A Meta Path Based Method for Entity Set Expansion in Knowledge Graph

Yuyan Zheng

doi:10.1109/TBDATA.2018.2805366

Return

A Meta Path Based Method for Entity Set Expansion in Knowledge Graph

2022-06-01

PRE

郑玉艳 (Yuyan Zheng)

Chuan Shi *

Xiaohuan Cao

Xiaoli Li (Xiaoli Li)

Bin Wu

DOI：10.1109/TBDATA.2018.2805366

Original

Original request for help

Save

Abstract

En 中文

Entity Set Expansion (ESE) is the problem that expands a small set of seed entities into a more complete set, entities of which have common traits. As a popular data mining task, ESE has been widely used in many applications, such as dictionary construction, query suggestion and new brand identification. Existing ESE methods mainly utilize text and Web information. That is, the intrinsic relation among entities is inferred from their occurrences in text or Web. With the surge of knowledge graph in recent years, it is possible to extend entities according to their occurrences in knowledge graph. In this paper, we consider the knowledge graph as a heterogeneous information network (HIN) that contains different types of objects and links, and propose a novel method, called MP_ESE, to extend entities in the HIN. The MP_ESE employs meta paths, a relation sequence connecting entities, in HIN to capture the implicit common traits of seed entities. In addition, an automatic meta path generation method, called SMPG, has been designed to exploit the potential relations among entities. Heuristic learning and PU learning methods are employed to learn the weights of extracted meta paths. With these generated and weighted meta paths, the MP_ESE can effectively extend entities. Comprehensive experiments on real datasets show the effectiveness and efficiency of MP_ESE.

Keywords:

Semantics

Task analysis

Resource description framework

Encyclopedias

Electronic publishing

Heterogeneous information network

entity set expansion

knowledge graph

meta path

PU learning

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.