Design and Benchmarking of a Citation Graph DB Across Neo4j, ArangoDB, and MASS Graph DB Systems

Abstract

Academic collaboration, citation influence, and institutional research visibility are increasingly reflected through scholarly relationship networks. However, existing academic platforms remain largely profile-centered and do not provide an institution-focused, interactive, and queryable graph system for multi-hop exploration across authors, works, affiliations, and citations. This research investigates the design and benchmarking of a UWB citation graph, seeded from CSS faculty scholarly activities for practical evaluation, and examines how effectively different graph database systems support this richer graph model for practical scholarly exploration. To address this problem, this work designs and implements a scholarly citation and co-authorship graph pipeline that constructs a heterogeneous Author–Work–Citation–Affiliation graph using institutional seed data and OpenAlex-derived metadata. The resulting graph is intended to support practical use cases such as collaborator discovery and referee explo-ration for UWB CSS faculty. The system is evaluated across three graph databases—Neo4j, ArangoDB, and MASS Graph DB—and is benchmarked using LDBC-aligned workloads and metrics, including bulk ingestion throughput, query throughput, and multi-hop traversal latency. In addition to the institutional citation graph, the evaluation framework includes public benchmark datasets of different graph types, densities, and scales to enable broader cross-platform comparison. The results support the hypothesis that a heterogeneous scholarly citation graph can enable richer institution-centered analysis while maintaining practical performance. Compared with a single-relation citation graph, the UWB-CSS Citation Graph supports collaborator discovery, citation-based visibility analysis, affiliation-aware filtering, and preliminaryexternal-referee identification. Although its heterogeneous author–work–affiliation–citation structure introduces additional ingestion and traversal cost, its sparse topology helps offset part of the performance cost introduced by heterogeneous node and relationship types. Platform-specific results further show that Neo4j provides strong interactive query and traversal performance for the UWB-CSS graph, MASS Graph DB supports high-throughput in-memory bulk ingestion, and ArangoDB Cloud offers useful managed deployment and distributed scaling potential, especially for bulk ingestion and 3-hop traversal from 4 to 8 nodes. Overall, the proposed graph remains practical on suitable platforms while providing broader analytical value than structurally simpler citation-only graphs.