中文摘要：

摘要利用网络科技信息支持科技动态监测和战略决策分析日益成为情报机构的一项重要工作，但面对良莠不齐的海量网络科技信息，如何快速而准确的对语篇论述的主题进行揭示，则是一个亟待解决的重要问题。本文通过把语篇映射成一个蕴含对象语法信息、语义信息、位置信息、共现信息、分布信息以及语篇结构信息的对象网格，将非结构化的语篇转变为可计算的知识单元。根据对象在网格中的分布规律，以及由这些规律凝练的对象凝聚度、活跃度和生命跨度三个指标维度，对语篇中重要的对象进行识别。利用具有明确概念的知识对象对语篇中蕴含的重要情报线索进行揭示。

英文摘要：

Nowadays, using web scientific information to support the analysis of dynamic monitoring and strategic decision-making is becoming an important task for intelligence analysis teams. How to correctly and quickly identify the core topic of a discourse from a large number of gathered web resources has become an urgent issue. In this paper, A new discourse representation model - object grid is used to map unstructured discourse into computable knowledge unit, which contains the information about object＇s grammatical role, semantic information, position information, co-occurrence information, distribution information, and discourse structure information. According to the distribution laws of objects in the grid, three dimensions of indicators about object＇s cohesion, activity and life span is used to identify important objects. Using explicit concepts reveal important intelligence clues inherent in the discourse.