Text analysis

• drew.georgopulos

  SSCarpal Tunnel

  Points: 4931

  More actions

  April 12, 2013 at 7:20 am

  #276442

  I built a "noise word" table, and parsed a free text field from customer care exchanges, so now i have a word with a primary key from the customer care table...about half a million records, so the most rudimentary thing to do with it was to count the frequencies, which was illuminating, but now, what is the next step to understand what is going on with those exchanges?

  For example, I can link back to the 3500 customer care records that mention Ambulance and learn the zip codes, ages and genders of the requestors, or the 6500 records that mention Membership cards, but I know zero about text analysis or its methodology, so whats to read about it, and what are my next steps to learn from that data store?

  One idea is that this probably lends itself to a cube because the text, by dint of being parsed, is multidimensional, that is, some words are positive, some are negative and some neutral (i filtered everything but nouns...if that was an error then i could rerun the parsing to include verbs and pronons too, but as it is i think i have a pretty rich field to investigate becase the pk links the words to their originating records, and the frequency liinks them to their repetitions, if you see what i mean

  thanks a lot

  drew