API,  Authoring,  Demo @ 50,  Perspective

Things to have relationships with in the glossary

So, we have this now in Liquid | Flow. Select text, keyboard shortcut for the menu and then simply ‘g’ for Add to Glossary. In this case I’ve added Chris:

At the bottom we have this pop-up for relationships which is are so far only relationship terms for concepts. But what about for other types of entries, such as people (in the case of Chris), places, documents, things, projects, companies and so on?

How should we organise the types?

For example, if the entry is about a person, we should have at least these options for defining the relationship with another entry:

  • Works with
  • Works for
  • Works on
  • Created
  • Created by
  • Is Related to
  • is Admired by

What are the categories we should cover? I don’t mind having a few, but it should be super-quick to assign from a pop-up or from buttons (ideally, since that’s faster), but it can’t be too many. I think once he category of ‘this is’ has been set, the pop-up options for relationships with other posts can dynamically change.

What do you think? My goal is to make this EASY and QUICK for a regular writer to add but also nicely connected in your knowledge graphs once ingested.

The list so far:


  • Christopher Gutteridge

    Heh. Frode, you’ve discovered graphs.

    What is going on here, I think, is that the definition blog post
    represents a concept. That concept is uniquely identifiable and
    addressable, and in RDF land should have a URI distinct from the URL of
    the page. Although if you’re super lazy you can just use the page URL
    and add a fragment ID like #concept. eg

    What you are asking is what predicates link concepts. The answer is…
    all of them.

    What you need to ask is what’s useful in a glossary… and that’s pretty
    broad still. I would absolutely use existing (RDF) terms for this as
    people have already done lots of this work, and it means you can
    trivially express your concepts as an RDF graph.

  • Gyuri Lajos

    My approach in (Mindgraph) to this is to let people enter freely some text to indicate the kind of relationship they had in mind. Ensure that the t relationships they introduced on the fly can are defined in their own a meta glossary.

    The key point is that the system maintains a list of existing relationships and can be accessed via autocomplete search boxes at the right places. In a private personal knowledge work setting this can be very powerful and useful way to help to make sense of what is being recorded.

    This could be done for the demo, just make sure that a nice neighborhood emerges.

    At some point, as people start collaborating they will encounter other peoples ‘relationships’ and can then decide to define equivalences and decide which ones to use.
    The system should cater for propagating these changes.

    Ultimately Team and Community servers in HyperKnowledge should provide the means for storing and developing conversations over emergent repositories for relationship concepts, and can even over recommendations for specific contexts.

    In the peer to peer world it is my ambition to build Conceptipedia, a graph based emergent encyclopedia of relationships and situated regulative meta concepts.

    Using WordPress as a collaboration point, these meta glossaries can be published and searched for, giving a way to share relationship names along with the content in which they figure.

  • Frode Hegland

    Thanks all.
    I think that different things have different relationships though, and we should make this easy for the author to specify. A person can have relationships like ‘friends with’ but a concept could have another unique one such as ‘thought up by’ and so on.

    Hence I’d like a useful set of ‘types’ to choose from, then the relationship list will automatically update. HOWEVER, I think I’ll follow Guyri’s method to start with and just allow the user to type in a relationship.

    BTW, all, does the formatting of the result suit you?

  • Gyuri Lajos

    Existing Ontologies were devised with a view to structure and make human knowledge amenable to be processed by machines in terms of propositions precisely because machines cannot make sense of text. People are capable of making sense of text, and the kinds of relationships they may be interested in not just between things, but more between ideas about things and much more, that are situated and higher level and go “Beyond Ontologies” https://hypothes.is/a/jCvjfkrKEeiA8EOxywU6GQ.

    Indeed Frode discovered graphs, but not so much the ones behind Linked Data, but the ones behind “Linked Text” https://jrnl.global/2018/11/20/linked-text/ that employ “high resolution linking” https://jrnl.global/2018/11/16/liquid-author/#annotations:jZnY0OwDEeiFt1_ozZc5Lw

    aimed primarily at augmenting human capabilities not making machines more powerful.

Leave a Reply

Your email address will not be published. Required fields are marked *