Text mining and content categorization

Can one control clus_desc in TextMiner cluster analysis?

Reply
Esteemed Advisor
Posts: 7,056

Can one control clus_desc in TextMiner cluster analysis?

Actually, I have a couple of related questions.  When ones does a TM cluster analysis, a file labeled textcluster_clusters.sas7bdat appears in the work directory and contains a field labeled clus_desc that contains a number of strings and + symbols.

Question 1: Do the strings in that field represent descriptors of the clusters?

Question 2: Do the + symbols represent parents or, if not, what do they represent?

Question 3: Can one control the contents of that field, e.g., to exclude particular strings?

TIA,

Art

SAS Employee
Posts: 7

Re: Can one control clus_desc in TextMiner cluster analysis?

Hi Author,

To answer your questions:

Answer 1: Yes, the strings represent the cluster descriptors

Answer 2: The plus symbols represent the parent terms

Answer 3: No, at least not explicitly via editing the raw sas7bdat file.  However, the contents of the fields can be adjusted by adding to the proceeding Text Parsing node's stop list and then re-run the workflow.

Thank you,

Michael Wallis

Ask a Question
Discussion stats
  • 1 reply
  • 290 views
  • 0 likes
  • 2 in conversation