JMIR Publications

ABSTRACT

Background:

Knowledge graph embedding is an effective semantic representation method for entities and relations in knowledge graphs. Several translation-based algorithms, including TransE, TransH, TransR, TransD and TranSparse, have been proposed to learn effective embedding vectors from typical knowledge graphs in which the relations between head and tail entities are deterministic.

Objective:

However, in medical knowledge graphs, the relations between head and tail entities are inherently probabilistic. Such a difference introduced challenge in embedding medical knowledge graphs: how to learn the probability values of triplets into representation vectors.

Methods:

To address the challenge, enhancements are made to existing TransX (X=E/H/R/D/Sparse) algorithms, including the following: 1) constructing a mapping function between the score value and the probability and 2) introducing probability-based loss of triplets into the original margin-based loss function.

Results:

The proposed PrTransX algorithm is performed on a medical knowledge graph that was built from large-scale real-world electronic medical records (EMR) data. The embeddings are evaluated using link prediction task. Comparing with the corresponding TransX algorithm, the proposed PrTransX performs better in all evaluation indicators, achieving a higher Hit@10 and NDCG@10 and lower Mean Rank.

Conclusions:

We can conclude that the proposed PrTransX successfully incorporated the uncertainty of the knowledge triplets into the embedding vectors.