Understand Online Medical Consultation Based on Dialogue State Tracking
Beiyu Xu, Wei Zhong Ren · 2020
With the continuous development of online clinical inquiry system, Chinese medical online inquiry community emerges actively. One of the core task of medical inquiry system is to extract the medical information from dialogue with specific significance for treatment including self-reported diseases, treatment means, doctor's advices and other kind of useful entities. However, due to the specific difficulties, such as colloquialism in dialogue text, it is still difficult to extract key medical information exactly. To alleviate this issue, this paper proposes a new neural network based method of information extraction for medical online consultation system with a coarse-granularity data annotation approach, which is more time-saving and robust compared with the traditional sequence labelling methods. Experimental results prove the effectiveness of our method.