Data Refinery is a capability designed to make it easy for business users and data scientists to wrangle and transform data. When working with csv files, there is often a series of steps to go through to define the schema / datatypes before anaysis / offload to DB / Watson Analytics, which can take a long time if the dataset is quite wide.
It is clear from the interface that Data Refinery is able to infer data types (e.g. integers) from the data when you start working in refinery. It would be good if you could "take recommended actions" on your data to speed up the process to generating insight.