BookmarkSubscribeRSS Feed
Calcite | Level 5



I know this is probably hopeless as the product is kinda deprecated, but I'm trying to retool some of our usage to try and get the best out of Categories and Classifier Concepts.


In other words, I'd like to use the accent insensitivity of Classifier Concepts  but have the added power of Categories.  I know I can create a reference/dependency between Classifier rules and Concepts with the [] syntax.


My problem is that I have some Concepts that are plugged in as include files and generated programmatically.  They can get quiet large and contain many different unique quantities. 


Just for example, it's like I have an include file for the Concept NFL, and inside I have match keys for all of the different teams, all of their known nicknames, etc.  Every match in the Concept file has the return value specified, so all synonyms/alternate phrasings come back to a consistent representation, e.g

New England Patriots,
Pats,New England Patriots
New York Jets,
Jets,__TGIF:{(OR,"football")}:New York Jets

I'd like to write a Category rule referencing the NFL concept, but tailor it to a specific return value.


Something like

(AND(OR,"Meadowlands","East Rutheford"),(OR,"[NFL]":"New York Jets"))

I tried approximating the xpath query syntax for a specific value.  It parses, but it doesn't match.


Is there any way to do this?  Or do I need to create a separate Concept uniquely for each thing in the include file?



SAS Employee

Thank you for your question! Yes, it appears that with the strategy you are using, you will get the results you are looking for only by separating your NFL list into the different sections you are interested in (probably by team based on my understanding of your use case).


Note: even though the product you are using is discontinued, the LITI syntax is alive and well at SAS. We use it for information extraction (concept and fact extraction) and it still integrates with the categorization syntax. There is a book on the  LITI syntax that may be useful to you published by SAS Press. Here is a link on Amazon (it is also available on O'Reilly and through your SAS representative):



Available on demand!

Missed SAS Innovate Las Vegas? Watch all the action for free! View the keynotes, general sessions and 22 breakouts on demand.


Register now!

How to choose a machine learning algorithm

Use this tutorial as a handy guide to weigh the pros and cons of these commonly used machine learning algorithms.

Find more tutorials on the SAS Users YouTube channel.

Discussion stats
  • 1 reply
  • 1 like
  • 2 in conversation