UMBC ebiquity

Automatic Alignment of Lecture Notes

Speaker: Andrew Fister

Start: Friday, October 17, 2008, 02:00PM

End: Friday, October 17, 2008, 03:00PM

Location: 325 ITE

Abstract: Notetaking as a phenomenon is a remarkable extension of the human brain's memory structure and ability. One definition states that "Notes can be defined as short condensations of a source material that are generated by writing them down while simultaneously listening, studying, or observing"(Piolat et. al., 2005). There are many research problems that make notetaking very intriguing indeed to the field of natural language processing. How does the source material of the notes(lecture, meeting, book, introspection) correspond to the notetaking content? Which semantic relations do non-linguistic elements of notes capture, and why aren't these elements encoded as linguistic elements? How does automatic notetaking differ from automatic summarization(Mani and Maybury, 1999;Sparck Jones 1999), indexation, and highlighting? In this talk, I will discuss the results of experiments in aligning the content of students' notes with the corresponding lecture, and resulting insights into the behavior of notetakers as well as motivation for creating an automatic notetaking system.

Host: Charles Nicholas