Let's begin by spending a few minutes understanding how the object2vec algorithm works. It is a multi-purpose algorithm that can create lower dimensional embeddings of higher dimensional objects. This process is known as dimensionality reduction, most commonly implemented through a statistical procedure called Principal Component Analysis (PCA). However, Object2Vec uses neural networks to learn these embeddings.
Some of the common applications of these embeddings include customer segmentation and product search. In the case of customer segmentation, similar customers appear closer in the lower dimensional space. A customer can be defined through multiple attributes such as name, age, home address, and email address. With regards to product search, because product embeddings capture the semantics of the underlying data, any combination...