学习支持系统问题理解模块的架构

The Infrastructure of LPSS Question Understanding Module

学习支持系统问题理解模块的系统功能是把自然语言描述的问题转化为计算器可处理的数据结构。本文提出的问题理解模块的架构，引入了人工神经网络、事例推理、知识表示、自然语言处理等理论，以及Sequential Model，Visible Markov Model（VMM）和Hidden Markov Model（HMM）等算法，它能够从学生自然语言描述中提取出三种计算器可处理的信息：关键词、语义分析数据和问题分类数据，为学习支持系统奠定了基础。

The mechanism of LPSS Question Understanding Module is to transform questions into the representations which can be processed by computers. We developed several learning components in the module， all of which consists of stages of learning from corpus， generating expressive semantic features and making use of the outcome of other components. The module yields three kinds of information：key phrases， question analysis record and classified question data， which are the primary resources of the answer retrieval.