When importing into Slate - whether importing Person records, Dataset records, or system objects associated with those records - an attempt to prevent duplicates will be made.
For person and dataset records, as soon as a record matches a mapped item, Slate considers the record to be found and will stop evaluating subsequent matching criteria. As such, the order of matching criteria is important to note. The criteria and order are explained in greater detail in the related Matching Criteria - Persons and Matching Criteria - Datasets documentation.
For system objects (such as addresses, jobs, relationships, gifts, entities, etc.), in some cases the objects are updated based on the imported data - in others, data is removed from the process in order to avoid creating duplicate system objects. The exact behavior for each type of system object is explained in detail in the Matching Criteria - System Objects documentation.