ELPKG: A High-Accuracy Link Prediction Approach for Knowledge Graph Completion

Jiangtao Ma,Guangwu Hu,Kai Ren,Yanjun Wang,Yaqiong Qiao,Huaiguang Wu,Yongzhong Huang,Arun Kumar Sangaiah,Chaoqin Zhang,Hongpo Zhang

doi:10.3390/sym11091096

Abstract

Link prediction in knowledge graph is the task of utilizing the existing relations to infer new relations so as to build a more complete knowledge graph. The inferred new relations plus original knowledge graph is the symmetry of completion knowledge graph. Previous research on link predication only focuses on path or semantic-based features, which can hardly have a full insight of features between entities and may result in a certain ratio of false inference results. To improve the accuracy of link predication, we propose a novel approach named Entity Link Prediction for Knowledge Graph (ELPKG), which can achieve a high accuracy on large-scale knowledge graphs while keeping desirable efficiency. ELPKG first combines path and semantic-based features together to represent the relationships between entities. Then it adopts a probabilistic soft logic-based reasoning method that effectively solves the problem of non-deterministic knowledge reasoning. Finally, the relation between entities is completed based on the entity link prediction algorithm. Extensive experiments on real dataset show that ELPKG outperforms baseline methods on hits@1, hits@10, and MRR.

Highlights

Knowledge graph (KG) refers to a network that contains specific topic related entities and the related information between entities
We propose a knowledge link prediction method named Entity Link Prediction for Knowledge Graph (ELPKG) to address the
We propose a knowledge link prediction method named ELPKG to address the problem of missing relations in the knowledge graph

Summary

Introduction

Knowledge graph (KG) refers to a network that contains specific topic related entities (i.e., nodes) and the related information (i.e., relation or predicate) between entities. A fact in a knowledge graph can be represented by the tuple relationship . Note that the entity in KG could be a specific object or an abstract concept, such as people, organization, dataset, and related documentation. As a promising artificial intelligence technique, KG has been widely adopted in many scenarios, e.g., question answering [1], recommendation system [2], co-reference resolution [3], information retrieval [4], and cross-language plagiarism detection [5]. The big issue in KG is that some link information is incomplete. In the Google knowledge vault project [6], 71% of the personal information lacks “place of birth”, while 75% lacks “nationality” information in Freebase [7]

Methods

Results

Conclusion