Stability analysis of incremental concept tree for concept cognitive learning

Abstract

Incremental concept cognitive learning is an emerging active issue that concerning incremental concept learning and dynamic knowledge processing in the dynamic context environments. As the development of that, how to measure the stability of a concept structure in the view of concept cognitive learning has become an urgent issue need to be solved. Motivated by that, we propose a method to analyze the stability of incremental concept tree (ICT) for concept cognitive learning. First, the incremental concept cognitive learning process is visualized in the form of ICT, and then the structure similarity based on concept subtree and the node similarity based on concept importance are carried out to measure the similarity along the evolution of ICT. At last, the global similarity of ICT is obtained by integrating these two similarity measurements by the regulatory factor according to different scenarios. Some numerical experiments compared with classical tree similarity algorithm were conducted to evaluate the effectiveness of our method. The results show that our method is effective to analyze the stability of concept cognitive learning by measuring the similar of ICT and promising to expand it to the field of incremental concept cognitive learning.