Sentence Extraction for Automatic Summarization and Notetaking Systems

In this book, we propose both automatic summarization and semi-automatic Notetaking systems as important tools that can engage students in learning activities and improve their learning, namely comprehending and recalling of the study materials. However, they can also be employed by lecturers to evaluate their students understanding. Many current systems summarize texts by selecting sentences with important content using methods generally known as sentence extraction. To deal with the development of a new sentence extraction method, we delve into text analysis at three levels: word, sentence, and text level analysis. At the word level, we consider word similarity and word disambiguation based on WordNet to compute the value for semantic relatedness. This feature is exploited by the proposed method we have developed for text similarity. For the sentence level, we analyze for its similarity using vector correlation. For text similarity, a cognitive method is used to identify the most important sentence. Our proposed unsupervised sentence extraction method is then used to identify the most salient sentences to produce high quality summarization and notes.