A spatially explicit reinforcement learning model for geographic knowledge graph summarization

Bo Yan*, Krzysztof Janowicz, Gengchen Mai, Rui Zhu

*Corresponding author for this work

Research output: Contribution to journalArticle (Academic Journal)peer-review

28 Citations (Scopus)

Abstract

Web-scale knowledge graphs such as the global Linked Data cloud consist of billions of individual statements about millions of entities. In recent years, this has fueled the interest in knowledge graph summarization techniques that compute representative subgraphs for a given collection of nodes. In addition, many of the most densely connected entities in knowledge graphs are places and regions, often characterized by thousands of incoming and outgoing relationships to other places, actors, events, and objects. In this article, we propose a novel summarization method that incorporates spatially explicit components into a reinforcement learning framework in order to help summarize geographic knowledge graphs, a topic that has not been considered in previous work. Our model considers the intrinsic graph structure as well as the extrinsic information to gain a more comprehensive and holistic view of the summarization task. By collecting a standard data set and evaluating our proposed models, we demonstrate that the spatially explicit model yields better results than non-spatial models, thereby demonstrating that spatial is indeed special as far as summarization is concerned.

Original languageEnglish
Pages (from-to)620-640
Number of pages21
JournalTransactions in GIS
Volume23
Issue number3
DOIs
Publication statusPublished - Jun 2019

Bibliographical note

Publisher Copyright:
© 2019 John Wiley & Sons Ltd

Fingerprint

Dive into the research topics of 'A spatially explicit reinforcement learning model for geographic knowledge graph summarization'. Together they form a unique fingerprint.

Cite this