Using Information Uncertainty for Keywords Extraction in Language System

Determining and distinguishing the capability of a node is an extremely important and appealing issue in complex networks, especially in many text processing tasks. There are many factors to inuence the evaluation, such as the topology structure character, the related statistical information and the transmission mechanism of the network. However, few previous studies are considered this issue from the information uncertainty or information missing of the network system in the transmission mechanism therein. In this paper, we focus on discussing the evaluation method of the importance of the node in the language system, in order to nd a more suitable evaluation and quantized metric to accomplish a well understanding of the text from the uncertainty of the whole system point of view. For doing this, a novel metric is proposed based on entropy to evaluate the importance of a node. The experimental results on actual information retrieval demonstrate that our approach can obtain well understanding of a text.