Circular Reference

KNIME Analytics Platform
1

Hi all,

I would like to know if there is a formula that can detect a circular reference.

Take the example below:

Employee Manager
Mickey Mouse Donald Duck
Donald Duck Mickey Mouse
Goofy Dog Donald Duck

Mickey Mouse and Donald Duck refer to each other as manager. Is there any way Knime can detect these cases and put them on a seperate list? The case of Goofy Dog is correct, and should be ignored.

Thanks

Kind Regards

 

 

 

2

Hi,

Provided that the circular reference can only be on one level (Employee to immediate Manager) I was thinking of something like this:

1) From the initial data table, you can use twice a String Manipulation node to create a new column as Manager|Employee and another one as Employee|Manager. If there is a circular reference, the first column will contain A|B and B|A, the second column will contain B|A and A|B

2) Join the table with itself (Inner Join) using the two added columns as Joining Columns. This will filter out all lines that do not have a match between the first and the second column.

3) You end up with a table containing only the circular references.

The only down-side is that each circular reference appears twice because A|B from the first column matches A|B from the second one (Manager to Employee circular reference) and B|A from the first column also matches B|A from the second one (Employee to Manager circular reference), but you get anyway your list of circular references.

Cheers,
Marco.

 

3

Hello Otter, hello Marco,

Attached you will find a solution that lists circular references of arbitrary lengths. It's not beautiful but it works.

Best,
Ferry

4