"Text Mining using Rapid Miner"

prakash_sridhar

Hi,

I'm working on building a Text Mining Engine for a Consumer Research Study using Rapid Miner. 

Now, what I'm looking to do is to identify common words used by people to express their experience with regards to some products.

The algorithm should typically be able to strip off the sentence all articles, prepositions etc and just list out the words. Post this I want to run a count to identify number of times people utilize specific words to describe consumer experience. I would actually create this for multiple consumer experience scenarios.

Does Rapid Miner have this capability? If Yes, Can you point me to the way I need to approach this. The input is typically a consumer research survey which is accompanied by a few open ended questions where the cnsumer is allowed to leave his comments on his experience with the product.

Thanks in advance for your help.

Prakash

IngoRM

Hi Prakash,

yes, this is actually possible with RapidMiner. The easiest way is to identify feature weights for the different experiences (found by classification, clustering, whatever...). However, those results are often not sufficient and lots of more or less complicated postprocessing has to be done afterwards. This is actually something, which has to be adapted for each new project and even for us it takes some time in project work to come up with the most appropriate solutions for this.

So yes, it is basically possible with RapidMiner. But no, it is not possible to give you a recipe and a complete process for this but you have to figure it out yourself (or we could make a project together  )

Cheers,
Ingo