The Semantic Data Warehouse for the AgriProKnow Project: A First Prototype

Contemporary dairy farming heavily relies on modern technology such as milking robots, feeding systems, and various sensors which track animal movement, micro climate, etc. All these systems produce vast amounts of data. These data contain potentially valuable information that could be used to increase efficiency of dairy farm operations. As of now this potential remains underused, which the AgriProKnow project intends to change. The AgriProKnow project develops a data analysis platform as a means to extract knowledge from the information contained in the data. In this thesis we present a first prototype of the AgriProKnow project's data analysis platform in the form of a semantic data warehouse (sDWH). The sDWH is realised using a combination of semantic technologies and a relational database management system. The schema and all instance data are described in RDF format using the RDF Data Cube Vocabulary. The RDF schema is mapped to a relational data model; the instance data in the sDHW are stored in a relational database. Furthermore, the sDWH provides intuitive query facilities for the stored data, the semOLAP patterns. The semOLAP patterns are defined by database and domain experts. Each semOLAP pattern contains wildcards. Based on the semOLAP patterns, users create queries and provide concrete values for the wildcards in the pattern. The combination of the semOLAP pattern and concrete values for its wildcards results in an SQL query which is executed in the relational database of the sDWH. If the concrete values for the wildcards of the semOLAP pattern are RDF elements, the export of the query results can be done in RDF as well. The query result is enriched semantically including, a definition of the result's structure and the underlying query.