Issues in extracting 050007_ForumAnalysis/workflows from Example Workflow

Hi Team,

I am new to KNIME and Text Mining. I have used most of the textmining example workflows to understand how each node works in text analytics. Thanks to theKNIME team!! 

I am trying to build a workflow to analyze open ended survey questions asked to new insurance customers. My objective of this study is to build a model to find out the on what type of topics did the customers expressed their opinion.  But unfortunately i am not sure how to do this. 

I was luck to stop a video presentation by Bernd Wiswedel on "Knowledge Extraction from a Web Forum using KNIME". I tried to download the Forum Analysis Classify post workflow in both the methods but none worked. The zip file from publicserver_workflow promts for a password while unzipping the file and couldnt copy the work seperately as well.

I have two requests:

1. Please share the forum analysis workflow to undersand the implimentation of nodes

2. Can you suggest me a way to identify topics and classify each responses into the various categories like price, claims, quotes, customer service, and ect.

 

I know its a huge ask and sorry for a long question. Thanks for all your support and i must agree KNIME forum, other resources and support provided by team is AWESOME !!

 

Thanks

sathish

Hi Sathish,

1.) you can download the example workflow easily from the example server. You can connect to the example server from within KNIME without the need to download it via browser and import it afterwards. See: http://www.knime.org/example-workflows for how to connect to the server. When you are connected navigate to the dir 050_Applications. Then drag and drop the dir 05007_ForumAnalysis into a dir in you local workspace in KNIME.

2.) To identify topics you can extract important words/terms. There are some methods available in KNIME, e.g. extracting terms based on their TF*IDF value or using the Keygraph Keyword extractor nodes. Usually this can be combined with filtering all terms but nouns. The second part of your question is imo a classification problem. You want to assign class labels to documents. See: http://tech.knime.org/document-classification-example for an example of how to classify text. Usually the text is filtered and preprocessed first, then transformed into in bag of words and then into document vectors, which is a numerical representation of documents. Based on these vectors classifier can be trained, e.g. decision trees or SVMs.

Cheers, Kilian

Hello everybody,

I am also new in this community and wandering to know about the forum classify posts workflow.

I am currently having problems though when copying the file from the EXAMPLES directory. After downloading approximately 30% of the workflow the download stops and the console shows me the following.

ERROR     WorkflowDownload                   Unable to download workflow: nested exception is: java.rmi.MarshalException: CORBA MARSHAL 1398079699 Maybe; nested exception is: 
    org.omg.CORBA.MARSHAL:   vmcid: SUN  minor code: 211 completed: Maybe

Could you tell me how to overcome this error please? I have the 2.11.1 version and 64 bit processor.

Thanks a lot!

Stefano

 

Hi Stefano,

the download works fine for me. We rebooted the examples server the last days, maybe it was because of that. Could you please try again and send a stack trace in case of failure.

Thank you.

Cheers, Kilian