"Develop word-tree for specific word in a document"

m_kaufmann

Dear all,

 

I'm relatively new to rapidminer, so my apologies for potentially stupid questions in advance.

 

I was wondering whether it is possible to generate a "word tree" with rapidminer (i.e. a visualisation of the context in which a particular word occurs)? My aim is to analyse the context of a word in a particular policy document and compare it (probably qualitatively) with other policy documents. If I cannot generate a word tree with rapidminer, is there a different possibility to analyse the 'context' in which a word occurs?

 

What I tried up until now:

• I analysed the similarity of the documents as such. However, I think they are too similar as they all deal with the same topic (climate adaptation). Thus, I am more interested in comparing how the documents frame specific words (e.g. vulnerability).
• I also tried using "create association rules" operator, but the process fails as "the process would need more than the maximum amount of available memory [...]". See attachment for xml file of the process, but I think that doesn't work and I need another alternative.
• I thought about using the 'decision tree' operator (word tree - decision tree, I thought it might be connected...), but I am wondering if it is applicable to text and if so, how could I 'tell' the operator only to focus on one specific word? 

 

I hope this makes any sense and you could help me further.

Thank you,

Maria

MartinLiebig

Hi,

i am not aware of a way to vizualize these trees except for the visualizations which are part of the association rules renderer.

 

In general i would suggest to think about word2vec as an option to numerically interpret the meaning of a word.

 

BR,

Martin