models can effectively characterize linguistic patterns in text and generate high-quality context-aware representations. However, these models are trained in a way that leverages only input-output pairs. As a result, These models struggle to capture external world knowledge such as named entities and their relations, common sense, and some domain-specific content. So knowledge is important for language representation and should be included into the training and inference of language models. Knowledge is also an indispensable component to enable higher levels of intelligence which is unattainable from statistical learning on input text patterns
*Call for Papers
We invite submission of papers describing innovative research methods and applications in knowledge-enhanced NLP. Papers that introduce new theoretical proofs or methods, help to develop a better understanding of new emerging concepts topics extensive empirical experiments, or demonstrate a novel application in natural language processing of these methods to a domain are strongly encouraged.
The topics include but are not limited to the following:
Knowledge-augmented language model pre-training
Knowledge-augmented language model fine-tuning
Knowledge retrieval from unstructured data
Knowledge retrieval from structured data
NLP methods augmented by knowledge graphs
NLP methods augmented by commonsense
NLP methods augmented by heuristic rules
NLP methods augmented by dictionary
NLP methods augmented by linguistic features
NLP methods augmented by retrieved texts
NLP methods augmented by large language models
*Paper submission
Regular papers are limited to a total of 7 pages, excluding all references. We also encourage extended abstracts and short papers, which can be ranged from 2 to 4 pages excluding all references. All submissions must be in PDF format and formatted according to the new Standard AAAI Conference Proceedings Template. Following this AAAI 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.
*Submission link: https://cmt3.research.microsoft.com/KnowledgeNLP2023/Submission/Index
*Dates:
Paper submission deadline: Nov. 4, 2022
Author notification: Nov. 18, 2022
Camera ready deadline: Dec. 2, 2022
*Workshop organization: Chenguang Zhu (Microsoft Cognitive Services Research), Shuohang Wang (Microsoft Cognitive Services Research), Meng Jiang (Univerisity of Notre Dame), Wenhao Yu (Univerisity of Notre Dame), Lu Wang (University of Michigan), Huan Sun (Ohio State University)