A Survey on Summarization with Deep Learning

PhD Qualifying Examination


Title: "A Survey on Summarization with Deep Learning"

by

Mr. Yuxiang WU


Abstract:

With the rapid growth of textual information in Mobile Internet era, we 
are facing a challenging problem of information overload. We may receive 
dozens or even hundreds of news notifications every day, and it takes a 
long time to read all these articles. It would help save reader's time if 
we could convert these posts into a summary.

The goal of summarization is to produce concise but informative summaries 
for long documents. With the recent surge of Deep Learning and the 
availability of large-scale summary dataset, a considerable amount of 
works have been conducted to apply Deep Learning in summarization. In this 
survey, we study the development of Deep Learning-based summarization 
algorithms. These works are categorized into two branches: extractive 
summarization and abstractive summarization. We first review extractive 
approaches, which exploit Deep Learning either as a feature extractor or 
in an end-to-end fashion. Then several techniques used in abstractive 
summarization are introduced. We then review the commonly-used datasets 
and evaluation metric in this area and compare the performance of works 
presented. The survey is concluded with the discussion of a summary of 
current research status and future directions in summarization area.


Date:			Friday, 28 July 2017

Time:                  	2:00pm - 4:00pm

Venue:                  Room 2612B
                         Lifts 31/32

Committee Members:	Prof. Qiang Yang (Supervisor)
 			Prof. Lei Chen (Chairperson)
 			Dr. Qiong Luo
 			Dr. Xiaojuan Ma


**** ALL are Welcome ****