Skip to main content
European Commission logo print header

Resource Bounded Graph Query Answering

Periodic Reporting for period 4 - GRACE (Resource Bounded Graph Query Answering)

Reporting period: 2020-05-01 to 2021-10-31

Graphs are a ubiquitous model to represent objects and their relations, such as social networks, transportation networks, telecommunication networks, the World Wide Web, biological networks, transportation systems and hidden terrorist networks. The need for querying big graphs is evident in search, social media marketing, knowledge discovery, route planning and fraud detection, among other things.

Querying big graphs has introduced a number of challenges, from fundamental problems to practical techniques. It demands a departure from the traditional query evaluation paradigm. Vital to any systems for querying big graphs are a number of technical questions. What graph patterns should we support to query big graphs? How can we identify associations of entities in real-life graphs? What queries are BD-tractable, i.e. “tractable” on big graphs? What queries are parallel scalable, i.e. guarantee to reduce running time when more resources (processors) are used? How can we make queries BD-tractable? What parallel model should we use to query big graphs? When exact query answers are beyond reach in big graphs under constrained resources, can we compute approximate answers with accuracy guarantees? Can we extend the techniques to big relations, beyond graph queries?

In response to the challenges, this project aims to extend the conventional query paradigm, establish methodological and algorithmic foundations, and provide effective resource-constrained techniques for efficiently querying big real-life graphs and relations.
The project has produced an array of results, as indicated by the following highlights.

(1) Awards. The research has received the Best Paper Award for SIGMOD 2017 (the premier international database systems conference), the Best Demo Award for VLDB 2017 (the leading international all-round database conference), and ACM SIGMOD Research Highlight Award. The work also received the Royal Society Wolfson Research Merit Award 2018.

(2) Publications. The project has produced 51 papers, including TODS (7), SIGMOD (14), PODS (2), PVLDB (12), ICDE (1, a leading database systems conference), TKDE (3, a major database journal), and 16 invited papers (including 4 as the best of SIGMOD 2017-2018/PODS 2016-2017).

(3) Systems. As proof of concept, the project has developed two prototype systems, GRAPE/GraphScope and BEAS, for querying big graphs and relations, respectively.

Below we summarize publications generated from each work package (WP) of the project. Here [X] refers to paper X on the list of Publications.

WP1: Graph Patterns. WP1 published 10 papers, including SIGMOD (3), PODS (1), PVLDB (3), TODS (2) and 1 invited paper besides [29]. Paper [29] is invited as the Best of PODS 17 [13].

(1) Pattern languages [6, 36].

(2) Association rules on graphs [10, 42].

(3) Dependency languages [7, 8, 13, 29, 30].

(4) Unifying logic reasoning and machine learning [38, 42].

WP2: Theory for Querying Big Data. WP2 produced 10 papers, including SIGMOD (3), PODS (1), PVLDB (1), ICDE (1), TODS (2) and TKDE (1), and another invited paper besides [44]. Paper [44] is invited as Best of PODS 2017 [3].

(1) A theory of bounded evaluability [3, 4, 21].

(2) Incrementalization [14, 41, 47].

(3) Parallel scalability [39].

(4) An axiom system and parallel reasoning [13, 18, 25, 39, 44].

WP3: Effective Methods. WP3 published 16 papers, including SIGMOD (6), PVLDB (3), TODS (3), TKDE (1), and another 3 invited papers besides [27, 28, 43]. Paper [15] received the Best Paper Award for SIGMOD 2017, paper [43] is invited as Best of SIGMOD 2018 [18]. Paper [27] received the SIGMOD Research Highlight Award.

(1) Resource-bounded query answering [5, 33, 46].

(2) Parallel graph programming model [15, 23, 26, 27, 28].

(3) Parallel computation model [20, 43].

(4) Parallel scalable algorithms [19, 31, 32, 37, 45, 51].

WP4. Approximate query answering. WP4 produced 6 papers, including SIGMOD (1), PVLDB (1), and another 4 invited papers.

(1) Data-driven approximation [17].

(2) A new query paradigm [12, 24, 34, 35, 40].

The work on bounded evaluation (WP3) and data-driven approximation (WP4) received the Royal Society Wolfson Research Merit Award (2018).

WP5: Systems. WP5 published 3 papers, all in PVLDB. Paper [22] received the Best Demo Award for VLDB 2017.

(1) GRAPE: Prototype system [22].

(2) GraphScope ( Industry-scale system [49, 50].

WP6: Beyond graphs. WP6 generated 6 papers, including SIGMOD (2), PVLDB (1), TODS (1), TKDE (1), and another 2 invited paper.

(1) BEAS: A system for Bounded EvAliuation of Sql [16].

(2) Applications: Querying and cleaning big relations [1, 2, 5, 11, 48].
The project has delivered results on each of the six work packages proposed, from theory to techniques and systems.