An Extensible Model for Turkish Single Document Summarization

Abstract

The development of Automatic Text Summarization systems with the help of a computer is one of the most challenge for two decades. The amount of information currently available and the time problem of those who need information reveal the importance of such summaries.

Generally, successful summaries are produced by people. However, it is obvious that human resources will not be enough to summarize the daily flow of information. Today, work is focused on designing and improving systems that automatically extract the abstract from the text.

In this study, a summarization system based on a parametric method has been proposed and implemented. It summarizes a single Turkish text. The compression ratio of the summary can be determined by the user. The system also offers title and keyword support to the user. The evaluation reveals an average model, however the hardness of Turkish language considered it is promising.