Discovering the Knowledge Hierarchy of MathWorld for Web Intelligence

One important demonstration of Web intelligence is learning oriented search, which means to find the right knowledge from the right place and present to the user in the right order so that the user can learn fast and manage the knowledge easily. The most difficult question here is how to present the knowledge to the user for fast learning, which requires to discover the knowledge hierarchy from the Web automatically. We study this problem using the MathWorld as an example, which is an online encyclopedia of mathematics. We study various statistical properties of the network, and find that the network of mathematical knowledge is a small world and is not created through preferential attachment. Only a small number of knowledge with large between ness centralities are the basic and important mathematical knowledge, and should be learned first by a user. We hope this work sheds insight to the study on Web intelligence in general.