count unselected items

[ulf1]

Reason

• unselected items $C_i$, $C_j$, $C_k$ => $N_{ij}+=1$, $N_{ik}+=1$, $N_{jk}+=1$ (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 #10