Speaker: Eben Blaisdell

Title: Parametric FCA for Fun and Profit

Abstract: Much of modern software accepts data from some users in one format simply to display the same data in another format to other users.  In a slightly different way, all declarative programming can be seen as a translation of data from a human-readable/semantic format to a machine-readable/more basic format.  This talk aims mainly to explore the mechanics of such format translation.

We start with a very brief overview of formal concept analysis (FCA), with the exclusive goal of showing that Galois connections give an incredibly natural semantics for data representation.  With this tool in our toolbelt, we generalize FCA via ‘covariant parameters’.  Not only is such a setup Turing complete, we argue that it is one of the most natural regimes for data manipulation.

As further proof of naturality, we end by demoing Ontolog, a complete implementation of a programming language in this new paradigm.  All of the preceding math is directly in the service of very practical problems.  If you’ve ever asked

 – How do I merge PDFs without getting crazy malware?

 – Where did I save that darn Pythton script?

 – What does this program want me to pass in?

 – Do I really have to learn a new tool just to do _that_?

then Ontolog might be useful to you!