Transforming Agricultural Research with CABI Thesaurus and PoolParty

Over the last year, there has been a greater push for taxonomy and knowledge graph based solutions to cover use cases across various industries. An area where this is proving to be true is in research and publishing, especially for industries where information is dense and scientific. CABI, a customer of the PoolParty Semantic Suite, uses a taxonomy at the core of their offerings to provide research and expertise to solve problems in agriculture and the environment. 

CABI is an international, intergovernmental, not-for-profit organization whose work focuses on combating crop and livestock losses from pests, invasive species, and diseases, promoting biodiversity and sustainability, and advocating for natural solutions like biocontrols instead of harmful pesticides. Through innovative scientific research, collaborative projects, and comprehensive publishing efforts, CABI aims to empower individuals with the necessary tools and insights to tackle these critical issues.

Recently, CABI relaunched their CABI Thesaurus which is used to provide a controlled vocabulary and indexing mechanism for the search functionality in the CABI Digital Library. The CABI Digital Library contains roughly 16 million records across multi-disciplinary subjects which include peer-reviewed articles, abstracts, and 80,000+ datasheets from the CABI Compendium. The Library helps people discover validated, evidence-based information to help them overcome the world’s biggest challenges.

We are proud that our PoolParty tools help CABI fulfil its mission, especially as it pertains to CABI users being able to access and use information in the Digital Library more quickly and efficiently. 

This blog dives into our story of working with CABI plus all the product updates that came recently from their side.

Improving the CABI Thesaurus

Prior to working with PoolParty, CABI was experiencing the following challenges:

• The CABI Digital Library was indexed on a legacy thesaurus that could benefit from integrating machine learning
• The CABI Thesaurus was disconnected from other projects/processes
• The CABI Compendium datasheets sat in a separate silo from the library repository
• The user search in the digital library had an opportunity for optimization to return more targeted results

The CABI Thesaurus was in need of modernization both from an internal working perspective and its frontend usability. The process to update CABI Thesaurus was mostly manual, meaning that changes to the Digital Library would take longer to see effect. 

Now with PoolParty, automated processes complement the ongoing work of manual curation. Datasheet edits automatically introduce new concepts and refine relationships between existing ones, while machine-generated suggestions improve the accuracy of document tagging.

This enables more consistency with the tagging so that content is easier to retrieve regardless of the exact keywording that the user inputs in the search field. 

Aside from searching in the CABI Digital Library, the CABI Thesaurus is also available as a SKOS/RDF knowledge graph. CABI’s Content Metadata Controller Gary Leicester advocates for the use of the CABI Thesaurus in third-party integrations such as SharePoint or other CMS which could benefit from a tailored and pre-defined dataset. The thesaurus has been carefully curated by experts in the field using industry-specific terminology, and can therefore be easily incorporated into existing systems as a foundation for further knowledge discovery.

Better search experiences

The improvements to the CABI Thesaurus serve as the first step to a better search in the Digital Library. In his blog about the launch, Gary Leicester highlights the search benefits that are now available to users. 

Gary’s blog includes short video demos of the new features which show you how you can curate a search according to your needs, so it is worth exploring to see how it all works.

In short, however, Gary’s demos show how users can filter terms by relevance, hierarchical relations, synonyms, and related concepts, which opens up the scope of the results that are returned by the search engine while still remaining precise to their query. The intelligent capabilities of semantic search improve the accuracy and relevance of search results in CABI’s extensive database.

The CABI Thesaurus plus AI

Conversational AI, which is now a common staple in the modern workplace, benefits from taxonomies and knowledge graphs especially – providing guardrails against hallucination and creating more comprehensive answers to user questions. 

This approach of combining taxonomies and knowledge graphs with large language models (e.g. ChatGPT) is called Graph RAG and is used to strengthen the output of the AI platform. It uses natural language processing and semantic search to find documents based on the user’s intent of the query, as well as generates new content about this topic while in dialogue with the “bot.” 

The CABI Thesaurus can be used in a Graph RAG scenario to feed a Conversational AI platform on the topics that the thesaurus serves. A similar example can be found in our knowledge-hub.eco, which is a Conversational AI platform that uses an ESG taxonomy to answer questions about the environment and social topics. 

Gary gives a great demonstration on how the CABI Thesaurus specifically can generate conversations with the Graphwise application, Talk To Your Graph. Check out his blog to watch the video and read his explanations. 

By modernizing the CABI Thesaurus and leveraging automatic indexing with the help of PoolParty, CABI has streamlined the process of users accessing critical, evidence-based information. The improvements not only optimize the search experience of the CABI Digital Library, but also pave the way for advanced AI applications that combine taxonomies and knowledge graphs with large language models to deliver more accurate and relevant results. 

Through this collaboration, CABI continues to empower researchers and professionals in agriculture and environmental sciences to tackle global challenges more effectively.