Automatic Text Summarization (The state of the art 2007 and new challenges)
Karel Jeæek, Josef Steinberger · 2008
The headline of this paper names a research area originating from the late 50's but not loosing its popularity until the present time. Moreover, one of the most relevant today's problems caused by the rapid growth of the Web, which is called information overloading, has increased the necessity of more sophisticated and powerful summarizers. This paper shortly introduces a taxonomy of summarization methods and an overview of their principles from classical ones, over corpus based, to knowledge rich approaches. We consider various aspects which can affect their classification. A special attention is devoted to application of recent information reduction methods, based on algebraic transformations. Further, we introduce experiences with the development of our own summarizing method. Finally, some new ideas and a conception for the future of this field are mentioned.