The values are sets or percentages or a number of other things with the requirement that, as NL is an extension of (among other things) Robinson's non-standard analysis and as we're dealing with a triplet, the summation of the components must lie in the "interval" from -0 to 3+. How the values for each component are determined depends upon application. In particular, you can just pick the values you think best and use NL as a classifier or to update the state of an inference system based upon new information. You may find the attached helpful (or remedial, or perhaps you're familiar with them already).

You can use the following formula:

s=(T-F)/(1+I+|T+F-1|)

See my papers on Researchgate site related to Neutrosophic theory.

Please what is this symbol means |T+F-1|?

It represents de absolute value, namely

max(T+F-1,1-T-F)