Introduction

Huge dataset of hundreds of class are usually full of mistakes, incompletions and lack a clear definition for the annotation semantics. The general belief that smart models will eventually figure out semantics by itself is not satisfactory. We think it's possible to bring some lightweight structure and completeness to these datasets. We show that doing allows some serious improvements such as mistake cleaning and data augmentation, to the benefit learning algorithms. We also show that this allows to use semantic metrics on models. Last but not least we present our work on a Music Tag Dataset, built from Deezer Playlists Titles as annotation as an example of lightly structured, massively multi-label dataset for machine learning experiments. 

• Class identity or labels on large datasets often come from poorly qualified sources, and most importantly the number and background of annotators is probably high which will inevitably lead to annotation noise in the form of mistakes, incompletions and more worryingly an absence of consensus on the semantics. A major example of this is the Google Audioset, where thousands of labels of many different nature are combined to be used seemingly together by learning algorithm.
• The relationship between a tag and an item is rarely defined itself (except if the tagging arise from a well defined ontology).