In this paper, an approach for semantic compositionbased on space projection operations over basic geometriclexical representations is proposed. Syntactic bi-grams arehere projected in the so called Support Subspace, aimed atemphasizing the semantic features shared by the compoundword. Empirical results are discussed over two tasks: a finegrainedsimilarity estimation task over syntactically typedword pairs and the Semantic Textual Similarity (STS) taskas defined by SemEval 2012. The generalization capabilities ofthe proposed compositional operators are also investigated ina cross-linguistic scenario, i.e. over English and Italian.

Semantics, Vectors, Syntactics, Humans, Correlation, Additives, Context, Semantic, Distributional Compositional Semantics

D. Croce, V. Storch, P. Annesi and R. Basili, "Distributional Compositional Semantics and Text Similarity," 2012 IEEE Sixth International Conference on Semantic Computing(ICSC), Palermo, Italy Italy, 2012, pp. 242-249.
doi:10.1109/ICSC.2012.63