AGROVOC

Annual AGROVOC Editorial Community Meeting 2022

FAO

Organized by the Partnerships and UN Collaboration Division of the Food and Agriculture Organization (FAO), the 5th Annual AGROVOC Editorial Community Meeting took place online on Tuesday 19 July 2022, bringing together 50 participants from more than 20 countries and organizations. The main objectives of the meeting were to revise the work carried out during the last months and to define priorities for the next ones.

The annual meeting was divided into three main sessions: 

  • Latest updates;
  • Ongoing collaboration with networks and consortia; and
  • Panel session about the use of vocabularies as research Data Interoperability framework.

First session

During the first session the latest updates were presented. The discussion included how to improve the editorial processes, a presentation of the usage statistics, and an overview about the national and institutional outreach activities.

In the last 12 months, three institutions joined the AGROVOC Editorial  Community:

Also new editors from existing institutions (the Land Portal Foundation, Embrapa and the Iranian Fisheries Science Research Institute) joined AGROVOC and were welcomed during the initial session.

After officially introducing the new editors to the AGROVOC editorial community, the progress of the previous 12 months (See Figure 1) as well as the future goals were presented by Imma Subirats, team leader of the Agricultural Information Management and Knowledge Sharing Team at FAO. 

Figure 1. AGROVOC´s content evolution.

 

The first session, presented by Kristin Kolshus (FAO) and Esther Mietzsch (Kuratorium für Technik und Bauwesen in der Landwirtschaft e.V., KTBL), focused on two important aspects of the editorial process:

  1. What topics are missing to fill content curation gaps. Some of the emerging and highlighted themes were forestry, fishing gears, social science related to agriculture and climate change.
  2. Good practices used by the community in their work on AGROVOC. Editors mentioned meetings with experts in the discipline, looking for other alternative resources, a deep communication with language institutes/experts at the national level or the creation of a consolidated team, and efficient collaboration with domain experts.

The final part of the first session included a presentation by Daniel Martini (KTBL) about how the AGROVOC Statistics Task Force has been studying the data in detail. Usage statistics on the legacy web services, downloads, SPARQL, Loddy and Skosmos were presented as well as the usage by country and access by language content. 

A brief presentation by Imma Subirats and Veronica Venturini (FAO), and Esther Mietzsch (KTBL), on national and institutional outreach activities closed this first session: results from the polls showed that the majority of participants carried out outreach activities during the last 12 months such as the publication of articles, delivering presentations in conferences and public events, sharing information and news about their work on social media channels. Part of those who didn't carry out such activities, faced as main barriers lack of resources and institutional support. The end of the first session saw very lively discussions, during which the importance of collaboration and outreach was underlined. 

Figure 2. Participants to the Annual AGROVOC Editorial Community Meeting 2022.

Second session

The Annual AGROVOC Editorial Community Meeting was also the chance to present two important ongoing collaborations between AGROVOC and two other networks: the CGIAR task group, and the ASFA network. This session was facilitated by Carsten Hoffmann from the Leibniz Centre for Agricultural Landscape Research (ZALF).

The audience was particularly interested in the methodology used to send new concepts to ASFA, and the selection criteria for choosing concepts in the context of CGIAR. Tasmin Vicary from the Aquatic Sciences and Fisheries Abstracts (ASFA) presented the ASFA Thesaurus used to index records in the OpenASFA, while Sara Jani, from the International Center for Agricultural Research in the Dry Areas (ICARDA), introduced the CGIAR task group, which uses an online shared form with a template for AGROVOC concepts. Accepted concepts are then added to VocBench by the Monitoring, Evaluation & Learning (MEL) team of ICARDA on behalf of the CGIAR task group. 

Third and last session

The panel session organized by Daniel Martini (KTBL) about the use of vocabularies as research data interoperability framework brought together three panelists from different organizations to present their experience: Dr. Clement Jonquet, from the University of Montpellier, Professor Christopher Brewster, from the Maastricht University, and Davide Gibin, from the European Food Safety Authority (EFSA).

The discussion circled around three focus questions:

  1. What mechanisms and tools should be foreseen or do you foresee in your projects for users to interact with vocabularies in research data management (navigational aids, search, autocomplete…)?
  2. Where do you see gaps in current vocabularies? and
  3. How do you intend to fill them?

Several initiatives were highlighted, including for example research outputs, an EFSA AI tool on food safety, Agroportal, FAIRness assessment, use of semantic artifacts, and the D2KAB project (knowledge graphs in agriculture). Different use cases exist for vocabularies versus ontologies, depending on the appropriate level of semantic and logical commitment needed. Domain ontologies in specific areas are different from application ontologies, and from reference thesauri like AGROVOC or NALT, which are used to index data. 

Challenges exist in metadata annotation, especially as several resources describe and classify information differently. The agricultural sector has over 300 vocabularies and ontologies, and selecting the right one depends on several aspects. For example, when collecting data from different information points, in practice, it is necessary to deal with different vocabularies and there is a need to train people on how to map their data.The FAIR data principles are also relevant in what is selected. Mappings as well as vocabularies to map should be as FAIR as possible.

Regarding alignments and mappings, vocabulary maintenance requires tools and resources. Approaches currently tested include the use of machine learning to identify candidate terms, and quickly identify what is missing, at a wider scale, but some degree of human curation is still needed. Procedures and tools that support the vocabulary development and maintenance process are needed, for example to capture suggestions for vocabularies and for mappings between semantic resources. From creation on for the whole life cycle of ontologies and vocabularies (conception, design, development, to operation and maintenance), there is a need to track changes in these standards used to collect data and to continuously add and maintain mappings to different vocabularies.

Possible suggested topic gaps in current vocabularies were precision agriculture, phenological aspect of different crops and climate/weather data, and data quality. There was considerable interest in the panel discussion, and there is potential to explore the topics further.

To close the event, two presentations on VocBench and ShowVoc were presented, followed by AGROVOC Outreach activities. Andrea Turbati (FAO) presented new technical developments about VocBench. It is the official tool to edit and maintain AGROVOC, a web-based, multilingual, collaborative development platform for managing OWL ontologies, SKOS(/XL) thesauri, Ontolex-lemon lexicons and generic RDF datasets. New features in Vocbench relevant to AGROVOC include that Custom Forms and Custom Views are now independent, and it is possible to create different types of Custom Views for a project, and graphic widgets for Custom Views. 

A similar tool to VocBench, ShowVoc, was also highlighted and briefly presented. It can be seen as a Metadata Catalog, a Dataset Catalog, and a Dataset/ontology/thesaurus/lexicon browser. Global Search can cover all datasets with a single action. For now, VocBench remains the official tool to edit AGROVOC.

Veronica Venturini (FAO), last presenter of the day, presented the AGROVOC outreach activities carried out during the last year. She showed the editorial community several channels and opportunities to communicate their work to FAO AGROVOC core team but also inside and outside their institutions.

A very pleasant and relaxing social session was organized for the AGROVOC editorial community by Lilian Lee (LandPortal Foundation) just after the end of the main program, where about 20 participants enjoyed a less formal exchange on a number of topics.

We would like to thank all the editors and experts who took part in the Annual AGROVOC Editorial Community Meeting.