Solutions to "Just KNIME It!" Challenge 06 - Season 2

:mega: A new Just KNIME It! challenge has just been released! :partying_face:

This week we have our first hard challenge, focusing on sentiment analysis :sob: :grin: over tweets, and relying a lot on text processing :spiral_notepad: techniques. :boom: To make the task easier for those who do not know much about text processing, we are sharing two components that may come in handy and can be adapted by you for your solution. :blush:

Here is the challenge. Let’s use this thread to post our solutions to it, which should be uploaded to your public KNIME Hub spaces with tag JKISeason2-6 .

:sos: Need help with tags? To add tag JKISeason2-6 to your workflow, go to the description panel on the right in KNIME Analytics Platform, click the pencil to edit it, and you will see the option for adding tags right there. :slight_smile: Let us know if you have any problems!

3 Likes

Link goes to Challenge 5.

Hey @rfeigel it is working on my end. Maybe it was a cache problem? Can you try again and please let me know? Thanks!

I use Incognito Mode for the right webpage to appear.

Got it. Another problem. I can’t get either of the components you recommended. I’ve tried both drag and drop and download. Get this error message:

@rfeigel Hey, I tested it a few times and it’s working on my end, so I will forward your error to our support team and look into a solution for it. Thanks for pointing this out!

Hi all,
Here is my solution.
The XGB model with optimized parameters was more accurate than the decision tree model.

4 Likes

Hello everyone, this is my solution.The hyperparameter optimization is performed manually, and the regular parameters are added strongly.

Discussion: It may be necessary to pay attention to the recall rate of negative evaluations. I understand that airline customer service needs to respond to negative evaluations in a timely manner.

1 Like

Can columns other than “document” (like “negativereason”, “negativereason_confidence”, etc.) be used for prediction ? (or only document?)

Here is my solution (made by only “Document” column).

2 Likes

Hi all,

here my solution:

I compared four methods and the winner is…

6 Likes

@rfeigel Could you please send me:

  • Your KNIME logs (how to get logs is here)
  • KAP version you used

This may help us understand what’s going on better! Thanks!

Last few days of log attached. I’m using v. 4.7.2. This problem is affecting my installed components as well. See pic below. A component already in a workflow will run, but I can’t add a component to a new workflow.
2023-05-02 185028,102 ERROR Worke.txt (35.3 KB)

2023-05-04_18-53-27

Here’s the error I get when I double click a component. This one is the XAI View. The same component works fine if its already in a workflow.
java.lang.NullPointerException Cann.txt (7.7 KB)

Hi all,
This is my solution.

I created a workflow with the provided components.
Positive, negative and neutral were randomly selected 1,000 each to save time.
Apart from that, I did without a dictionary tagger for my study, referring to the education material [L4-TP].

4 Likes

I have learned from your solution, and I will also give it a try. This is an imbalanced dataset, and balanced sampling may be beneficial for learning. thank you

1 Like

Thanks for your comment. I am an ML beginner, so I don’t really know the importance of bias, but I’m glad I could help you.

1 Like

Thanks, @rfeigel! I’m in touch with our team and will get back to you as soon as possible with more help.

Hello KNIMErs, Here is my Solutions for “Just KNIME It!” Challenge 06 – Season 2

2 Likes

I didn’t diagnose the problem, but I fixed it. I did a clean reinstall and everything seems to be fine. I’m still curious about what caused this. Identifying the cause may help avoiding the problem in the future.

1 Like

Finally fixed my components problems. Here’s my solution.

1 Like