Readers and Writers
A Reader is an FME term for the object in a translation that reads a source dataset and a Writer is the FME term for the object in a translation that writes to a destination dataset.
Each representation in the Navigator is either a Reader or a Writer, as shown here:
There is a distinction between Readers/Writers and Datasets because many FME translations work on multiple datasets. Sometimes source datasets are identical in format and structure and can be handled more efficiently by a single Reader. Sometime source datasets – although identical in format – may have different structures that require separate readers to handle them.
Datasets
The Association for Geographic Information defines a dataset as an organized collection of data with a common theme, and this is a good description of what the term means within FME.
In FME a dataset refers to a collection of data you want to translate from or to. FME can read from or write to multiple datasets, and the datasets can all exist in the same location or be dispersed in multiple locations.
Feature Types
Feature Type is the FME term that describes an identifiable subset of records within a dataset; so each dataset may contain a number of different feature types.
Common alternatives for this term are layer, feature class, and object class.
For example, a collection of spatial data may relate to Planning Info within which is information on roads, properties, and schools. In FME, Planning Info would be the name of the dataset, and roads, properties, and schools would be different feature types.