Disruption index (CD index)

A way to measure Scientific disruption from bibliometric data (Funk2016dynamic).

For a focal document $f$ It is defined as $$CD(f) = \frac{n_i(f) - n_j(f)}{n_i(f) + n_j(f) + n_k(f)},$$ where $n_i(f)$ is the number of documents that cite $f$ but do not cite any of $f$‘s references, $n_j(f)$ is the number of documents that cite $f$ and some of $f$‘s references, and $n_k(f)$ is the number of documents that cite some of $f$‘s references while not citing $f$.

A review: Leibel2024what

Kim2025uncovering reviews issues of the index and proposes a new, continuous version of the measure based on Graph embedding, Embedding disruptiveness measure

Implementations

• https://gist.github.com/skojaku/cee26755645b133a69d6630c79307cde
• https://github.com/yy/embedding-disruptiveness/blob/main/workflow/embedding_computation/scripts/Distance_disruption.py