2
$\begingroup$

I'm using Google API to categorize and predict problems with laptops (Hardware, Software, Network, Customer satisfaction, No reason). I inserted my training data and categories are very well evaluated and predicted. Training data only uses 5 main categories as it comes from free text and being specific requires some manual work. Now I want to go a level deeper and tag these categories: Example. Laptop was returned because Display was cracked: Hardware category was defined already, but I have a dictionary where I can also look for the word "Display". Example:

Hardware:
   Display:
      failure: True
   Info:
      Physical_damage: Display cracked

Unable to install Skype:

Software:
   Applications:
      Skype:
       failure: True
   Info:
      Description: Unable to upgrade    

I want to categorize first using main categories and after main category is chosen, look into my text for possible options from a bag of words (Example: {"display": ["screen", "display", "monitor"] }, {"applications", "skype","office","adobe","spotify"}

Should I run these in 2 stages? Any recommended pattern or solution?

$\endgroup$
2
  • $\begingroup$ Somehow I am not able to get a clear understanding of what you are trying to achieve. I mean can you specify what data does Google API provide you and what results you get after first iteration of clustering and so on. $\endgroup$ Commented Feb 2, 2016 at 18:47
  • $\begingroup$ I edited the question $\endgroup$
    – gogasca
    Commented Feb 2, 2016 at 19:18

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.