Vector embedding spaces are representational structures that can capture both the similarity relationship between items and various other semantic relationships. Current state-of-the-art embedding models can generate embedding vectors for individual words and longer strings of text, enabling the vector space to encode the similarity between entire documents of text. We explored whether semantic relationships besides similarity are represented in this space, focusing on the relationship between arguments and counterarguments as a specific example. While there was not a linear subspace that captured the semantic relationship between an argument and its counterargument, we found that a neural network with a single hidden layer could partially learn the transformation between an argument's embedding vector and the corresponding counterargument's embedding vector. This trained model generalized across three different datasets of arguments, suggesting that the partially learned transformation is applicable to arguments and counterarguments in general, not just tied to the semantic context of the model’s training dataset. This approach has practical applications in designing information retrieval systems for intelligent agents and, potentially, in models of cognition that use vector embedding spaces as a representational structure.

Authors:

Cherrie Chang: Vassar College; Josh de Leeuw: Vassar College