Abstract: The need of inductive learning of complex structure grows rapidly recently.The approaches adopted by inductive learning of complex structure can be classified into logic-based ones and graph-based ones according to knowledge representation.This paper firstly provides a survey of background and context from which inductive learning from complex structured data arises.Secondly，different learning approaches based on different knowledge representation are introduced，analyzed and compared.Finally，several challenging researching problems are identified.

Key words: complex structure, inductive learning, graph-based concept learning, inductive logic programming, higher-order logic

摘要： 复杂结构归纳学习的需求近年来快速增长。复杂结构归纳学习方法按照知识表示方式不同分为基于逻辑的方法与基于数学图的方法。阐述了复杂结构归纳学习研究的历史沿革，介绍、分析和对比了不同知识表示方式下的学习方法，给出了复杂结构归纳学习将来发展面临的挑战和需重点解决的问题。

关键词: 复杂结构, 归纳学习, 基于图的概念学习, 归纳逻辑程序设计, 高阶逻辑