Linear Compositional Distributional Semantics and Structural Kernels

In this paper, we want to start the analysis of the models for compositional distributional semantics (CDS) with respect to the distributional similarity. We believe that this simple analysis of the properties of the similarity can help to better investigate new CDS models. We show that, looking at CDS models from this point of view, these models are strictly related with convolution kernels (Haussler, 1999), e.g.: tree kernels (Collins and Duffy, 2002). We will then examine how the distributed tree kernels (Zanzotto and Dell’Arciprete, 2012) are an interesting result to draw a stronger link between CDS models and convolution kernels.