Relevant Works

Overview

• Results of the fifth edition of the BioASQ Challenge

Baselines

• OAQA at BioASQ3B
• Learning to Answer Biomedical Questions: OAQA at BioASQ 4B
• Tackling Biomedical Text Summarization: OAQA at BioASQ 5B

Query Processing and Document Retrieval

• A Multi-strategy Query Processing Approach for Biomedical Question Answering: USTB PRIR at BioASQ 2017 Task 5B

Document Ranking

• A Multi-strategy Query Processing Approach for Biomedical Question Answering: USTB PRIR at BioASQ 2017 Task 5B
• Biomedical Question Answering via Weighted Neural Network Passage Retrieval
• Seq2seq-Attention Question Answering Model
• Teaching Machines to Read and Comprehend
• Reading Wikipedia to Answer Open-Domain Questions
• Siamese Recurrent Architectures for Learning Sentence Similarity
• Learning to Rank Short Text Pairs with Convolutional Deep Neural Networks
• How to predict Quora Question Pairs using Siamese Manhattan LSTM

Answer Selection

• Macquarie University at BioASQ 5b – Query-based Summarisation Techniques for Selecting the Ideal Answers
• Making Neural QA as Simple as Possible but not Simpler (FastQA)
• Neural Question Answering at BioASQ 5B
• Neural Domain Adaptation for Biomedical Question Answering
  • Contains comparison w/o SQuAD

Resources

• Word Embedding

Source	Year	Dimension	Size	Corpus
Biomedical natural language processing	2013 Dec	200	22,120,269 articles (PubMed), 672,589 articles (PMC), 24,181,640 types, 5.5B tokens in total	PubMed/PMC [+Wikipedia]
BioASQ word vectors	2014 Mar	200	10,876,004 English abstracts, 1,701,632 distinct words	PubMed
GloVe	2015 Oct	25-300	6-840B tokens	Wikipedia 2014 + Gigaword 5/Common Crawl/Twitter