Exploring a Rich Design Space for Text Summarization

Abstract

Text summarization is a critical task to deal with the ever increasing amount of information available online. There is abundant research in this area, however, because inputs and outputs of the summarization process can vary substantially, it is often difficult to compare and integrate the existing methods. To tackle this challenge, in this talk I will introduce a set of dimensions to characterize a rich design space for summarization systems. I will then present two summarization frameworks we have been working on, one for evaluative text (e.g. customer reviews) and one for conversations (e.g., discussion forums), and discuss how they fit into the proposed design space. It will be clear that our two frameworks represent points which are quite far apart in the design space. One summarizer generates abstractive summaries of evaluative text (e.g. customer reviews) by applying a probabilistic discourse parser and by aggregating the extracted information. In contrast, the summarizer of conversations generates extractive multimedia summaries by first performing topic modeling and by then visualizing the source text annotated with the resulting topics. A comparison between these two very different frameworks will hopefully shed some light on the pros and cons of different approaches to summarization and on how to move forward in this important area of NLP.