When building your Knowledge Graph


Aim to Keep Your Concepts Small

Humans can only understand and absorb what they can hold in short-term memory. You want every user query to come back with a single digestible answer. If a concept can't be described in one or two paragraphs, its scope is too broad.

Large concepts - referred to as macro-concepts - may be useful for Generative AI purposes (e.g. with ChatGPT), but you should always aim to break them down later.


Don't Give them Too Many Facts

For example, a "laptop" concept with a 1000 models and their configurations is just a meaningless list with little semantic value. Better to use a meaningful description with a narrow scope like (say) "Lenovo business laptop" that you can attach fewer facts to.


If you're new to knowledge graphs and you're looking for a comprehensive step-by-step guide this article is a good reference.