Abstract
This paper provides a comprehensive technological classification and analysis of open data research within the scientific community. By examining 893 peer-reviewed articles published between 2006 and 2019, the authors explore six key facets: publication venue, impact, subject domain, life-cycle phases, and research type. The findings suggest that while the initial "hype" surrounding open data is stabilizing, the field is reaching maturity. Key trends include a significant focus on Semantic Web technologies and software systems for linked open data, alongside a notable research gap in early-stage data life-cycle phases like curation and harmonization.
Introduction: The Technological Gap in Open Data
While open data has been a prominent topic for nearly two decades, academic surveys have historically prioritized its social, political, or legal implications, such as government transparency and policy. There has been a distinct lack of structured overviews focusing on the technological infrastructure and methodologies. This paper addresses this gap by mapping the technical landscape: the tools being developed, the active technical domains (e.g., IoT, Geospatial), and the existing voids in the data life-cycle. By answering these questions, the study provides a roadmap for researchers to identify well-trodden paths versus "blue ocean" opportunities for innovation.
Methodology
The authors conducted a Systematic Mapping Study (SMS), a rigorous secondary research method typically utilized in software engineering to provide a broad overview of a research area.
As illustrated in the process diagram above, the methodology followed four primary stages:
- Definition of Research Questions: Establishing six RQs to guide the analysis of venues, impact, domains, topics, life-cycle phases, and research types.
- Search and Screening: Querying major databases (ACM, IEEE, Springer, IOS Press, and ScienceDirect), resulting in a corpus of 893 relevant papers.
- Classification Scheme Definition: Categorizing papers across six "facets," including technical topics (e.g., Semantic Web, Security) and life-cycle phases (from creation to exploitation).
- Mapping and Analysis: Synthesizing the data to identify trends and gaps. The raw classification data has been made available in a Zenodo open-access repository.
Key Insights and Results
The analysis reveals several critical trends in the evolution of open data technology:
Maturity Over Hype
Research interest peaked between 2016 and 2018, followed by a slight decrease in 2019. Rather than a decline in relevance, the authors interpret this as a sign of field maturity. The "hype cycle" has concluded, and the discipline is consolidating into a stable, productive research area.
Technological Drivers
- Semantic Web and Linked Data: Accounting for 22% of the corpus, Semantic Web technologies are the primary engine of open data research. This emphasizes that the core challenge has shifted from simple availability to machine-readability and interlinking.
- The Role of Infomediaries: Approximately 43% of papers focus on the "Infomediaries" domain—developers and organizations that transform raw data into value-added applications.
- Life-cycle Imbalance: There is a heavy "Exploitation" bias (40% of research). Conversely, early-stage phases like Curation (3%) and Harmonization (1%) are severely under-represented, leading to potential "garbage in, garbage out" issues in downstream applications.
Quantitative Findings
- Dominant Topics: Semantic Web (22%), Software Engineering (19%), and Government (18%) lead the technical discourse.
- Research Type: 55% of papers are "Solution Proposals," while only 8% focus on "Evaluation Research," suggesting a field that prioritizes building new tools over rigorously testing existing ones.
- Geographic Scope: 67% of studies have an international impact, indicating that technical solutions are generally not restricted by specific national regulations.
Future Directions and Limitations
The mapping study identifies several critical gaps in the current research landscape:
- Data Quality and Curation: There is an urgent need for automated, scalable methods to clean and standardize data before it reaches the exploitation phase.
- Security and Privacy: Despite the sensitivity of domains like Health and IoT, security accounted for only 2% of the literature. This represents a significant vulnerability.
- Entrepreneurship: There is a lack of research on how technology can specifically lower barriers for open-data-driven startups.
- Automation of Research: The authors suggest using AI and Natural Language Processing (NLP) to automate future systematic mapping efforts, reducing the labor-intensive nature of manual classification.
Summary updated