top of page

Knowledge Graphs for AI

AdobeStock_218869095.jpeg

Fact Based Context for AI

AI is fast becoming part of how farmers make decisions. Chat-based advisory services already answer questions about pests, soils, crop varieties and planting times, often by phone or messaging app and in local languages, and they can reach far more farmers than traditional extension services ever could. As these tools improve, more and more everyday decisions on the farm will be shaped by the advice they give. That makes the quality of that advice matter a great deal: a recommendation on what to plant, when to spray or how to treat the soil has real consequences for a farmer's yield, income and land.

 

The risk is that AI without context gives advice that sounds right but isn't. A language model can produce a fluent, confident answer that is wrong for the farmer's region, crop, season or regulatory setting. It may recommend a product that isn't registered in that country, draw on research from a very different farming system, or fill a gap in its knowledge with something plausible rather than admit it doesn't know. And because the answer comes without a clear trail back to its source, neither the farmer nor the advisory service can easily check it. As AI advice becomes more widely used, these errors scale with it.

 

Knowledge graphs give AI the context it needs to get this right. Used alongside retrieval-augmented generation (RAG), a knowledge graph supplies AI tools with structured facts and evidence rather than loosely matched passages of text: which crops, pests and practices are connected, which findings apply to which conditions, and where each piece of evidence came from. This means advice can be checked against its source, can be filtered to what genuinely applies to a farmer's situation, and can say clearly when the evidence isn't there. Because concepts are linked to shared standards, the graph can also connect local names for crops and pests to the same underlying knowledge, so advice holds up across languages. At Knowmatics, we build and curate these knowledge graphs so that AI advisory services can give advice that is accurate, relevant and traceable.

  • twitter

©2018 by knowmatics. Proudly created by Bibliobox

​

bottom of page