This talk presents my PhD work on how to approach semi-structured data as a novice or as a data scientist who has to work with new data. The two main axes are: (a) E-R diagrams built out of any structured or semi-structured dataset ; (b) entity-to-entity path enumeration and ranking.