MKGB: A Medical Knowledge Graph Construction Framework Based on Data Lake and Active Learning

Medical knowledge graph (MKG) provides ideal technical support for integrating multi-source heterogeneous data and enhancing graph-based services. These multi-source data are usually huge, heterogeneous, and difficult to manage. To ensure that the generated MKG has higher quality, the construction of MKG using these data requires a large number of medical experts to participate in the annotation based on their expertise. However, faced with such a large amount of data, manual annotation turns out to be a high labor cost task. In addition, the medical data are generated rapidly, which requires us to manage and annotate efficiently to keep up with the pace of data accumulation. Prior researches lacked efficient data management for massive medical data, and few studies focused on the construction of large-scale and high-quality MKG.