Knowledge graphs for global and regional geologic time scales and an associated R package

Abstract:

Data heterogeneity is one of the challenges for data-intensive geoscientific study. One typical heterogeneity is at the semantic level. For example, scientists may use different words to represent the same concept, or a same word may have different meanings in different databases. Geologic time scale is a such system of concepts that faces these problems of semantic heterogeneity. International geological time scale (IGTS) has different versions (e.g. 2012 version and 2020 version) that may have different boundary ages for a same geological time concept. Moreover, different regions may have their own geological time scale concepts (e.g. China and North America). These heterogenous concepts are widely used in geoscientific publications and thus appear in lots of data and metadata in data-intensive research. Such heterogeneity impedes the data collection, interoperability, and data-driven discovery. To address these problems, we created knowledge graphs for several regional geologic time scale (RGTS), including North China, South China, North America, New Zealand, British, etc. Along with our previously constructed knowledge graph of all versions of IGTS, they are integrated into one knowledge base. To facilitate querying the knowledge of the concepts of IGTS and RGTS and interacting with other databases, we built an R package: DeepTimeKG. A use case of interaction between our knowledge graph, DeepTimeKG and Paleobiology Database (paleobiodb.org) has been established. It demonstrates that purely using the names of regional geologic concepts does not work in Paleobiology Database, but integrating our work will complement the workflow. We are also designing other use cases in mineral evolution to apply the knowledge base to leverage existing data resources and build executable and reproducible workflows. The developed knowledge base will be made open and shared to support a national open knowledge network.