A method for Automatic Text Summarization using Consensus of Multiple Similarity Measures and Ranking Techniques

In the era of information overload, text summarization can be defined as the process of extracting useful information from a large space of available content using traditional filtering methods. One of the major challenges in the domain of extraction based summarization is that a single statistical measure is not sufficient to produce efficient summaries which would be close to human-made ‘gold standard’, since each measure suffers from individual weaknesses. We deal with this problem by proposing a text summarization model that combines various statistical measures so that the pitfalls of an individual technique could be compensated by the strengths of others. Experimental results are presented to demonstrate the effectiveness of the proposed method using the TAC 2011 Multiling pilot dataset for English language and ROUGE summary evaluation tool.