The Noun Compound Synonym Substitution in Books (NCSSB) datasets contain in-context instances of potentially idiomatic English noun compounds, obtained by substituting idioms for synonyms occurring in public domain books forming part of the Project Gutenberg corpus.