When you run a mapping with a Cleanse transformation that specifies a dictionary, the transformation compares the input field data to the data in the dictionary. If the transformation finds a match between an input value and a dictionary value, the transformation performs an action that you define in the corresponding step in the cleanse asset.

At least one dictionary column must contain the set of standard or preferred values for your current data project. The other columns can contain alternative versions of the values. The column that contains the standard or preferred values is called the valid column.

In search and replace operations, a Cleanse transformation searches every column in the dictionary except the valid column that the asset specifies. If you want the search operation to include the valid column data, add a copy of the valid column to the dictionary. By default, the valid column is the first or left-most column in a dictionary. You can create a dictionary that contains two identical columns of data.

A dictionary might contain public terms, such as telephone area codes or address abbreviations. Or, a dictionary might contain values that are specific to an organization, such as employee codes or product codes. You can populate a dictionary with any combination of values that suits your project. The data in each column does not need to be formally correct.