Sentiment_Analysis_Training_With_Integrated_Deployment

KCH Workflows
1

Sentiment_Analysis_Training_With_Integrated_Deployment

This is a companion discussion topic for the original entry at https://kni.me/w/-q3w1nIJwiGHxGMT
2

Hello. I would like to see the workflow running but it does’t have the input data table Imdb.table. Is it possible to get it please? Thnaks

4

Hello @lsandinop,

this workflow is part of Codeless Deep Learning with KNIME book and you’ll find all input data in data folder one level up.

Br,
Ivan

1 Like
5

To analyse new reviews does the trained workflow need to be connected to the deployment workflow?

And if it does how is this done?

6

Hello @stockdaa,

and welcome to KNIME Community!

In above example training workflow creates new workflow called Sentiment Prediction (located under data → Created_workflows) which is then called from when new predictions are done using Call Workflow (Table Based) node. For more on Integrated Deployment I check out following guide:

https://docs.knime.com/latest/integrated_deployment_guide/index.html#introduction

and this blog post series:

Br,
Ivan

7

Thank you for responding. Unfortunately your response is very general, whereas my question is specific.

Are you able to give me more detailed help?

8

Hello @stockdaa,

indeed is general although should provide more clarity and set you on path to explore more about Integrated Deployment. However let me try to provide more details.

Sentiment_Analysis_Training_With_Integrated_Deployment → training workflow which creates prediction workflow
Sentiment_Analysis_Deployment_with_Call_Workflow → deployment workflow which calls prediciton workflow
Sentiment Prediction → prediction workflow

So, if when you say “trained workflow” you mean prediction workflow from above then answer is yes and they are connected via already mentioned Call Workflow node. If “trained workflow” from your question is training workflow then answer is no.

Hope this helps!

Br,
Ivan

9

Hello Ivan

Thank you for persevering.

I am new to Knime and I am not a data scientist. I am interested in exploring how useable Knime is for noncomputing researchers. I want to run the Sentiment Analysis Workflow to see how it works from a practicable viewpoint. I came across this case in a book called Codeless Deep Learning with Knime by Kathrin Melcher.

This case appears in the book in different chapters. The first developed the application the second showed how it could be deployed. I downloaded the workflows for each of these into Knime. The data for this case was stored separately as .h5 files.

What I would like help with is how to put these different parts together and run the application on some data to see how it works. At the moment I do not know how to do this. I had already read the documents that you sent me originally and I have a reasonable understanding of how this works from a theoretical point of view.

What I need is some practical guidance to run this case, hence my questions. I did not use the words “trained workflow” in my questions. I have not heard of a Call Workflow node. This was not mentioned in the book; just the integrated deployment workflow, nor the prediction workflow. How are these connected together?

Is there any way that I can get this practical help?

Best wishes

Alan

10

Hello @stockdaa,

Sry for late response. I understand you have been in touch with Scott about this so will stay away from it. Hope you get your answers.

Br,
Ivan