Sentiment Analysis Error

Hi everyone,

I recently downloaded Knime to work on a project to analyse some user reviews.

The first step was to analyse the reviews and break down whether they were overall positive or negative based on their word contents.

I run this workflow and it seems fine until towards the end where a number of errors appear.

The original CSV of data is just a column where each row contains text and an adjacent cell with a unique reference code.

I have made sure i formatted the cells so there are no line breaks.

Could someone please assist and explain some of these errors and how to work around them please?

WARN GroupBy 0:758:310 No grouping column included. Aggregate complete table.
WARN GroupBy 0:758:310 No grouping column included. Aggregate complete table.
WARN GroupBy 0:758:310 No grouping column included. Aggregate complete table.

Does this mean i needed to add a column to help it group?

Kind Regards,

Krish

Hey @Krishan,

the node dialog of the GroupBy node has four tabs. The Groups tab and three Aggregation tabs. In the Groups tab you can select the column(s) which you want to group. In the Aggregation tabs you can select aggregation columns and specify the type of aggregation.

I can see that you’ve got a GroupBy node in your workflow to count the number of words for each document. To do so, select the Document column as column to group and the TF column as aggregation column (in the Manual Aggregation tab). As aggregation type you can use count for number of unique words or sum for number of total words. Afterwards you should get a table with one row per documents and a word count column.
If you want to count number of positive and negative words in each document, you have to select the Document column and the sentiment column, so that you get a table with two rows per document and a count for positive and negative words.

I hope this helps.

Kind regards,

Julian

1 Like

Hi Julian,

I really appreciate your response!

I am very new to the software so am trying to make sure I am following your instructions:

I still appear to be getting some error messages.

Not really sure how to configure these:

Kind Regards,

Krish

Hey @Krishnan,

regarding the first screenshot: You don’t need to add the Term column, if you want to count the number of words per document. It will group equal documents and count the number of rows for each unique document (which is the number of unique words per document).

The second screenshot looks alright. You don’t need to enter anything in the Type Based Aggregation tab (third screenshot) in this case. It has basically the same options as the Manual Aggregation tab, but instead of selecting columns, you can select a Type of a column. For example, if you select Number (Integer), all integer columns will be aggregated based on the specified aggregation type (e.g. sum).

The warnings below don’t seem to come from the selected GroupBy node but from succeeding nodes. I guess, there are some nodes in the Calculate Score meta node that aren’t configured correctly. It seems that you used the Sum aggregation type before. The created new column is then called Sum(TF). However since you use count now, the new column is then called Count(TF) and the succeeding Column Rename node can’t find the column anymore, so you have to reconfigure the node.
If you don’t want that the GroupBy node renames the aggregation columns, you can change the name handling at the bottom of the node dialog.

Cheers,

Julian

Cheers Julian,

I think the next section of the workflow is another can of worms…

Not really sure how to configure any of this:

Krish

This node in particular is very confusing:

Krish

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.