Keywords

[alex boulton]

Dear all

I've been wondering about the keyword settings. Currently, with a target corpus of 200 files, if I set Range to 5 and Freq to 5, the keyword list might include an item that appears only 9 times in total (5 times in one file, and 1 time in each of 4 others); and another that could be near the top of the list if it appears 100 times in one file (and just 1 time in each of 4 others). This doesn't strike me as indicative of "aboutness". 

What I really want is to choose keywords that occur at least 5 times in each of 5 fies (so for a minimum total of 25 times). 

I'm sure I'm not the first person to have this concern, does anyone know of any research on the topic? or solutions, other than going down the list and removing such 'noise' manually?

Greetings from TaLC (Laurence is here too!)

Best

alex