ASTON: AUTOMATIC SUMMARIZATION FOR NEWS
This is a group project for CSCE 638 Natural Language Processing Foundation and Techniques (Fall 2018). First advocate of this project, I found the DeepMind Q&A Dataset and implemented a modified version of the NetSum algorithm. Modified NetSum is found to perform the best among all three algorithms. I was also responsible for project management.
ASTON (Automatic SummarizaTion fOr News) includes three algorithms: LexRank, LSA, and NetSum. LexRank and LSA are unsupervised models, whereas NetSum is a neural-network based model. Given a sample news article and a selected algorithm, an auto summary consisting of three sentences will be produced. All algorithms are essentially extractive methods.