A Factorization Machine Framework for Testing Bigram Embeddings in Knowledgebase Completion
This work addresses knowledge base completion for AI systems, but it is incremental as it builds on existing embedding methods with a focus on bigram types.
The paper tackled the problem of knowledge base completion by exploring bigram embeddings for entity and relation pairs, achieving relative improvements over a compositional model on the fb15k237 dataset.
Embedding-based Knowledge Base Completion models have so far mostly combined distributed representations of individual entities or relations to compute truth scores of missing links. Facts can however also be represented using pairwise embeddings, i.e. embeddings for pairs of entities and relations. In this paper we explore such bigram embeddings with a flexible Factorization Machine model and several ablations from it. We investigate the relevance of various bigram types on the fb15k237 dataset and find relative improvements compared to a compositional model.