TL;DR: Automatic Summarization With Textual Annotations

Abstract

Automatic summarization is a powerful means for compressing large quantities of text into manageable chunks for human consumption. Despite the growth in blogs and other platforms that facilitate human interaction with text, there have been relatively few studies aimed at incorporating the auxiliary annotation data provided by these platforms into the summarization task. In this paper, I introduce a suite of summarization algorithms that utilize textual annotations (highlights and comments) to effectively summarize text. Specifically, two of the systems developed, ATMS-A (Hdp) and the Hotspot Summarization are shown to outperform all competitive baseline systems.