A New Effective Criterion to Select Sentences in Extractive Text Summarization

This paper introduces a new criterion to get better performance in selecting the most important sentences of text for extractive text summarization. There are two kinds of criteria to find the most relevant sentences of text: statistical criteria and semantic relations between text sentences. The proposed technique is a statistical criterion. The idea behind our approach is to consider the position of sentence words relative to words in the sentence occurring in title and keywords. We evaluate this criterion in combination with other statistical criteria. The results show that using this criterion in selecting the most important sentences of text has good results