Hode is, loosely speaking, a hypergraph editor.
The quickstart guide is a super-short (730 word) introduction to Hode. Once you've slugged that down, come back to this README for a deeper look.
These are not a complete guide to Hode, but they quickly show what it's good for. Hode has gotten slicker since these videos were made, and it has new features (e.g., Hode handles order now), but the basic idea remains the same.
The video on navigation is probably more interesting. I suspect it will make sense even if you haven't watched the video on editing, but I'm not sure.
Video: How to navigate (search and crawl) .
The in-app help describes all the UI and language features of Hode.
You'll still need to use the out-of-app documentation for a good introduction to the app -- one that covers what and why, not just how. But you don't have to study the docs too thoroughly, because once you've seen what can be done, the in-app help can quickly reveal how to do it.
The in-app help is more up-to-date than the docs -- it might cover advanced features not otherwise documented, and the keywords and keyboard shortcuts it describes are guaranteed to be correct and complete, whereas the documentation outside of the app might not be.
Access the help by pressing M-?.
(That probably means Alt-?, unless you use a Mac).
Hode is an editor for higher-order data.
There are three branches to Hode: The RSLT data structure, the Hash language for describing subsets of an RSLT, and the UI, which lets you use the previous two things.
A Rslt (Reflexive Set of Labeled Tuples)
is a generalization of a knowledge graph.
It lets a user easily represent any natural language expression.
(A Rslt is isomorphic to what some programmers call a "hypergraph" --
but mathematicians claimed that term first,
and in math it means something much less expressive.)
A Rsltis a collection of expressions,
each of which is either a phrase (like "cats"),
or a relationship (like "cats have noses")
or a template (like "_ have _") shared by many relationships.
What distinguishes a Rslt from a graph
is that relationships can involve any (positive) number of members,
and a relationship can itself belong to other relationships.
If you already read the quickstart guide, you might not even need to read about the Rslt in detail. But it's a super-short document -- under 600 words.
Hash is a language,
close to ordinary natural language,
for talking about expressions in a Rslt.
It offers a concise representation,
both for individual Exprs (expressions) in a Rslt,
and for queries to retrieve sets of Exprs.
Hash doesn't do anything. It is used only to describe parts of a graph: The expression you'd like to add, the subset you'd like to search for, etc.
To run simple searches you won't need to know anything about Hash.
Most of the searches I run are simple -- e.g. /find Steve
will tell Hode to find the phrase "Steve".
To run fancier searches -- "find me everyone who came to my wedding",
"find every food Sam likes that Ella isn't allergic to" --
you'll need to
read about Hash.
Txhe UI lets you do stuff to your data -- add to it, modify it, search for it, load and save it, etc.
To use Hode you'll probablyxneed to read about the UI.
(Actually, the UI is optional. The module Hode.NoUI lets you use Hode directly from GHCI, without a UI. The NoUI module is not as complete as the UI, though -- in particular, transitivity and order are hard to deal with outside of the UI.)
The fourth branch of Hode, Qseq is a metalanguage for search. It lets you use existential and universal quantifiers to string together searches in some underlying search language.
If you're a curious hacker, Qseq has been implemented, but it hasn't made it's way into the UI yet.
That's because Hode relies on the Brick terminal interface library. There are, fortunately, lots of ways to run Linux in Windows -- Docker, for instance.
Before installing Hode, you'll need to first have installed
-
Stack, the Haskell toolkit -
Git, the version-control software -
Optionally, you can install
xselfor some clipboard functionality.
In the folder stack-options you'll find two files.
One of them is what to use if you run NixOS.
The other is what to use if you run anything else.
Copy one of them to the root of the project, and rename it stack.yaml.
Open a console (a.k.a. "terminal", "command line", "command prompt"). Clone the repo, enter it, and start ghci:
git clone https://github.com/JeffreyBenjaminBrown/hode
cd hode
stack run
The first time you try that, it will take a long time to start.
Depending on your system, Stack might complain that
you need a few more things installed.
(libxrandr-dev, lbxss-dev and libx11-dev, maybe?).
If it does, install those too.
Then return to the cloned repo and try stack ghci again.