arXiv 2503.227482025-03-27cs.LGcs.AI

以SPARK点燃预测：一种用于时序知识图谱预测中优化LLM的高效生成框架

Ignite Forecasting with SPARK: An Efficient Generative Framework for Refining LLMs in Temporal Knowledge Graph Forecasting

School of Cyberspace Science, Harbin Institute of Technology（哈尔滨工业大学网络空间安全学院）

Gongzhu Yin, Hongli Zhang, Yi Luo, Yuchen Yang, Kun Lu, Chao Meng

备注：: To be published in the 30th International Conference on Database Systems for Advanced Applications (DASFAA 2025)

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

针对LLM在时序知识图谱预测中面临输入长度和资源优化问题，提出SPARK框架，通过束序列级生成和TKG适配器优化，实现高效且即插即用的预测。

摘要：

时序知识图谱（TKG）预测对于利用历史数据预测未来事件至关重要。随着大型语言模型（LLMs）的兴起，近期研究开始探索将其整合到TKG预测中，并取得了一定成功。然而，它们仍面临输入长度有限、输出生成效率低下以及优化资源密集等局限，这些削弱了其性能和实际适用性。为解决这些局限，我们提出了SPARK，一种用于在TKG预测中优化LLM的序列级代理适配框架。受推理时算法在控制生成中的启发，SPARK通过两项关键创新提供了成本效益高、即插即用的解决方案：（1）束序列级生成，将TKG预测重新定义为top-K序列级生成任务，使用束搜索在单次前向传播中高效生成下一实体分布。（2）用于优化的TKG适配器，采用传统TKG模型作为可训练的代理适配器，以利用全局图信息并优化LLM输出，克服了输入长度和资源密集型微调问题。跨多种数据集的实验验证了SPARK的预测性能、稳健的泛化能力和高效率。我们在https://github.com/yin-gz/SPARK发布源代码。

英文摘要：

Temporal Knowledge Graph (TKG) forecasting is crucial for predicting future events using historical data. With the surge of Large Language Models (LLMs), recent studies have begun exploring their integration into TKG forecasting and achieved some success. However, they still face limitations such as limited input length, inefficient output generation, and resource-intensive refinement, which undermine their performance and practical applicability. To address these limitations, we introduce SPARK, a Sequence-level Proxy-Adapting framework for Refining LLMs in TKG forecasting. Inspired by inference-time algorithms adopted in controlling generation, SPARK offers a cost-effective, plug-and-play solution through two key innovations: (1) Beam Sequence-Level Generation, which reframes TKG forecasting as a top-K sequence-level generation task, using beam search for efficiently generating next-entity distribution in a single forward pass. (2) TKG Adapter for Refinement, which employs traditional TKG models as trainable proxy adapters to leverage global graph information and refine LLM outputs, overcoming both the input length and the resource-intensive fine-tuning problems. Experiments across diverse datasets validate SPARK's forecasting performance, robust generalization capabilities, and high efficiency. We release source codes at https://github.com/yin-gz/SPARK.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。