What up

Social science yo

• Meta

For ontology files put the following at the top of the file:

:- module(test1, [test1/0]).
:- reexport('folder/child').
:- include('theoryToolbox2').

For theories add the following::

:- use_module(library(scasp)).

From sCASP docs:

Op	Description
?—	Prove and only show the bindings
?+-	Prove, show bindings and model
?-+	Prove, show bindings and justification (tree)
?++	Prove, show bindings model and justification)
??+-	As above, but using human language output
??-+
??++

To-Do’s

• Do I need an Obsidian plugin? Could be useful to automate
• How to use canvas view in this case?
  • Should be easy because any item in canvas can be a md page
  • might need to add canvas labels manually to match named relations from dataview
    • can also make a script for that?
• Degree of standardization? Standard tags? Classes/hierarchies?
• need to improve logic in general and specifically code to represent hierarchies inside praxis
  • see chapter in Simply Logical book

Approach

1. Semantic layer/Ontology/Glossary/Dictionary
   • ontology in obsidian + dataviewjs + graph-link-types with labeled links or frontmatter
     • turn links into Prolog statements with a script like so
     • Make named links with link :: . (optional) use aliases for notes and tags for structure
     • optionally serialize for storage/exchange via RDF/turtle/any other format
   • hierarchy with folder note and waypoint (basically markdown links)
     • pick and node in the folder hierarchy as root
     • create a “Main” file in this node which will hold Prolog code
     • parent “Main” files can import children. There can be one big top-level root file (very useful for inference)
     • root file acts like a package manifest in software development. can include more stuff like links, relation definitions etc
       • Make a script that will automatically update parent root files to import children
     • can optionally do types and any kind of organization, tags, supertags etc
   • query this knowledge base with Prolog or whatever you want e.g. SparQL !
2. Logic/structure/inference layer
   • Try it
   • inference in prolog; see Modes of Inference
     • express theories and logic/structures with prolog code (see theoryToolbox)
     • induction with FOLD
     • deduction and abduction with sCASP
3. Prediction/magic layer
   • Can pass the output to an LLM or other algorithms to answer questions or conduct search. Both to extend knowledge base and for inference
     • Try it
   • … what else?

Advantages of this approach and goodies

• Can be easily stored in git, it’s just markdown and prolog with one js script on top
• Can “enrich” with external ontologies e.g. DBPedia.
  • Actually let’s add an “externalDefinition” frontmatter item that will use any external link
• Can then import generated Prolog or generate code for Logtalk/Logica/other more advanced languages or compile to SQL
• Can optionally use SparQL anything or similar to parse the whole contents of markdown pages as structured content, or use advanced frontmatter for more structure
• How to group terms with similar meanings
• Can visualize prolog logic
  • Visualise relationships, logical rules (AND, OR) and tables in Praxis
  • Just import the whole root code into Praxis 🤯