A Case Study of eXplainable AI in Smart Agriculture. Lessons Learned for Information Systems Engineers

Document Type

Conference Proceeding

Source of Publication

Lecture Notes in Business Information Processing

Publication Date

6-7-2026

Abstract

While Vision-Language Models (VLMs) show promise in smart agriculture, they often function as black boxes, lacking the domain-specific transparency needed for farmer trust. Current models suffer from hallucinations or fail to provide actionable context. This paper addresses these limitations by proposing a novel integration of the BLIP-2 multimodal model with the Olive Leaf Disease Ontology. Following design science principles, our approach grounds AI-generated explanations in a structured knowledge graph to ensure semantic consistency. Results demonstrate that combining VLMs with ontologies effectively eliminates hallucinations and provides transparent, practically valuable insights into disease causes and treatments for agricultural decision support.

ISBN

[9783032279965]

ISSN

1865-1348

Publisher

Springer Nature Switzerland

Volume

587 LNBIP

First Page

28

Last Page

36

Disciplines

Computer Sciences

Keywords

eXplainable AI, Knowledge graphs, Ontologies, Sustainable agriculture, Vision–language models

Scopus ID

105041905638

Indexed in Scopus

yes

Open Access

no

Share

COinS