An Exploration of Data Augmentation and RNN Architectures for Question Ranking in Community Question Answering

摘要: The automation of tasks in community question answering (cQA) is dominated by machine learning approaches, whose performance often limited the number training examples. Starting from a neural sequence approach with attention, we explore impact two data augmentation techniques on ranking performance: method that swaps reference questions their paraphrases, and examples automatically selected external datasets. Both methods are shown to lead substantial gains accuracy over strong baseline. Further improvements obtained changing model architecture mirror structure seen data.