Giving generated Clusters a semantic meaning

Muhammed_Fatih_

Dear RM-Community, 

which possibilities does Rapid Miner offer with regard to the content based analysis of generated Clusters? My goal is to give the respective clusters a sense. Clusters could be generated, but it is currently unknown what the cluster groups hide in detail. 

Which methods or rather operators does RapidMiner offer? What did you use? What is your experience? 

I thank you in advance for your hints and feedback! 

Best regards

Fatih

MartinLiebig

Thats what i usually do:

https://towardsdatascience.com/understanding-clustering-cf0117148ef4

Best,

Martin

Muhammed_Fatih_

Hi Community, 
Hi @mschmitz, 

the consideration of Decision Trees in oder to describe detected Cluster patterns is definitely an interesting approach which I have also targeted. But there is for sure not one and only option to describe patterns. Hence, I am interested in the variability and different methodological ways of describing methods. 

I am generally interested in e.g. white box algorithms that ensure the prescriptive understanding of structural patterns. As an example, the literature mainly describes Topic Modeling and Concept Extraction approaches to describe the underlying content of Clusters. 

Are there other methodological possibilities that are offered by Rapid Miner? Has anyone had any experience with Topic Modeling and Concept Extraction? 

Best regards, 

Fatih  

MartinLiebig

Hi @Muhammed_Fatih_ ,

i think you want to check my favourite text mining operator: Extract Topcis (LDA).

Cheers,

Martin

Muhammed_Fatih_

Hi @mschmitz

many thanks for your feedback!

Do you know if RapidMiner also offers operators for the Log Likelihood Ratio (LLR), LSI, PLSA as extended Topic Modeling and Concept Extraction techniques? I couldn't find operators or rather extensions refering to those approaches.

Best regards, 

Fatih