中文摘要：

Sowa在1984年提出了一个抽象模型，即概念图，作为基于语言学、心理学和哲学的知识表示语言。但是其概念图的定义比较简单，对于刻画文本中的语言组块的语义及语言单元之间的组合运算比较困难。为了能从形式上为将来的组块内部和组块之间的组合分析提供支持，深化概念图的表示形式，我们提出了一种适用于自然语言自动化分析的概念图的形式化描述，它可以看成是Sowa概念图的递归扩展。然后我们将这种新颖的知识表示方式应用到文本检索领域的文本标引中，同时给出了以此为基础的概念图匹配算法。在和布尔检索的比较实验中，概念检索体现出比较明显的优势。

英文摘要：

Sowa proposed an abstract model, Conceptual Graph, in 1984, which is a knowledge representing language based on linguistics, psychology, and philosophy. But it is difficult to describe the semantic meaning of chunks and the combined computation of semantic units with his relatively simple definition. In order to provide support for the combined analysis of chunks and intensify the form of Conceptual Graph, we put forward a formal description of Recursive Conceptual Graph, which could be easily adapted to the automatic analysis of natural languages and it can be regarded as a recursive extension of Sowa＇s Conceptual Graph. And then we apply it to the indexing of texts in region of text retrieval, and meanwhile propese a matching algorithm. In the experiment, Conceptual Retrieval have quite a obvious advantageous over Boolean Retrieval.