MiMI employs deep merging techniques to identify entities that are the same across data sources. Once identity has been determined, the entries are merged and their provenance is recorded. Since data sets often contain contradictory or overlapping information, the deep-merging corrects for duplications, synonyms, and redundant interactions across identifiers. The deep-merging provenance model tracks where each piece of data came from and what processes have been performed upon it. The integrated data reside in a relational database.