CSV-X: A Linked Data Enabled Schema Language, Model, and Processing Engine for Non-Uniform CSV

摘要

At present there are numbers of tools and schemas to aid in publishing CSV as RDF. However, there are still many non-uniform CSV datasets, which cannot effectively be processed using available solutions. This paper proposes CSV-X, a novel schema language, model, and processing engine for non-uniform CSV enabling annotation, validation, cross-referencing, linked data, RDF serialization, and transformation to other formats. Common non-uniform patterns are identified and used as a basis to create language constructs that can describe these patterns. The value reference via dynamic variable declaration making it possible to describe arbitrary relations between schema entities. Each schema entity can be mapped with a parameterizable template, allowing flexible transformation to other formats. The language processor is implemented and evaluated against real-world non-uniform CSV datasets. The results show that CSV-X can effectively process CSVs with non-uniform patterns into more advance data model in RDF.