Simplifying Data Preparation for Analysis using an Ontology for Machine Data

摘要

Vehicle manufacturers gather large amounts of data through on-board sensors and other systems, for applications, such as real-time diagnostics, prognostics, design improvements, etc. However, a lot of time is spent in preparing the data for specific analyses. Moreover, this data preparation requires people having expert knowledge about various data schemas and structures used, as well as the specific domain or vehicle systems that the data pertains to. This paper proposes an approach using a formal Ontology to capture knowledge about the domain, and a reasoner to query and prepare data. Using a demonstrative example, the paper presents a comparison of the current approach to preparing data using experts with the proposed approach. The preliminary findings from the study suggest that the proposed approach is promising, and provides unique advantages specifically when faced with distributed, polymorphic data structures, that may change over time.