Generating Questions for Abstractive Answers with a Multi-Channel Fusion Encoder-Decoder Model | IEEE Conference Publication | IEEE Xplore