Kapsama katsayisi tabanli kümeleme ile belge özetleme

Multimedia documents, especially the text documents are occupying an important time in our daily life. Reading the summery of documents will allow us to spend less time and understand the document faster. A summary consist of the subset sentences of documents and they needs to cover the main content of document as a brief. In this paper, we have proposed an automatic, language independent text summarization system, that is based on Cover Coefficient-Based Clustering Methodology (C3M) With our methodology C3M extracts the summary sentences from the similarity between the document sentences where their features extracted by Natural Language Processing (NLP) methods such as Term Frequency (TF) and Feature Extraction. Performance evaluations evaluated on Turkish document summaries were approved by 10 human evaluators such as research assistants and graduate students. We have evaluated our machine summaries with using Precision, Recall and Rouge-N evaluation methods. Performance evaluation results shows that our system results are better than most of similar systems.