Semantic projection: recovering human knowledge of multiple, distinct object features from word embeddings

Understand

The words of a language reflect the structure of the human mind, allowing us to transmit thoughts between individuals.

• However, language can represent only a subset of our rich and detailed cognitive architecture.
• Here, we ask what kinds of common knowledge (semantic memory) are captured by word meanings (lexical semantics).
• We examine a prominent computational model that represents words as vectors in a multidimensional space, such that proximity between word-vectors approximates semantic relatedness.