What is a personal knowledge graph?

A personal knowledge graph is what happens when your saved articles, ideas, references and notes stop being a list and start being a map — with explicit links between the things you know.

From folders to graphs

The traditional way to file information online is a tree: folders inside folders, notebooks inside notebooks, tags on a single axis. Trees are clean to render but they fight you the moment a piece of information belongs in more than one place. A paper on machine-learning fairness sits awkwardly between ML, ethics, and policy — and you end up either picking one folder and losing the others, or duplicating the entry and creating drift.

A graph dissolves that problem. Each thing — a bookmark, a topic, a person, a project — is a node. Each relationship between them is an edge. The ML-fairness paper can connect to ML and to ethics and to policy at the same time, without copies and without contradiction.

What makes it "personal"

Wikipedia is a knowledge graph. So is the dataset behind Google's rich-result panels. The point of a personal knowledge graph is that it's yours — the nodes are the things you care about, the edges are the relationships you see between them, and the whole structure lives in storage you control.

That last property matters. A personal graph is only durable if you can keep it across tool changes, vendor closures, and price hikes. Tools come and go; your graph shouldn't.

Where to start: bookmarks

A personal knowledge graph sounds ambitious, but the lightest version is a bookmark archive with one extra rule: every saved link carries at least one topic, and topics can be related to each other. With a few hundred entries, patterns emerge — clusters of interests, gaps in your reading, ideas that cross-pollinate between domains.

mnera.io is designed around exactly this idea. Every bookmark is a node, every topic is a node, every edge is stored as W3C-standard linked data, and the whole graph lives in a private Solid pod you own. When you open the Atlas view, you're looking at the shape of your interests as they evolved over time.

Things you can do with one that you can't do with a list

• Find what you forgot you knew. "What did I read about Byzantine fault tolerance two years ago?" A graph surfaces it through its connections, even when you can't remember the title.
• Spot related ideas. A new article on incentive design connects to game-theory pieces from years earlier you'd otherwise miss.
• See your interests change over time. An activity heatmap or topic-share-over-time chart turns a saved-link archive into a clearer map of your attention.
• Share a slice cleanly. Pull every bookmark under a topic, with structure preserved, into a reading list you can hand to someone else — without exposing the rest of your graph.

How to build one from where you are now

1. Pick a tool that stores links as linked data, not in a proprietary silo (mnera.io, or any other Solid-aware bookmark app).
2. Import your existing bookmarks — browser, Pocket, Raindrop, Evernote — so the graph starts with the material you already have.
3. Add at least one topic per bookmark on the way in. Don't optimise the taxonomy yet.
4. After 50–100 entries, look at the topic cloud. Merge near-duplicates, introduce parent topics where natural, and split topics that are doing too much work.
5. Keep going. The graph compounds — every new addition slots into existing structure, every existing entry gains new neighbours.

If you want to explore further, the next two pieces are Personal knowledge management systems, explained and Why browser bookmarks fail.