Partial Knowledge In Embeddings

TL;DR

This paper explores methods for representing partial domain knowledge in embeddings, introducing ensemble and aggregate embedding techniques to enhance expressiveness and encode incomplete information.

Contribution

It introduces ensemble and aggregate embedding methods specifically designed to encode partial knowledge within embedding spaces.

Findings

Ensemble and aggregate embeddings effectively encode partial knowledge.

These methods improve the expressiveness of knowledge representations.

The techniques demonstrate potential for more flexible knowledge systems.

Abstract

Representing domain knowledge is crucial for any task. There has been a wide range of techniques developed to represent this knowledge, from older logic based approaches to the more recent deep learning based techniques (i.e. embeddings). In this paper, we discuss some of these methods, focusing on the representational expressiveness tradeoffs that are often made. In particular, we focus on the the ability of various techniques to encode `partial knowledge' - a key component of successful knowledge systems. We introduce and describe the concepts of `ensembles of embeddings' and `aggregate embeddings' and demonstrate how they allow for partial knowledge.