Associative structure is best illustrated by the knowledgebase everyone is familiar with: Wikipedia.
MediaWiki (the software that powers Wikipedia) supports linking to other pages, while software such as RoamResearch supports linking to blocks (e.g. sentences in other pages). Regardless of the technical details, in such a structure, knowledge is associated with other knowledge through mechanisms like links or tags.
It's a powerful structure precisely because of how unopinionated it is. There's an almost unlimited amount of flexibility, which helps you avoid being overly dogmatic about a single "point of view", and allows the underlying structure of your knowledge to emerge from these many interrelationships. Over time, you'll discover that some concepts are frequently referenced, while others may largely link to others or be isolated.
Speaking in generalities, this structure tends to do better at the "lower" levels of your PKB, as we tend to desire a little bit of encapsulation/abstraction in order to work with big ideas. Chunking your knowledge into "domains" that are loosely coupled to other domains, but are also highly-cohesive, richly connected graphs within, can be a great strategy to get the benefits of an organic, flexible structure without being swamped by the complexity.
Users of associative structures will often utilize:
A Graph View (e.g. obsidian)
Index notes: Instead of using folders to group notes under a "concept", other notes (or perhaps "higher-order notes"?) will be used to collate notes under a concept via links.