论文部分内容阅读
给出了在一个面向Agent、本体和关系的常识知识库中如何实现常识知识查询的过程。在查询过程中,用自然语言形式描述的查询先被转化成一种我们所定义的面向Agent和本体的描述性查询语言。接着,描述性查询被翻译成SQL查询语句。用SQL查询语句对已转化成为关系形式的常识知识库进行查询后,自然语言生成算法把以关系表为形式的查询结果转化成自然语言。实验表明,新的查询方法在查询时间上优于原有的直接在agent和本体上查询知识的旧方法。实验数据表明,查询时间缩短了大约80 %。
The process of how to search common sense knowledge in a common knowledge base facing Agent, ontology and relation is given. In the process of querying, the query described in natural language is first transformed into a kind of descriptive query language which is defined by both agent and ontology. Next, the descriptive query is translated into a SQL query. After querying the common knowledge base that has been transformed into a relational form with a SQL query, the natural language generating algorithm converts the query result in the form of a relational table into a natural language. Experiments show that the new query method outperforms the old method of querying knowledge directly on the agent and ontology. Experimental data show that the query time is reduced by about 80%.