Building Memory with Concept Learning Capabilities from Large-scale Knowledge Base

TL;DR

This paper introduces a neural knowledge base embedding framework that models symbolic knowledge and its learning process, enabling reasoning and the ability to learn unseen entities from natural language descriptions.

Contribution

It proposes a novel framework that regularizes neural KB embeddings and incorporates concept learning from natural language descriptions for unseen entities.

Findings

Improved reasoning performance on KB tasks.

Capability to learn embeddings for unseen entities.

Regularization enhances embedding quality.

Abstract

We present a new perspective on neural knowledge base (KB) embeddings, from which we build a framework that can model symbolic knowledge in the KB together with its learning process. We show that this framework well regularizes previous neural KB embedding model for superior performance in reasoning tasks, while having the capabilities of dealing with unseen entities, that is, to learn their embeddings from natural language descriptions, which is very like human's behavior of learning semantic concepts.