TL;DR
This paper introduces a foundational vocabulary and ontology for modeling insect natural history data, addressing the significant challenge of integrating heterogeneous and scattered information. By developing a structured approach, the authors aim to enable more robust data aggregation and analysis for ecological and evolutionary research on insects.
What Problem Does This Paper Address?
Insects, with their immense diversity, offer unparalleled opportunities for ecological and evolutionary research. However, harnessing this potential is severely hampered by the fragmented and inconsistent nature of insect natural history data. This data is scattered across specimen labels, obscure publications, field notes, and various databases, making it difficult to access, integrate, and analyze. The heterogeneity in data collection methods, observational detail, and terminology further complicates efforts to build comprehensive information systems. This paper tackles this challenge by proposing a standardized vocabulary and ontology to bring order to this complex data landscape.
Methods
The authors' approach involved a multi-stage process initiated at a workshop bringing together entomologists, computer scientists, and data modelers.
- Assembly of Example Data: Five small groups were formed, each focusing on one of the five mega-diverse insect orders (Hemiptera, Coleoptera, Diptera, Lepidoptera, Hymenoptera). They compiled concise datasets of natural history information typically found on specimen labels, drawing from literature, online databases (like iNaturalist and GloBI), and digitized specimen labels (from iDigBio). This process helped delimit the scope of the new vocabulary and ontology.
- Analysis of Example Data and Ontology Scoping: The assembled example data was analyzed to identify high-level conceptual areas required for a comprehensive ontology. This involved summarizing the types of biological information and observational details present.
- High-Level Ontology Design and Concept Identification: Based on the analysis, ten top-level conceptual areas were identified. The authors then prioritized three critical areas for initial development: "Observations and observing processes," "Relationships and interactions," and "Positional (spatial) information." They began sketching out data models and ontology design patterns for these areas.
- Identifying Use Cases and Authoring Ontology Competency Questions (OCQs): The final task involved defining potential users and their use cases for the ontology. This led to the development of OCQs, which are specific queries that the ontology should be able to answer, thereby testing its utility and guiding its development.
Datasets
The paper introduces a newly compiled dataset of 189 insect natural history observations. This dataset, drawn from specimen labels, literature, and online databases, covers a wide range of concepts and observation types for five major insect orders. The dataset is made available as supplementary material in PDF and CSV formats, and also in a public GitLab repository in various other formats (HTML, Markdown, SQLite). This dataset serves as a benchmark for developing and testing the proposed vocabulary and ontology.
Key Insights and Ideas
- The Power of Ontologies for Data Integration: The paper strongly emphasizes that ontologies are crucial for automating complex data integration and querying tasks, especially when dealing with large, heterogeneous datasets. They provide computable semantics that allow computers to infer relationships and answer complex questions that would be impossible for humans to manage manually.
- Prioritizing Core Concepts: The identification of "Observations and observing processes," "Relationships and interactions," and "Positional (spatial) information" as critical initial areas highlights the fundamental nature of these concepts for modeling natural history data. These areas form the backbone for understanding what was observed, how organisms interact, and where these observations occurred.
- Addressing the "Interaction Metadata" Challenge: A significant data modeling challenge identified is accurately capturing metadata about interactions between organisms. This is a common and important type of natural history data, but current technologies like RDF struggle to express this information efficiently and without complications.
- The Importance of Standardized Vocabularies: The paper draws a parallel to the success of the "Darwin Core" vocabulary for species occurrence data, suggesting that a standardized vocabulary for insect natural history data can similarly revolutionize large-scale biodiversity science.
Key Results
- Ten High-Level Conceptual Areas Identified: The research identified ten essential conceptual areas for a comprehensive insect natural history ontology, including observations, relationships, behaviors, ontogeny, organism products, habitat, spatial information, weather, collecting methods, and curation.
- Prioritization of Three Key Areas: "Observations and observing processes," "Relationships and interactions," and "Positional (spatial) information" were prioritized for initial ontology development due to their foundational importance.
- Development of an Example Dataset: A dataset of 189 insect natural history observations was compiled and made publicly available, serving as a valuable resource for the community.
- Creation of Use Cases and Competency Questions: A comprehensive set of use cases and ontology competency questions (OCQs) were drafted, covering seven main user groups and providing a framework for testing the ontology's functionality.
Limitations and Future Work
- Scope of the Example Data: While the authors are confident their example data captured the conceptual breadth and depth for specimen labels, they acknowledge that a formal evaluation of its comprehensiveness across all insect natural history information was not possible.
- Interaction Metadata Implementation: The paper highlights the ongoing challenge of finding the best technological solution for modeling interaction metadata, which is a critical area for future work.
- Draft Ontology Release: The authors plan to release a draft ontology implementation for public review and welcome additional participants to contribute to these efforts.
- Broader Ontology Potential: The work is envisioned as a stepping stone towards a broader ontology for natural history data that could encompass other animal groups.
Summary updated