Text Analysis project help

[KevinG42]

I am a phd student at the Univ. of Memphis working on a project that I think SAS Text Analytics can help with but I need a little guidance.

Here is what I am trying to do: I have a list of words, about 100. I need to find all the words in a large volume of text (like from a book) that relate to the word in my list. How many times they are related and ideally, whether the relationship is positive or negative.  For example:  let's say "blue" is in my reference list and there is a sentence in the text that is, "The car is blue." Then "car" is related to "blue" and the relationship is positive, If the sentence was, " The car is blue" then the relationship would be negative.

I am not sure if what I am asking is possible or if SAS is the right program to use.  I would imagine my project would be similar to brand affinity which I have seen SAS used for.  Any suggestions would be greatly appreciated.

[art297]

Why is "The car is blue." positive, but "The car is blue" negative?

[KevinG42]

Sorry, that was supposed to be:  If the sentence is, "The car is NOT blue," then the relationship is negative.

[KevinG42]

CORRECTION: The last line of the second paragraph was supposed to be:  If the sentence is, "The car is NOT blue," then the relationship is negative.

[art297]

Here is one article you might find useful: http://www.mwsug.org/proceedings/2011/saspres/MWSUG-2011-SP06.pdf

[KevinG42]

The article pointed me in the right direction.  I still need to figure out the programing details.  Sentiment Analysis is a separate module from SAS 9.3. Thanks