count unselected items

[ulf1]

Reason

• unselected items $C_i$, $C_j$, $Ck$ => $N{ij}+=1$, $N{ik}+=1$, $N{jk}+=1$ (It's a symmetric matrix $N{ji}=N{ij}$)
• it is possible that the unselected items are too similar according some (user-specifc) decision critieria
• we could use such count data ($N_{ij}$) as a proxy for similar items => compute similarity metric

TODO

• [ ] brainstorm what to analyse with such data
• [ ] code it

[ulf1]

Duplicate https://github.com/ulf1/bwsample/issues/10