Fae is constituted by a fuzzy-set library, and a engine + query language which leverage that library in order to allow the user to make declarative tabular databases and then run fuzzy analysis on the database in real time via a REPL.
- Tables are simply titled fuzzy-membership functions (concepts)
- Evaluation of a table means to apply the function over the domain
- Fuzzy-set logic is used to analyse the database using a simple DSL
- New tables (concepts) can be innovated by combining tables functionally (this is done by currying) E.G., the fuzzy-intersection of hotness wetness may be used as the concept for 'humidity'.
- Compilation and parsing of scripts on startup
- JSON loading + reading of domain data (currently supports only lisp files)
- Cascading
- Error handling (let's leverage the excellent CL condition system please!)
- Data simulations
- Bitemporal domain handling
(init!) from a REPL to hook into Fae. You'll begin with an empty domain and only one table, t, which is crisp & true. Entities are maps/plists
(load filename) load domain data from a lisp file
(save filename) write domain data to a file
(rule name body) is used to define new rules via combinations of tables
(table name body) is used to define a new table programmatically (allowing you to break into Lisp to define a membership function directly).
(attr keyword) makes a crisp membership function for an attribute of an entity (accessed by the relevant symbol)
(tables) displays tables
(cut r) and (s-cut r) are the alpha-cut and strong-alpha-cut of an evaluated result
(eval fn) is used to evaluate tables over the domain
(delete table/keyword) deletes from the domain via applying a table and deleting those resultt or deleting via an attribute keyword
(mutate table/keyword) mutates the domain via applying a table and keeping the results or by an attribute keyword
(undo) and (redo) do what they sound like
Fuzzy Set Theory Lectures By Prof S Chakraverty https://www.youtube.com/watch?v=oWqXwCEfY78&list=PLjZ9ULh2Ff9FWvv-DiZHP6RN0kMr_4aW4
"BayesDB: Query the Probable Implications of Data" by Richard Tibbetts https://www.youtube.com/watch?v=7_m7JCLKmTY
MIT License