Mobile medical question and answer system with auto domain lexicon extraction and question auto annotation

Over the past several years, China has entered an open and comprehensive two-child era, which has caused a scarcity in the medical resources available to older pregnant women. We have presented the MCQAs, which is a Chinese mother-to-child-domain question-answering system. We use deep learning as an algorithm for question classification, however, most of actual systems do not have labeled data for training deep learning models at initial. To solve those problems, in this paper, first we present an auto domain dictionary extraction method that combines the word vector and new word detection strategy. And then we automatically annotate unlabeled data, which solves the constraints of relying on a large number of labor and experts.