Research on web news keyword extraction and theme-oriented feature automatic construction

This paper proposes an algorithm to extract key words from web news, taking into consideration of the word´s location importance. Furthermore, the algorithm is built into the framework of constructing the web news´ feature automatically based on its theme. Then the algorithm and framework proposed in this paper are tested with an experiment, using news reports on the emergency event, A (H1N1) influenza event. The results demonstrate that the proposed method is effective.