Pandora: Leveraging Code-driven Knowledge Transfer for Unified Structured Knowledge Reasoning

2025-08-25

AI Summary

Existing unified structured knowledge reasoning (USKR) approaches rely on task-specific strategies or customized representations, limiting their generalizability across diverse tasks. To address this, we propose a code-driven unified reasoning framework: (1) adopting Python Pandas APIs as a universal knowledge representation backbone, mapping tabular data, relational databases, and knowledge graphs into a homogeneous code space; (2) introducing a cross-task memory mechanism that leverages executable code feedback to adaptively refine large language model (LLM) reasoning; and (3) integrating code-aware pretraining with structured knowledge transfer. Evaluated on six benchmarks spanning three structured knowledge modalities—tables, databases, and knowledge graphs—our method significantly outperforms prior USKR approaches and matches the performance of task-specialized models. It achieves, for the first time, genuine joint reasoning and generalization across heterogeneous structured knowledge sources.

Technology Category

• Knowledge Representation and Reasoning: Knowledge Acquisition
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Reasoning under Uncertainty: Uncertainty Representations

Application Category

• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Unified Structured Knowledge Reasoning (USKR) aims to answer natural language questions by using structured sources such as tables, databases, and knowledge graphs in a unified way. Existing USKR methods rely on task-specific strategies or bespoke representations, which hinder their ability to dismantle barriers between different SKR tasks, thereby constraining their overall performance in cross-task scenarios. In this paper, we introduce extsc{Pandora}, a novel USKR framework that addresses the limitations of existing methods by leveraging two key innovations. First, we propose a code-based unified knowledge representation using extsc{Python}'s extsc{Pandas} API, which aligns seamlessly with the pre-training of LLMs. This representation facilitates a cohesive approach to handling different structured knowledge sources. Building on this foundation, we employ knowledge transfer to bolster the unified reasoning process of LLMs by automatically building cross-task memory. By adaptively correcting reasoning using feedback from code execution, extsc{Pandora} showcases impressive unified reasoning capabilities. Extensive experiments on six widely used benchmarks across three SKR tasks demonstrate that extsc{Pandora} outperforms existing unified reasoning frameworks and competes effectively with task-specific methods.

Problem

• Unified reasoning across structured knowledge sources
• Overcoming task-specific limitations in cross-task scenarios
• Enhancing knowledge transfer with code-based representation

Innovation

• Code-based unified knowledge representation using Pandas API
• Knowledge transfer through cross-task memory building
• Adaptive reasoning correction via code execution feedback