lin-1998-similarity-formula

IN premise — summaries/2026/08/24/wiki-Semantic_similarity-chunk-1.md

Created 2026-08-25T02:58:52+00:00

Lin's (1998) similarity measure is sim(A,B) = 2·IC(LCS) / (IC(A) + IC(B)), normalizing by the information content of both compared concepts.

Summary

This is a way to measure how alike two concepts are by looking at the information they share and then dividing by the total "specificity" of both concepts combined. The normalization step matters because it keeps the comparison fair: a broad, common concept won't look artificially similar to everything just because it shares a lot of generic information, while two very specific siblings will score high because their shared details are rare.