Error in Chi-square Keyword Extractor result

Hi all,

In attached knime workflow I am simply reading data about movie reviews. If I set document title as “empty string” or “row ID” results are changing. Odd thing about this, anyways I am filtering title to empty string even if I set it as row ID with Number Filter.

C4_E2.knwf (22.2 KB)

Why and how are results changing? I think it should not change.

Thank you for discussion…

Hi @asenkron,

that sounds interesting…
Would it be possible to share also the data that you use in the workflow?
I’m not 100% sure that i fully understand what are the 2 alternative scenarios. Could you maybe add two parallel branches in a workflow doing slightly different processing and highlight what is different between the branches? This would help to start debugging faster.

Cheers,
Mischa

Best regards,
Mischa

1 Like

Hi,

did you manage to resolve the issue? If not, could you please provide further information? This will help a lot to debug what is going wrong.

Best,
Mischa

Hi,

I have employed parallel workflows with nuance. They resulted exactly the same in that case. But in another knime workflow results differ dramatically.

I attached data inside of C4_E2.knwf documents.

C4_E2.knwf (25.8 KB)
<a class=“attachment” C4_E2_parallel.knwf (42.5 KB)

Best,

Edo

Hi @asenkron -

When I look at your workflows, I’m not seeing any difference in the filtered Chi-square results based on whether the title is set to empty string or rowID.

I did notice that in the C4_E2.knwf workflow, the first Row Filter node excludes a “POS” value for the Category field, while in the newer workflow, the “POS” values is included.

Does this explain the discrepancy for you?

2 Likes

It makes sense. Sorry for inconvenience. it is my fault :frowning:

1 Like