Large Knowledge-Enhanced Models @IJCAI 2024

Overview

We are excited to announce the first International OpenKG Workshop on Large Knowledge-enhanced Models (LKM2024) held in conjunction with IJCAI 2024. The workshop aims to bring together researchers from academia and industry to discuss the latest advances and challenges on a variety of topics over knowledge-enhanced large language models in AI, and the integration of large models with symbolic KR such as KG.

Humankind accumulates knowledge about the world in the processes of perceiving the world, with natural languages as the primary carrier of world knowledge. Representing and processing these world knowledge has been central to its objectives since the early advent of AI. Indeed, both LLMs and KGs were developed to handle world knowledge but exhibit distinct advantages and limitations. LLMs excel in language comprehension and offer expansive coverage of knowledge, but incur significant training costs and struggle with authenticity problems of logic reasoning. KGs provide highly accurate and explicit knowledge representation, enabling more controlled reasoning and being immune from hallucination problems, but face scalability challenges and struggle with reasoning transferability. A deeper integration of these two technologies promises a more holistic, reliable, and controllable approach to knowledge processing in AI.

Natural languages merely encode world knowledge through sequences of words, while human cognitive processes extend far beyond simple word sequences. Considering the intricate nature of human knowledge, we advocate for the research over Large Knowledge-enhanced Models (LKM), specifically engineered to manage diversified spectrum of knowledge structures. In this workshop, we focus on exploring large models through the lens of “knowledge”. We expect to investigate the role of symbolic knowledge such as Knowledge Graphs (KGs) in enhancing LLMs, and also interested in how LLMs can amplify traditional symbolic knowledge bases.

Call for Papers

We welcome all submissions related to but not limited to the following topics:

Large model knowledge enhancement
Integration of LLM and symbolic KR
Knowledge-injecting LLM pretraining
Structure-inducing LLM pre-training
Knowledge-augmented prompt learning
Knowledge-enhanced instruction learning
Graph RAG and KG RAG
LLM-enhanced symbolic query and reasoning
Large model knowledge extraction
Large model knowledge editing
Large model knowledge reasoning
Knowledge-augmented multi-modal large models
Multimodal learning for KGs and LLMs
Knowledge-enhanced Hallucination Detection and Mitigation
Semantic tools for LLMs
Knowledgeable AI agents
Integration of LLM and KG for world models
Domain-specific LLMs training leveraging KGs
Applications of combing KGs and LLMs
Open resources combining KGs and LLMs

Submission URL: https://cmt3.research.microsoft.com/LKM2024

Format: Submissions are invited in the form of 7-page papers (with an additional 2 pages for references) for inclusion in the proceedings, or a 2-page abstract for poster and demonstration proposals. All submissions must adhere to the formatting requirements specified in the conference's author guidelines, available at https://www.ijcai.org/authors_kit. Submissions will be reviewed in a single-blind manner, and it is required to include all authors' names and affiliations. Papers that fail to adhere to the submission guidelines or fall outside the scope of the workshop's relevant topics will be desk rejected. Accepted papers will be featured in the workshop program and incorporated into the workshop proceedings, although authors may choose to opt out of this inclusion.

Dual-submission policy: We welcome ongoing and unpublished work. We also welcome papers that are under review at the time of submission, or that have been recently accepted. We welcome both accepted and rejected IJCAI 2024 papers for submission to this workshop.

Archival/Non-archival: The accepted papers may choose between two publication options: archival or non-archival. All archival papers will be published as a special issue in Data Intelligence Journal (https://direct.mit.edu/dint). Selected papers will be invited to submit extensional versions to the　Elsevier Journal of Big Data Research (SCI Indexed). To qualify for archival publication, submissions must be notably original and not previously published in other venues or journals. Non-archival papers, on the other hand, are permitted to be works that have been presented or published in another venue or journal.

Presentation: accepted papers can choose between an in-person presentation or opting out of an onsite presentation.

The workshop is organized by OpenKG, an open research community committed to the innovation on open technologies for KGs and their integration with modern large language models.

Important Dates: