Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

OneEdit: A Neural-Symbolic Collaboratively Knowledge Editing System

Ningyu Zhang, Zekun Xi, Yujie Luo, Peng Wang, Bozhong Tian, Yunzhi Yao, Jintian Zhang, Shumin Deng, Mengshu Sun, Lei Liang, Zhiqiang Zhang, Xiaowei Zhu, Jun Zhou, Huajun Chen

TL;DR

OneEdit, a neural-symbolic prototype system for collaborative knowledge editing using natural language, which facilitates easy-to-use knowledge management with KG and LLM and can achieve superior performance on two new datasets.

Abstract

Knowledge representation has been a central aim of AI since its inception. Symbolic Knowledge Graphs (KGs) and neural Large Language Models (LLMs) can both represent knowledge. KGs provide highly accurate and explicit knowledge representation, but face scalability issue; while LLMs offer expansive coverage of knowledge, but incur significant training costs and struggle with precise and reliable knowledge manipulation. To this end, we introduce OneEdit, a neural-symbolic prototype system for collaborative knowledge editing using natural language, which facilitates easy-to-use knowledge management with KG and LLM. OneEdit consists of three modules: 1) The Interpreter serves for user interaction with natural language; 2) The Controller manages editing requests from various users, leveraging the KG with rollbacks to handle knowledge conflicts and prevent toxic knowledge attacks; 3) The Editor utilizes the knowledge from the Controller to edit KG and LLM. We conduct experiments on two new datasets with KGs which demonstrate that OneEdit can achieve superior performance.

OneEdit: A Neural-Symbolic Collaboratively Knowledge Editing System

TL;DR

OneEdit, a neural-symbolic prototype system for collaborative knowledge editing using natural language, which facilitates easy-to-use knowledge management with KG and LLM and can achieve superior performance on two new datasets.

Abstract

Knowledge representation has been a central aim of AI since its inception. Symbolic Knowledge Graphs (KGs) and neural Large Language Models (LLMs) can both represent knowledge. KGs provide highly accurate and explicit knowledge representation, but face scalability issue; while LLMs offer expansive coverage of knowledge, but incur significant training costs and struggle with precise and reliable knowledge manipulation. To this end, we introduce OneEdit, a neural-symbolic prototype system for collaborative knowledge editing using natural language, which facilitates easy-to-use knowledge management with KG and LLM. OneEdit consists of three modules: 1) The Interpreter serves for user interaction with natural language; 2) The Controller manages editing requests from various users, leveraging the KG with rollbacks to handle knowledge conflicts and prevent toxic knowledge attacks; 3) The Editor utilizes the knowledge from the Controller to edit KG and LLM. We conduct experiments on two new datasets with KGs which demonstrate that OneEdit can achieve superior performance.
Paper Structure (30 sections, 11 equations, 6 figures, 3 tables, 2 algorithms)

This paper contains 30 sections, 11 equations, 6 figures, 3 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Large Language Models.
  4. Knowledge Graph.
  5. Knowledge Editing.
  6. System Desgin
  7. Knowledge Editing for LLMs
  8. Neural-Symbolic Knowledge Editing
  9. Interpreter
  10. Controller
  11. Conflict Resolution
  12. Coverage Conflict.
  13. Reverse Conflict.
  14. Knowledge Graph Augmentation
  15. Editor
  16. ...and 15 more sections

Figures (6)

  • Figure 1: OneEdit for neural-symbolic collaboratively knowledge editing with KGs and LLMs.
  • Figure 2: The detailed workflow of OneEdit in handling conflicts: the natural language input from the user is extracted into knowledge triples by the Interpreter, then processed by the Controller to generate sequences of editing triples and rollback triples, which are finally sent to the Editor.
  • Figure 3: The variation of one-hop metrics with changes in generation triples in GPT-J-6B.
  • Figure 4: The impact of adding logical rules on the One-Hop results in OneEdit.
  • Figure 5: A coverage conflict within OneEdit: OneEdit first rolls back the conflicting knowledge before editing the new knowledge. Without using OneEdit, previous edited knowledge may remain.
  • ...and 1 more figures