Towards Time-resilient MIR Processes

Abstract

In experimental sciences, under which we may likely sub- sume most research areas in MIR, repeatability is one of the key cornerstones of validating research and measuring progress. Yet, due to the complexity of typical MIR exper- iments, ensuring the capability of re-running any experi- ment, achieving exactly identical outputs is challenging at best. Performance differences observed may be attributed to incomplete documentation of the process, slight vari- ations in data (preprocessing) or software libraries used, and others. Digital preservation aims at keeping digital objects authentically accessible and usable over long time spans. While traditionally focussed on individual objects, research is now moving towards the preservation of entire processes. In this paper we present the challenges of pre- serving a classical MIR process, i.e. music genre classi- fications, discuss the kinds of context information to be captured, as well as means to validate the re-execution of a preserved process.