GQBE: Querying knowledge graphs by example entity tuples

We present GQBE, a system that presents a simple and intuitive mechanism to query large knowledge graphs. Answers to tasks such as “list university professors who have designed some programming languages and also won an award in Computer Science” are best found in knowledge graphs that record entities and their relationships. Real-world knowledge graphs are difficult to use due to their sheer size and complexity and the challenging task of writing complex structured graph queries. Toward better usability of query systems over knowledge graphs, GQBE allows users to query knowledge graphs by example entity tuples without writing complex queries. In this demo we present: 1) a detailed description of the various features and user-friendly GUI of GQBE, 2) a brief description of the system architecture, and 3) a demonstration scenario that we intend to show the audience.

I. Introduction

Consider the scenario where a computer historian is interested in preparing an article on university professors who have designed a programming language and also won an award in Computer Science. If the historian only knows of a few professor-university-award triples, to start writing the article she must have a more comprehensive list of such triples. In general, users are interested in finding entities of various types (e.g., persons, products, organizations) that are related in certain ways. Tasks like the above one are becoming increasingly common in several applications, including search, recommendation systems, and business intelligence.