Front Matter

Lei Chen, Fatma ÖzcanPVLDB 2018

Proceedings of the VLDB Endowment logo

Abstract

Volume 12, Number 4 of the Proceedings of the VLDB Endowment (PVLDB), published in December 2018, presents ten peer-reviewed papers curated for the 45th International Conference on Very Large Data Bases (VLDB 2019). This issue highlights a deepening integration of machine learning with core database systems, alongside significant advancements in graph analytics, stream processing, and data marketplaces.

The PVLDB Publishing Model

The "Front Matter" of this volume addresses the ongoing challenge of managing and disseminating high-quality research within the rapidly evolving data management field. The PVLDB model was pioneered to resolve the traditional "conference vs. journal" dilemma in computer science.

While traditional conferences often rely on a single high-stakes deadline and journals can suffer from multi-year publication lags, PVLDB employs a hybrid approach. It features monthly submission deadlines, a rapid six-week review cycle, and a journal-style revision process. This ensures that the research presented at the annual VLDB conference—held in Los Angeles in 2019 for this specific volume—is both timely and rigorously vetted.

VLDB 2019 Conference Banner

Editorial Methodology

The "Letter from the Program Chairs" outlines a rigorous editorial process designed to maintain high standards:

  • Expert Peer Review: A Review Board of 166 expert researchers, overseen by 17 Associate Editors, manages the evaluation process.
  • Revision Cycle: Unlike standard conference models, PVLDB allows for a three-month revision cycle. This enables authors to substantively improve their work based on reviewer feedback before final acceptance.
  • Thematic Focus: For Vol. 12, No. 4, the editors prioritized papers that bridge the gap between traditional data structures and modern computational needs, particularly within distributed environments.

Key Research Insights

The papers included in this issue reveal several critical trends in the current state of database research:

1. Machine Learning Integration

A notable shift is occurring where Machine Learning (ML) is moving "inside" the database engine. Rather than treating ML as an external application, research such as ColumnML and Helix explores integrating ML directly into column-store transformations and iterative optimization loops.

2. Scalable Graph Analytics

Research is advancing beyond simple graph processing toward complex partitioning policies for large-scale distributed platforms. This issue includes work on microbenchmark-based graph database evaluation to address the need for standardized performance metrics in this space.

3. Consistency and Integrity

The collection tackles the perennial challenge of maintaining data integrity in weakly consistent replicated databases, specifically through work on "Invariant-preserving Applications."

4. Human-in-the-Loop and Data Markets

There is a continued focus on data quality, utilizing "Oracles" (human or automated) to clean noisy crowdsourced data. Additionally, the issue explores the economics of data, evaluating correlation analysis and cost-efficient acquisition within online data marketplaces.

Conclusion

The ten research papers accepted for this issue cover a broad spectrum of topics, including graph systems, holistic ML optimization, relational interfaces for data explanation (the DIFF system), and stream frequency handling. As noted by the Program Chairs, these contributions are intended to inspire novel systems research and further exploration at the intersection of machine learning and data management. The PVLDB model continues to prove its value by balancing the speed of conference publication with the depth of journal-style review.

Summary updated

BibTeX
@article{chen2018front,
  author = {Lei Chen and Fatma Özcan},
  title = {{Front Matter}},
  journal = {Proc. VLDB Endow.},
  year = {2018},
}
Figure