A number of Japanese morphological analyzers have been developed so far, but they are not accurate enough to analyze medical texts. One of the reasons for this is that conventional morphological analysis suits general-purpose texts such as newspapers more, and is not specialized for analyzing medicinal texts. In addition, the unit of morphology is smaller than the name of a drug or disease from which we want to extract the morpheme. Therefore, medical terms getting fragmented into very small pieces has been a long persistent problem. To begin with, many of the technical terms are compound nouns, and it is not possible to extract them by morphological analysis alone.
In order to solve these problems, MedNER-J is made to specialize in disease term extraction, not usual morphemes. In addition, clinically important post-processing, such as standardization to ICD10 and factual determinations, are also incorporated.
MedNER-J is equivalent to the advanced version of MedEX/J.