S-PIC4CHU: Semantics-Enriched Techniques for Data Preparation in Data Science

The S-PIC4CHU project deals with the crucial issue of data preparation for Data Science and Machine Learning, and aims to offer new models and techniques for fighting inaccuracy, noise, uncertainty, bias, and incompleteness of data. While, at the core, the project embraces a semantics-based approach, the proposed data preparation pipeline includes data cleaning —also from the ethical viewpoint—, transformation, reduction as well as deduplication, error detection, missing value imputation, and space transformations for multimedia data. This paper illustrates the advancements on all these fronts, achieved during the first months of work on the project, and sets out the forthcoming actionable objectives.