product master data (wrong data) - pattern matching

KNIME Analytics Platform
1

Hi everyone,

I need your help to correct the product master data that has been entered with errors over the years, for example,
bolt (correct) bolt… (wrong)
hammer (correct) hamer (wrong)

How can I structure a workflow that recognizes these errors (for example, special characters or extra spaces?

Thanks everyone

2

You can use the Column Expressions or String Manipulation nodes to do a Regex replace formula that only retained alphanumeric numbers as a starting point [^a-zA-Z0-9 ]

Capitalization can also be addressed by formula in those nodes.

After that you can clean up the spellings various ways. If statements, string matcher / similarity search, dictionary rule node, custom components.

1 Like
3

For another approach, you might try Fuzzy matching. This workflow might be helpful if you already have a dictionary of correct names:

2 Likes